What the researchers actually tested

Alongside the conversations themselves, participants in the initial two lab studies pairing up strangers also completed measures of four personality traits that seemed like plausible candidates for explaining who underestimates being liked the most: narcissism, shyness, rejection sensitivity, and self-esteem.

Boothby and her co-authors ran a statistical test for each of the four, checking whether it predicted a bigger or smaller gap between how much a participant thought they were liked and how much their partner actually reported liking them.

The one trait that mattered

Only shyness moved the needle, and it moved it in a clean, dose-dependent way: the shyer someone was, the wider the gap between how liked they thought they were and how liked they actually were.

Rejection sensitivity didn’t predict the gap at all. Neither did self-esteem. Neither did narcissism — three traits that, on paper, sound like exactly the kind of thing that should track with underestimating how much people like you, and none of them did. The researchers had also built in a statistical safeguard for testing four traits at once, the kind of correction designed to catch a fluke result before it gets mistaken for a real one.

Shyness passed that test. The other three didn’t even come close.

Why shyness specifically, and not the others

The result lines up with what shyness actually is, as distinct from the traits it gets lumped in with. Low self-esteem is a general negative view of one’s own worth; rejection sensitivity is a heightened alertness to signs of social rejection specifically; narcissism involves an inflated self-view that, if anything, might be expected to shrink the liking gap rather than widen it. Shyness is neither of those — it’s closer to a chronic discomfort in the moment of interacting, paired with a tendency toward self-critical, self-monitoring thought during and after a conversation.

That specific combination — watching yourself closely while assuming the worst about how it’s landing — lines up with the mechanism the broader liking-gap research points to: people miss signs that their partner likes them because they’re too absorbed in their own self-critical read of the interaction to register the signal.

Boothby has described this and related findings as part of a broader family of “social illusions” that distort self-perception in fairly specific, individual ways rather than as a single uniform bias everyone shares equally.

What this means for anyone building an audience through writing rather than conversation

Shy people are, on average, drawn toward exactly the kind of communication a blog or newsletter offers: asynchronous, revisable, and free of the live, unfolding judgment a face-to-face exchange demands in the moment. Susan Cain’s Quiet: The Power of Introverts in a World That Can’t Stop Talking made the broader case that a culture built for extroverts systematically undervalues introverts’ contributions. The liking-gap data points to something related but distinct: shy people aren’t actually valued less by the people they talk to — they’re just the ones least equipped to see how much they already are.

A shy writer who assumes a lukewarm response to a post, a pitch, or a comment thread is applying the same distorted instrument the study measured directly: not a neutral read of how the interaction went, but a self-critical one, calibrated by a trait that has nothing to do with how the other person actually felt.

The finding also rules out a tempting alternative explanation. It would be easy to assume that anyone anxious about social standing — someone high in rejection sensitivity, or low in self-esteem — would show the same distorted read. They don’t. Only shyness does. Whatever is driving the gap in shy people specifically, it isn’t a general fear of rejection or a general dislike of oneself; it’s something more particular to shyness itself, and it was strong enough to survive statistical correction for testing four traits at once.

The post The liking gap turned out to be largest specifically in the shyest people studied, meaning the people most convinced they weren’t liked were also the same people whose own self-criticism was distorting the read in the first place appeared first on The Blog Herald.

The original design, and what it actually measured

The first study, run by psychologist James Pennebaker and his student Sandra Beall, only had access to a fairly blunt outcome measure: health center visits. Subjects wrote about either “personally traumatic life events” or trivial topics on four consecutive days, fifteen minutes a session, and researchers then pulled their health center records for the following six months. As Pennebaker later described the finding, students who had written about trauma “ended up going to the student health center over the next 6 months at about half the rate of students in the control condition.”

The trauma group in that first study was actually three groups — subjects wrote about the facts of their trauma, their feelings about it, or both — and the health-center drop was concentrated in the group that wrote about both. The result was interesting enough that it raised an obvious follow-up question: if something measurable was changing, could it be detected more directly than through a count of doctor’s visits?

The follow-up that added blood work

Two years later, Pennebaker teamed up with psychologist Janice Kiecolt-Glaser and immunologist Ronald Glaser, both at Ohio State, to rerun the design with fifty healthy undergraduates, stretching the writing sessions to twenty minutes a day and adding blood draws at three points: the day before writing began, about an hour after the final essay, and six weeks later. The researchers measured how strongly each subject’s blood sample responded to two substances, phytohemagglutinin and concanavalin A, that are known to trigger T-lymphocytes to proliferate — a standard lab proxy for how vigorously the immune system mobilizes against a challenge.

The result: on one of the two immune measures researchers tracked, the trauma group’s response rose further above its own starting point than the control group’s did — a difference too consistent to be chance. The second measure moved in the same direction but wasn’t strong enough on its own to count as a reliable finding.

What subjects actually wrote about, and what it cost them in the short term

The trauma group’s essays covered adjusting to college (19%), parental problems including divorce and family conflict (14%), a death (13%), injury or illness (12%), and sexual abuse (9%), among other topics — specific, concrete material grounded in real events. The immediate experience of writing it was not pleasant: compared with the control group, trauma subjects reported measurably higher negative moods and more physical symptoms right after each session (both p < .001), a cost the researchers were careful to report rather than smooth over. Whatever benefit showed up in the six-week and six-month data came after four nights that, by subjects’ own accounts, felt worse rather than better while they were happening.

The caveats the researchers themselves flagged

The 1988 study’s authors noted their own complication: the writing sessions took place in early February, one of the highest-illness weeks of the academic year, and the six-week follow-up blood draw landed just before spring break, a comparatively low-illness period. Some of the apparent shift in health center visits, they acknowledged, could reflect that seasonal swing rather than the writing itself — which is part of why the lab-measured mitogen data mattered as a separate, independent line of evidence pointing the same direction. Pennebaker has also been candid that the underlying explanation for why any of this happens was, in his own words, “still debated—and continues to be debated today,” more than three decades and dozens of replications later.

None of this makes writing about a hard experience a substitute for medical or psychological care, and the researchers never claimed otherwise — a handful of fifteen- or twenty-minute sessions moved a lab measure of immune response, not a diagnosis. What the two studies do establish is that the difference between describing a difficult event in specific, personal detail and describing an ordinary one was large enough to register in blood drawn six weeks later, in subjects who felt worse, not better, on the nights they wrote it. That is a strange enough result to explain why the same basic design — four days, a blank page, a hard thing named directly on the page — has been rerun by other labs for decades since, on a phenomenon researchers still can’t fully explain. It is also the closest thing personal blogging and journaling-style writing have to an actual research pedigree: the confessional post that names something specific and difficult sits in the same lineage as Pennebaker’s essays, whatever the underlying mechanism eventually turns out to be.

The post A 1988 study had one group of people write about their most traumatic experience for 20 minutes a day, four days in a row, while another group spent the same four days writing about something completely ordinary, and when researchers measured their immune response afterward, the group who wrote about the hard thing came out with a measurably stronger one appeared first on The Blog Herald.

Psychologists borrowed that image, minus the murder, to describe something ordinary: people about to speak in front of others routinely believe their nervousness is far more obvious to the room than it actually is.

Where the name comes from

Cornell psychologist Thomas Gilovich, along with Kenneth Savitsky and Victoria Husted Medvec, gave this mismatch a name in a 1998 paper that opened with the Poe passage: the illusion of transparency, a term borrowed from Dale Miller and Cathy McFarland. In one of their experiments, participants took turns telling truths and lies to a small group and then guessing which of the others had lied. Liars estimated that 48.8% of the group would correctly identify them as the liar. The actual detection rate was 25.6% — indistinguishable from chance. People lying about something as mundane as their preferred shampoo brand were convinced their deception was nearly twice as visible as it actually was.

The same gap over something involuntary

A second experiment removed the element of deliberate deception entirely. Participants tasted a series of drinks on camera, five of them deliberately foul, while trying to keep a neutral expression, and predicted how many of ten observers would later guess correctly which drinks had disgusted them.

Tasters predicted 4.91 correct guesses on average; the observers actually managed 3.56, a rate no different from chance guessing at 3.33. The gap held even without a lie being told — just an involuntary reaction the taster was trying to mask, and assumed was leaking through more than it was.

Taking the same idea to a podium

Five years later, Savitsky and Gilovich tested the same illusion specifically in public speaking, where the stakes of feeling exposed are higher than a card game or a taste test. Their design was later replicated in detail in a 2011 master’s thesis, which put speakers through a three-minute speech under the same three conditions: some were told about the illusion of transparency beforehand, some were simply reassured not to worry about their anxiety, and a control group got no instruction at all.

That replication, run on a smaller and more socially anxious sample, found the same direction of effect but not enough of one to call statistically significant. Participants and independent observers then rated the speeches on quality and on how anxious the speaker had appeared.

The study is listed among Savitsky’s publications as appearing in the Journal of Experimental Social Psychology in 2003.

What just knowing about the illusion did

The speakers who had been told about the illusion of transparency — nothing more than an explanation of the phenomenon itself — rated their own speeches as significantly higher quality than speakers in either of the other two conditions, and rated themselves as appearing more relaxed.

The effect wasn’t confined to the speakers’ own generous self-assessment: independent observers, watching the same speeches without knowing which condition a speaker belonged to, also rated the informed speakers as more relaxed and as giving better speeches.

Simply naming the gap between felt anxiety and visible anxiety changed the speech itself — the observers grading it had no opinion to be flattered.

Why the mismatch happens at all

Gilovich, Savitsky, and Medvec traced the bias to something more mundane than social paranoia: people anchor on their own internal experience and then adjust away from it to guess what others perceive, and that adjustment is consistently too small. A nervous speaker is acutely aware of a racing heart and a dry mouth: information no one else in the room has access to. Trying to imagine the audience’s perspective means starting from that vivid internal signal and discounting it, and the discount reliably falls short, leaving the speaker convinced that some of what they’re feeling must be visible when, for anyone watching from the outside, it mostly isn’t.

What makes the finding durable rather than a one-off lab curiosity is how little it depends on the specific setting: a card game about lying, a taste test involving a foul drink, a three-minute speech to strangers — different rooms, different stakes, the same gap between how exposed people feel and how exposed they actually are. Writers who hesitate before hitting publish on something personal, bracing for a comment section to notice every hedge and uncertainty in the piece, are running the same distorted calculation Poe’s narrator ran under the floorboards: mistaking the loudness of a feeling for the volume at which anyone else can actually hear it.

What this means for the room you’re dreading”

That’s a strange kind of relief for anyone bracing for a room, a stage, or a comment section to notice every crack in their composure: the noticing was never as thorough as it felt. Simply hearing about the gap, with nothing else changed, was enough to make the speakers in Savitsky and Gilovich’s study come across as calmer to observers who had no reason to be generous about it. Whatever room someone is dreading, the heartbeat under the floorboards is usually audible only to the person standing on them.

Rationale: Ties off the piece by handing the reader something to actually use, rather than just restating the finding — it points to the one result in the article that’s directly actionable (knowing about the bias measurably changed how speakers came across, per blind observers) and closes the loop on the Poe/floorboards image from the opening line rather than introducing a new device.

The post People about to give a speech consistently believe their own nervousness is far more visible to the room than it actually is, a specific mismatch researchers gave its own name decades ago and the pattern keeps showing up appeared first on The Blog Herald.

The averages were lopsided from the start, and the researchers, Monique Pollmann, Tyler Norman, and Erin Crockett, were specifically testing whether the volume of texting could substitute for time spent together in a way earlier, less granular studies had suggested it might.

Participants logged about four and a half hours of face-to-face contact with their partner on a typical day, against eight minutes of phone calls, eight minutes of video calls, and just over an hour of texting. When the researchers ran the numbers, only one of those channels moved either outcome.

What actually predicted feeling understood

Time spent face to face was the one variable connected to both measures. On days when participants logged more in-person time with their partner than usual, they reported feeling more understood and more satisfied with the relationship — on days one standard deviation above the average for face-to-face contact, relationship satisfaction was 0.26 points higher on a 5-point scale, a real and specific shift tied to time in the same room. Phone calls and video calls, despite also carrying voice and in the case of video, faces, didn’t predict either outcome on their own once face-to-face time was accounted for.

What texting didn’t do

For relationship satisfaction, texting’s effect was flat. The researchers report that there was no main effect of texting on relationship satisfaction, and no interaction between texting and face-to-face time on satisfaction either — texting a lot on a low-face-to-face day didn’t move satisfaction any more than texting a little did on a high-face-to-face day. That result cut against the study’s own prediction going in; prior cross-sectional research had suggested texting could carry more weight than that, and this daily-tracking method found otherwise.

The one place texting did register

Understanding told a more layered story than satisfaction did. Texting showed no association with feeling understood on days when face-to-face contact was already high — once a couple had spent hours together in person, extra texting added nothing measurable. But on days when face-to-face contact was low, texting was positively associated with feeling understood, functioning as a partial stand-in for the in-person time that wasn’t happening. Understanding was still highest, across the board, on days with high face-to-face contact; texting only ever closed part of the gap left by its absence, never matched it.

Why phone calls worked and video calls didn’t

What’s notable here isn’t a null result for voice — it’s a split one. Phone calls, despite averaging just 8 minutes a day, significantly predicted both outcomes on their own, controlling for face-to-face time and texting: more phone contact was associated with feeling more understood and more satisfied with the relationship. Video calls were the one channel that came back null on both measures.

That lines up with a separate line of research on reconnecting with old friends by phone versus email, which found that hearing a voice, even without seeing a face, produced measurably more felt connection than exchanging text did — and the researchers behind that study found the same voice-based advantage held for strangers getting acquainted as well as for old friends reconnecting. Here, even a small daily dose of voice contact showed up as a real predictor, while texting’s much larger volume didn’t move satisfaction at all.

What this suggests about relationships built mostly in writing

Writing partnerships and long-distance collaborations run on almost the inverse ratio of what this sample logged: hours of Slack messages, comments, and shared docs against a handful of calls, if any. The study doesn’t speak directly to co-authors or critique partners, but the mechanism it isolates — that understanding tracks face-to-face time specifically, with texting only ever partially covering for its absence — is a caution against assuming a long, responsive DM thread with a writing partner is building the same mutual understanding that an actual call or meetup would. A collaborator who replies within minutes to every message may still be someone their co-author doesn’t understand especially well, if all of that responsiveness has happened entirely in text.

None of this argues for abandoning text messages, which still handled the logistics of coordinating a life together in this sample the same way they do in most relationships. It argues against treating the volume of messages as a proxy for the health of a relationship, or assuming that a long thread of texts throughout the day is doing the same relational work as being in the room.

The couples in this study who spent more time face to face felt more understood and more satisfied, on the same day, regardless of how many texts had gone back and forth — and the texts, however many there were, never showed up as the reason why.

The post A study that tracked how couples split their time between texting, calls, and being in the same room found that face-to-face time — and, more surprisingly, even brief phone calls — predicted how understood a partner felt, while hours of texting moved satisfaction not at all appeared first on The Blog Herald.

What the experiment actually asked people to do

In a study published in the Journal of Experimental Psychology: General by marketing professor Amit Kumar and psychologist Nicholas Epley, participants were asked to identify an old friend they had lost touch with and would like to reconnect with. Before any contact happened, they predicted how awkward the exchange would be and how connected they’d feel afterward, by phone and by email. Then they were randomly assigned to actually reach out using one medium or the other and reported back on how it went. Participants consistently predicted the phone call would be more awkward than the email, even though they also, somewhat contradictorily, expected it would leave them feeling more connected. Given that tension, most said they’d rather email.

What happened when they actually called

The calls didn’t go the way anyone expected. People who phoned their old friend reported forming a significantly stronger bond than those who emailed, and — the detail that overturns the whole prediction — they did not find it any more awkward than the people who emailed did. As Kumar put it in an interview about the research, “We think it’s going to be awkward to talk to somebody, but that just turns out not to be the case. Instead… people form significantly stronger bonds when they’re talking on the phone than when communicating over email.” The researchers even timed the calls against how long it took people to compose and send an email: reconnecting by phone took roughly the same amount of time as reconnecting by text, which knocks out the other common justification for avoiding it.

The same pattern shows up with total strangers

A second experiment in the same study removed the old-friendship variable entirely. Participants were paired with someone they’d never met and assigned to get acquainted using voice chat, video chat, or text chat, working through a set of increasingly personal questions designed to build closeness quickly.

Once again, people didn’t even expect a difference — they predicted feeling equally connected and equally awkward no matter which format they were assigned. And once again, their actual experience diverged from that expectation: hearing another person’s voice produced significantly more closeness than typing did, with no corresponding rise in awkwardness, and video added nothing measurable beyond what audio alone provided.

The effect wasn’t limited to a narrow kind of exchange, either: participants worked through prompts ranging from ordinary icebreakers to more personal ones, like describing an embarrassing moment — the full range the “fast friends” procedure is built to cover — and the voice-based conditions still came out ahead overall.

Why the voice itself seems to be doing the work

The researchers’ explanation centers on what a voice carries that text strips out. Kumar has pointed to related research showing that people who merely heard a job pitch rated the applicant as more thoughtful and competent than people who read the identical pitch as a transcript, and that people who listened to a political opponent’s argument were less likely to dehumanize them than people who read the same words on a page. None of that requires anything visual — video conferred no advantage over audio alone in either the original study or these related ones — which points to something specific about vocal tone, pacing, and inflection carrying signals of thought and feeling that flattened text simply does not transmit, regardless of how well-written the message is.

Why this cuts against the grain of how most outreach gets done

Email and its faster cousins — a DM, a comment reply, a scheduling link — have become the default for almost every kind of professional reconnection a writer does: pitching an editor who hasn’t replied in months, following up with a podcast host after a guest appearance, checking in with a co-author on a stalled project. Each of those defaults exists because writing a message feels lower-risk than picking up the phone, exactly the calculation Kumar and Epley’s participants made before their calls. The research does not suggest email is a bad tool for scheduling or quick logistics — participants in the study still found it useful for exactly that. What it suggests is that when the actual goal is repairing or deepening a relationship rather than coordinating a task, the medium people instinctively reach for is the one built to under-deliver on that specific goal.

The gap between prediction and experience is the part worth sitting with. It isn’t that people know calling would feel better and email anyway out of habit; the study’s participants got the emotional trade-off backward, expecting connection without paying the cost of awkwardness that never actually showed up. For anyone maintaining a list of people they mean to reach out to — a source who hasn’t returned an email, a fellow writer they haven’t spoken to since a conference two years ago — the research suggests the outreach method that feels the safest to send is quietly the one least likely to produce the connection they’re reaching out for in the first place.

The post Researchers asked people to reconnect with an old friend either by phone or by email, and the people who predicted the phone call would feel awkward were wrong — it left them significantly more connected than email did, with no extra awkwardness at all appeared first on The Blog Herald.

In a series of studies that paired up strangers for conversations, tracked first-year college students moving in with new roommates, and followed members of the public through a personal-development workshop, researchers found that people consistently underestimate how much a conversation partner liked them — a gap between how much people are actually liked and how much they think they were liked that the researchers named the liking gap.

What the mismatch actually looks like

In the first test of the effect, researchers at Yale, Cornell, Harvard, and the University of Essex paired up strangers, gave them a getting-to-know-you conversation to have, and then asked each person two things: how much they liked their partner, and how much they thought their partner liked them.

On a 7-point scale, participants reported liking their partner at an average of 5.82. But they predicted their partner liked them at only 5.17 — a real, measurable gap between the warmth people actually felt and the warmth they assumed was being returned, in the same conversation, among the same two people. Because both members of a pair cannot simultaneously like the other person more than they are liked back, the gap has to be an error in perception rather than an accurate read of a colder reality.

Why the gap opens up in the first place

The researchers point to a few reasons this happens. Conversations run on a kind of enforced politeness that keeps people from broadcasting exactly how much they’re enjoying themselves, so neither party gets a clear signal back. People are also reluctant to show too much interest in someone new, in case it isn’t returned, which mutes whatever warmth they do let through. And conversation itself is demanding enough — tracking what to say next, managing turn-taking, reading tone — that people often miss the signs of liking their partner is actually sending, because they’re too busy monitoring their own performance to notice.

Following it past a single conversation

A single awkward first exchange might be explained away as first-meeting nerves, so the same research team went looking for the effect somewhere conversations keep happening: a dorm. Working with a separate, larger study of Yale students who had been assigned by the university to live together as suite mates, they surveyed participants five times across an academic year — once shortly after move-in, then again in October, December, February, and May — asking each time how much they liked their suite mates and how much they thought their suite mates liked them.

At every time point except the last one, students underestimated how much their suite mates liked them. The gap didn’t close after the first semester of daily contact, or the second. Offering people a cash incentive to report their honest, most accurate estimate at the final survey didn’t move the number either — accuracy wasn’t being blocked by an unwillingness to admit the truth, since even paid-to-be-accurate participants gave the same low estimates as everyone else.

What finally closed it

By May, after eight months of living down the hall from the same people, the gap was gone: how much students thought they were liked by their suite mates finally matched how much they actually were. Later work on the same phenomenon, including a 2021 follow-up study that tracked engineering teams working on real projects together, found the same underestimation showing up in group settings and workplace teams, with the size of the gap tied to how much people fixated on the worst moments of their own performance rather than how the interaction went overall.

What closes the gap isn’t reassurance or being told the numbers. It’s sustained, ordinary contact — the unglamorous accumulation of a semester’s worth of hallway small talk doing what no single conversation, however good, can do on its own.

Why this matters for anyone building a network of readers

Blogging and newsletter writing run on exactly the kind of short, high-stakes conversations the lab studies used: a first message to another writer proposing a guest post, an introduction at a conference, a cold pitch for a newsletter swap. These interactions carry the same asymmetry that the strangers in Study 1a and Study 2 experienced — both people are trying to read the other’s reaction in real time, both are distracted by their own performance, and both walk away afterward assuming they came across worse than they did. A writer who cold-emails ten other writers in their niche and hears back from three is likely to read the seven silences as a verdict on the pitch itself, when the more mundane explanation, backed by the same mechanism documented in these studies, is that the writers who didn’t reply probably liked the exchange more than they let on and simply got busy.

None of this means every guest-post pitch or panel introduction is secretly going better than it feels in the moment, but the research gives a specific reason to discount the read. Psychologist Erica Boothby, one of the study’s authors, has discussed the finding as one of several social illusions that distort how people believe they come across to others — distortions that, unlike a genuine cold shoulder, tend to soften with nothing more than time and repeated contact. The bloggers driving home from that conference, running the conversation back and coming up short, are working from the same miscalibrated instrument everyone else in these studies was. The instrument gets less wrong the longer the two people keep talking.

The post People leaving a conversation with someone new consistently underestimate how much that person actually liked them, and the effect held up even when researchers followed college roommates for months after they first met appeared first on The Blog Herald.

Somewhere below your address and your date of birth, it asks for your emergency contact — and it gives you exactly one line. Not ten. Not “list your social circle.” One name: the person who picks up.

Society, in its bureaucratic way, has always known what actually matters when it matters. And psychology has spent more than half a century confirming it — which is worth remembering the next time someone in their 60s gets pitied for having a small circle.

Scientists figured this out in 1968

Long before loneliness became a headline, two researchers named Marjorie Lowenthal and Clayton Haven studied older adults and asked what protected people’s morale through the hard weather of later life — retirement, widowhood, the losses that pile up.

The answer wasn’t social calendars or big networks. It was the presence of a single close confidant. Their study became a classic, and the decades of research it launched kept landing in the same place: having one close confidant is associated with greater psychological wellbeing in older adults, full stop. People with one person who truly knew them weathered losses that flattened the merely popular.

Think about what that finding actually says. It’s not “friends are good,” which everyone knows. It’s that the protective ingredient comes in a unit of one — a threshold, not a scale. One genuine confidant moves the needle enormously. And past that, the returns get strange.

The part nobody expects: the count doesn’t count

Because here’s what modern studies found when they looked closer.

Researchers interviewing adults from their late 50s into their 90s measured both how many confidants people had and how close those relationships were, then checked the results against anxiety and depression. Closeness predicted better mental health, strongly. The number of confidants predicted nothing at all.

Sit with that. Ten shallow confidants delivered no more protection than two deep ones. The system runs on depth, and depth alone — headcount is decorative. (The same study produced a finding so impish it deserves its own sentence: having more children and grandchildren was associated with more anxiety, not less. Being surrounded, it turns out, is simply not the same product as being seen.)

And large circles aren’t just neutral — they carry freight. Research on older adults’ networks finds that bigger networks increase exposure to openness and support, yes, but also to criticism — more people who know your name means more people with opinions about your life and no depth to cushion them. A hundred acquaintances isn’t a hundred small comforts. It’s a hundred small audiences.

The 60-something with three friends isn’t behind. They’re finished.

Which reframes the person this article’s title describes — the one who arrives in their 60s with a circle you could fit in a booth.

They didn’t fail to build a network. In most cases they had the network, decades ago — the packed weddings, the forty-person birthdays, the hundred names. Then they spent thirty years running the longest relationship experiment there is, and watched what every decade confirmed: the crowd was lovely and interchangeable, and the same two or three people kept turning out to be the ones who showed up, kept the secret, noticed the bad week, asked the second question. So they invested accordingly. What looks from outside like a dwindled social life is usually a concluded search — the hundred names audited down to the ones that were ever going to matter, with everything reinvested in the survivors.

That’s not isolation. That’s the finding from 1968, arrived at independently, by living it.

And it’s why the pity is so misplaced. Loneliness was never the gap between you and a crowd — it’s the gap between the connection you have and the connection you need. The person with three real friends and one true confidant has closed that gap completely. Meanwhile plenty of people with three hundred contacts are carrying it around like a stone, because nothing in the three hundred goes deep enough to touch it.

The one honest caveat

Now, the part a fair version of this article has to say: a small circle is only the happy story when there’s a confidant in it.

The research divider was never circle size in either direction — big isn’t automatically lonely, and small isn’t automatically wise. The divider is whether anyone truly sees you. A small circle with a confidant is the winning configuration the studies keep endorsing; a small circle without one — where the shrinking wasn’t chosen but happened, through loss and drift, leaving nobody who knows the real news — is precisely the situation the confidant research warns about. Studies tracking older adults over time found that losing a confidant measurably raised depressive symptoms, and gaining one measurably lowered them.

So the test for your own circle, or your parents’, isn’t a headcount. It’s one question: is there a person who actually knows how things are? If yes, the small circle is a fortress, whatever it looks like from outside. If no — that’s the one gap worth working on, and the research above doubles as good news: the benefit of a confidant kicks in whenever the confidant does. It’s never too late to fund the position.

The emergency contact test

So here’s the takeaway, sized for a wallet.

Stop auditing your social life by width — yours or anyone else’s. The hundred people who know your name are pleasant, and they are not the infrastructure. The infrastructure is the emergency-contact line: the one who picks up, who knew before you said it, who gets the unedited version.

If you have that person, you’re not lonely, whatever your calendar looks like — you’re exactly what fifty-plus years of research describes as protected. Tell them something true this week; that’s how the position stays funded.

And if you’re lucky enough to be someone’s one line on the form — the name they wrote without hesitating —

then you already know the whole secret of this article from the other side:

it was never about being known by a hundred people. It was about being irreplaceable to one.

The post Psychology says people who reach their 60s without a large circle of friends aren’t lonely — they’re the ones who figured out the one relationship truth that emotionally intelligent people swear by, which is that one person who truly sees you is worth more than a hundred people who only know your name appeared first on The Blog Herald.

“No, I don’t think I will.”

No apology attached. No elaborate excuse. Not hostile — she’d say it pleasantly, the way you’d decline a second helping. But the family reeled anyway, because this was the woman who had hosted every Christmas for thirty years, driven everyone to every airport, listened to every crisis at any hour, and absorbed every awkward silence at every dinner like it was her job.

The verdict came quickly: she’s gotten difficult. Prickly. A bit bitter, maybe.

She hadn’t. She’d retired — from a position nobody ever officially gave her, doing work nobody ever officially counted. And what happened next in that family is the same thing psychology keeps finding when it looks closely at people who supposedly “harden” with age.

The invisible job

To see what actually changed, you have to see what she was doing before, because it never looked like anything.

Absorbing other people’s discomfort is a real job with real hours. It’s laughing at the joke that wasn’t funny so the teller doesn’t feel it die. Taking the worst bedroom, the middle seat, the inconvenient date, so nobody else has to negotiate. Listening to the monologue for the fourth time. Smoothing the argument at dinner by swallowing your own point. Saying “it’s fine, honestly” while rearranging your entire week — because your discomfort, handled privately, is cheaper for everyone than their disappointment, handled publicly.

Do that for forty years and something interesting happens: it stops registering as generosity and becomes the furniture. Nobody thanks the load-bearing wall. The family doesn’t experience Aunt-hosting-again as a gift; they experience it as how Christmas works. Which means the day she stops, there’s no vocabulary for it except the accusatory kind. Nothing was ever named as her labor, so its withdrawal can’t be named as retirement. It can only be named as a personality problem.

Science has watched this happen — on purpose

Here’s where the research gets genuinely interesting, because psychologists have studied older adults’ shrinking social worlds, and the lazy assumption — that people lose relationships as they age — turns out to be wrong in the most telling way.

In one of the classic studies, researchers tracked older adults over years and then asked about the people who had disappeared from their networks. The main reason relationships ended wasn’t death, distance, or drift. It was that the person was no longer interested — the ties were actively eliminated. The researchers’ conclusion is almost blunt: networks don’t shrink by default in later life; they get pruned, deliberately, down to what’s emotionally worth it. And the same body of work found the payoff: people whose circles narrowed this way reported better emotional experience in daily life, not worse.

Read that as a family member and it stings a little. Read it as a psychologist and it’s a picture of skill: with less life left to spend, older people quietly audit where it goes — and the draining acquaintance, the one-way friendship, the gathering that runs entirely on their absorption, don’t survive the audit. Not because the heart hardened. Because the accounting finally got honest.

Why it looks like bitterness from the outside

So why does this healthy pruning read, from the receiving end, as she’s gotten difficult?

Because a subsidy you’ve stopped noticing feels, when it ends, exactly like an attack. The nephew who always got picked up from the airport experiences the first “no” as coldness — not because anything cold occurred, but because his comfort had been silently underwritten for so long that its full price appearing on his side of the table feels like a fee being imposed rather than a discount expiring. Multiply that across a whole family, a whole friend group, and you get the consensus: harder to be around.

Notice what the phrase actually means, though. Harder to be around, in these cases, almost always translates to: harder to use frictionlessly. The person is usually just as warm to the people who never ran a tab. It’s specifically the heavy withdrawers who find the new atmosphere chilly — which is itself the diagnostic.

And to be fair to everyone: this isn’t villainy on the family’s side either. They genuinely never saw the labor, because the whole point of absorption done well is that it’s invisible. They’re not mourning her kindness. They’re mourning a service they never knew was being rendered.

The real test: bitter or boundaried?

None of this means every prickly older person is a misunderstood saint — bitterness is real, and it’s worth knowing the difference, because they look similar for about five minutes and then not at all.

Bitterness is aimed. It travels toward people: the digs, the scorekeeping, the cold recitations of old debts. It wants the other person to feel something bad. And crucially, bitter people are usually miserable company to themselves too — the resentment runs at home, alone, with nobody to absorb it.

Boundaries are placed. They don’t travel anywhere; they just stand there. “No, I don’t think I will” contains no attack, no lesson, no bill for the past — and the person saying it is typically more pleasant inside the lines they’ve drawn, not less, because they’re no longer performing through gritted teeth. My aunt’s Christmases, the ones she chose to attend after her retirement from hosting, were reportedly her warmest in years. She wasn’t harder to be around. She was, for the first time, actually around — present as herself instead of as the household’s shock absorber.

That’s the tell, if you’re trying to read someone: bitterness subtracts warmth everywhere. A boundary just moves where the warmth is spent.

If the “difficult” person is you

And if you’ve felt this shift in yourself — the dwindling willingness to eat discomfort for the table — the research above is your permission slip: it’s not decay, it’s calibration, and it’s associated with feeling better, not worse.

Just keep the two halves together. Drop the absorbing; keep the kindness. Decline the airport run warmly. Skip the draining gathering and genuinely delight in the one you attend. The goal was never to make people earn you.

It was to stop paying for everyone’s comfort with the only currency that was ever really yours —

and to finally find out who stays at the table when nobody’s being absorbed.

Usually, it’s the ones who were never running a tab. And they were the whole point all along.

The post Psychology says people who become harder to be around as they get older aren’t becoming mean — they’re becoming less willing to absorb other people’s discomfort at the expense of their own, and that’s a skill most people mistake for bitterness appeared first on The Blog Herald.

What the note actually recommends

The note itself is narrower than the online conversation around it suggests. Ibarra isn’t telling people to quit Substack or stop publishing; he’s recommending a private habit — pick a topic, write about it daily, skip the optimization. “Pick a topic you love and write about it,” he wrote. “Without AI. Without templates. Without the feeling that you need to go viral. Just you, your thoughts, and a piece of paper.” The claim that writing this way “will literally rewire your brain” is the one line in the note that outruns what any neuroscience actually shows, and it is worth separating from the more defensible part of his point: writing without an audience in mind changes what actually ends up on the page.

The research behind why that might be true

That narrower claim has real backing. In a study of 72 people already active in creative writing, psychologist Teresa Amabile had participants write two poems, with a brief questionnaire in between that primed either intrinsic reasons for writing — enjoyment, self-expression, the pleasure of the work itself — or extrinsic ones, such as publication, recognition, or career advantage. There was no difference between groups on the first poem. On the second, the poems written after five minutes ranking extrinsic reasons were rated meaningfully less creative by a panel of twelve poets, several of them published: a mean creativity score of 15.74 out of 40, against 18.78 for the control group and 19.88 for the group primed with intrinsic reasons, a statistically significant drop (F(2,69) = 4.66, p < .013). Amabile herself called the size of the effect “particularly impressive” given that her subjects were already experienced writers with strong existing interest in the craft — this wasn’t a case of talking novices out of a hobby, it was a temporary five-minute reframing measurably dulling the work of people who wrote for hours every week. Amabile’s 1985 paper opens with a line from Sylvia Plath’s journals describing her own writer’s block as a product of wanting “acceptance” from “editors and publishers and critics and the World” — a want she said “corrupts my nunnish labor of work-for-itself-as-its-own-reward.” Ibarra’s note is a compressed, non-academic version of the same argument: bring the audience into the room, and the work changes shape around it, even when the audience is only being imagined.

Why five minutes was enough

What makes Amabile’s result useful here is what it did not require. She didn’t need to impose a deadline, a reward, or an evaluator watching over anyone’s shoulder — the studies that had already shown creativity declining under those conditions. All she did was have subjects spend five minutes reading and ranking a list of extrinsic reasons for writing, making that motivation briefly salient before they picked up a pen. The mechanism wasn’t a constraint on the work; it was a shift in what the writer was thinking about while doing it. A feed that displays a like count, a restack count, and a reply count under every post someone writes is, functionally, a machine for making extrinsic reasons salient on a loop, with no single deadline or reward required to do it. The platform doesn’t have to force anyone to chase an audience. It just has to keep showing them the number.

The mechanism working in the other direction

The friction is that the note exists inside a system built to reward exactly the orientation it argues against. When Substack launched Notes in 2023, its founders wrote that they wanted to build something that “fosters deep connections and quality over shallow engagement and dopamine hacks,” turning away from “the attention economy.” In the same announcement, they described restacking — the exact mechanic that moved Ibarra’s note past nine hundred shares — as giving writers “a powerful growth channel as these interactions find new audiences.” Both descriptions are true of the same feature. A note about writing for no one still travels through a tool explicitly built to convert attention into subscriptions. Nothing about posting the advice to Notes, where it can be restacked, replied to, and counted, sits outside the audience-chasing behavior it describes. The same pattern predates Substack by two decades. Blog platforms have displayed comment counts, subscriber numbers, and page-view stats since the format existed, and the effect Amabile measured in a lab in 1985 does not require a restack button specifically — any dashboard that keeps a running tally of who is paying attention will do the same work of keeping the audience salient, on any format, in the background of every post.

What the replies gave away

The replies underneath the note make the same tension visible from the other side. Naman Gupta wrote that “writing every day is like having a conversation with yourself. Over time, the conversations get a lot more interesting” — a reply that picked up its own likes and its own sub-replies. Ashbi Abraham added that “expressive writing has been shown to reduce stress markers while helping us regulate our emotions,” delivered in the register of a pitch rather than a private observation. One writer, posting under the name Restitched Surgeon, described starting a Substack days earlier and framed the daily-writing habit as something already worth narrating publicly, describing the habit, four days in, as something already unraveling into its own project. None of this is hypocrisy. It is evidence for how hard the audience orientation is to switch off once a platform is watching and counting, even in the specific act of agreeing that it shouldn’t matter.

The interesting finding is not that the advice was right, or that its own popularity undercut it. It is that both are true in the same note, at the same time, without contradiction. Amabile’s writers produced worse poems after five minutes spent thinking about publication and recognition. Ibarra’s readers produced a wave of replies and restacks after reading a note telling them not to think about publication and recognition. The habit the note recommends — writing daily, for its own sake, off the feed — is the one thing none of its engagement metrics can actually measure.

The post A Substack post telling writers to stop chasing an audience and just write for themselves picked up close to three hundred comments and nearly a thousand reposts, which is either proof the advice works or proof that no one, including the people agreeing with it, can actually stop chasing an audience appeared first on The Blog Herald.

The bills are manageable. The family calls. Health is stable enough. There is affection, comfort and no immediate crisis.

And yet something feels absent.

Often, what is missing is not love. It is consequence. Nobody is waiting for your judgement. Nobody needs the thing you know how to do. The people around you are kind, but they have quietly reorganised life so that you are always the person being helped and never the person whose help makes a difference.

Researchers studying later life use words such as mattering, purpose and generativity to describe parts of this experience. They are not identical, but they point towards a common human need: to feel that your presence adds something to another person’s world.

No serious study proves that usefulness matters more than health, love or financial security. Those foundations remain important. What the research suggests is that receiving care does not entirely replace the experience of contributing care, knowledge, attention or effort.

Being loved tells us we belong. Being useful tells us that our belonging has movement in it.

Being loved and being needed are different experiences

Love flows towards us. Usefulness flows through us.

That may sound like a poetic distinction, but it describes something practical. A person can be surrounded by affection while feeling increasingly peripheral to the decisions, routines and small responsibilities that once structured life.

This often becomes visible after retirement. Work can be exhausting and imperfect, but it also provides repeated evidence that someone expects us to arrive, solve a problem or carry a piece of responsibility. When the role disappears, the salary is not the only thing that goes. A daily answer to the question “Why am I here?” can disappear with it.

The same shift can happen when adult children become protective. They take over the cooking, driving, planning and practical decisions out of love. Each act is reasonable on its own. Taken together, however, they can turn an older person from a participant into a recipient.

There is a deep difference between having people around and having a place in what those people are doing.

The research is stronger than a comforting idea

One of the clearest studies followed adults aged 70 to 79 in the MacArthur Study of Successful Aging. Participants were asked how useful they felt to friends and family, then followed for seven years.

Those who said they rarely or never felt useful were more likely to develop greater disability or die during the follow-up than those who frequently felt useful. The association remained after the researchers accounted for health, behaviour, social circumstances and other psychosocial factors.

That does not establish a simple cause. People whose health is beginning to decline may feel less useful before the decline becomes obvious. Feeling useful may also lead to more movement, social contact and daily structure. The relationship probably travels in both directions.

Still, similar patterns appear elsewhere. A ten-year analysis from the national MIDUS study examined generativity among adults aged 60 to 75. Its measure included whether people felt needed, had skills to pass along, influenced others positively and were approached for advice. Higher perceived generativity was linked with more favourable later health trajectories, although the authors were careful about the limits of observational data.

The interesting detail is the word perceived. It was not enough to count volunteer hours or family duties from the outside. The studies asked whether the person themselves felt that their contribution mattered.

Researchers call this generativity

Generativity is the concern for supporting people who will continue after us. It often means guiding a younger generation, but it is broader than parenting or grandparenting.

It can be teaching someone a skill, passing on hard-earned judgement, tending a community, helping a neighbour or creating something another person can use. At its core is the feeling that part of what you have learned is still travelling forward.

A small randomised study gives this idea more weight. Seventy-three women aged 60 to 86 were assigned either to write weekly about their life experiences and share advice, or to write neutral descriptions. After six weeks, the generativity group showed improvements across several measures, including greater participation in social activities, less psychological distress, more positive expectations about physical ageing and reduced pro-inflammatory gene expression.

This was a preliminary intervention in a small, selective group of women with computer access. It should not be treated as a universal prescription. But it shows that contribution does not have to mean taking on a demanding job or committing to fifteen hours of volunteering each week.

Even organising experience into words that could help someone else changed something.

Why usefulness can reach places comfort cannot

Comfort removes friction. Usefulness creates agency.

When another person genuinely relies on something we offer, we receive several messages at once. I can still affect the world. My experience has not expired. I am not merely being managed. There is a future event that will be different because I took part in it.

This may explain why helping can add something that companionship alone does not. A large longitudinal study of Americans over 50 found that people providing substantial informal help to friends, neighbours or relatives later reported more positive affect, optimism, purpose and sense of mastery, alongside several favourable physical and social outcomes.

Again, this was an association. Healthier, more socially connected people have more capacity to help. But the pattern is consistent with the possibility that contribution reinforces the very resources that make continued contribution possible.

There is also a reciprocal quality to it. Research following older adults in Miami found that both feeling loved and expressing compassionate love towards others predicted greater positive affect. Wellbeing was not only about what entered a person’s life. It was also connected to what they could still send out.

The wrong lesson is that older people must earn their place

This research can be interpreted badly.

An older person does not need to remain productive in order to deserve love, dignity or care. Usefulness should never become a polite word for unpaid obligation. A person overwhelmed by caregiving, family demands or physical limitations is not failing because they need rest.

There is also no joy in a task that everyone knows was invented merely to keep someone occupied. Token responsibility can feel more diminishing than no responsibility at all. People can usually tell when their contribution has no consequence.

The healthier principle is not “keep older people busy.” It is “do not assume they have nothing left to give.”

That means asking rather than assigning. It means respecting a refusal. It means noticing what a person wants to contribute, not forcing them back into the role they performed at 45. Usefulness in later life should be chosen, proportionate and real.

It also does not require complete independence. Human beings are interdependent at every age. Someone can need help with transport and still be the person whose advice holds a family together. They can receive assistance with meals and remain the one who knows how the recipe should taste.

Small usefulness is still real usefulness

We tend to imagine purpose as something large: leading an organisation, building a legacy, mentoring dozens of people.

Late-life usefulness is often quieter.

It is a grandchild calling because you are the only person who listens without rushing to fix them. It is a neighbour trusting you with a spare key. It is showing someone how to fold dumplings, repair a hinge, prune a plant or recognise when an apology is overdue.

It is being asked, “What do you think?” and knowing the answer will be taken seriously.

These moments do not replace medical care, money or loving relationships. They give those things somewhere to lead. Health preserves the ability to participate. Financial security creates room to choose. Love makes contribution safe rather than compulsory.

The deepest lesson in the research is not that joy comes from constant activity. It is that people want to remain inside the circulation of life. We want care to move in both directions. We want our accumulated experience to have a destination.

To be loved is to know that someone would notice if you were gone.

To feel useful is to know that something is different because you are still here.

The post Researchers studying late-life wellbeing have found that one of the deepest sources of joy is not just being loved, healthy, or financially secure — it is still feeling useful to someone else appeared first on The Blog Herald.

What actually happened

Nazir was named the 2026 Commonwealth Short Story Prize’s regional winner for the Caribbean, one of five regional winners named that year alongside Africa, Asia, Canada and Europe, and the Pacific, for a story titled “The Serpent in the Grove.” The prize, judged by a panel chaired by novelist Louise Doughty, publishes its winning entries in Granta. Within weeks of the announcement, Mollick’s AI-detection claim, made using a tool called Pangram that its developers say is roughly 99% accurate, put Nazir’s win at the center of a dispute the prize had not anticipated having to manage.

What the evidence actually consists of

The accusation rests on two separate strands. The first is Pangram’s automated flag, which, according to reporting in The Week, was not limited to Nazir’s story alone; the same tool also flagged two other winning entries from the same year’s prize. The second strand is more impressionistic: readers who began scrutinizing the story after Mollick’s post pointed to specific stylistic tics, including vague intensifiers and a workshop-polished tone some described as reading like an MFA parody, as circumstantial signs of AI involvement. Neither strand is a confirmed finding. An AI-detection tool’s output is a probabilistic flag, not a determination, and stylistic resemblance to workshop prose is a common feature of contemporary literary fiction that AI didn’t invent.

What Nazir says

Nazir has denied writing the story with AI, saying it drew directly on his own childhood memories. Granta, which publishes the prize’s winning stories, said it was investigating the allegation, a detail also confirmed by The Free Press‘s reporting on the prize’s judging structure.

What the Commonwealth Foundation actually said

The idea that the prize committee has said nothing is, as of this writing, out of date. The Commonwealth Foundation’s own statements show its director-general, Razmi Farook, issued a public response on May 19, defending the judging process, stating that the Foundation does not run AI-detection checks on unpublished fiction submissions, citing consent and ownership concerns, and noting that every shortlisted writer had personally denied using AI. Farook also called detection tools “not infallible” and said the Foundation operates “on the principle of trust.”

Three days later, on May 22, the Foundation issued a second statement saying it “takes seriously” the allegations and was conducting a review of “all available evidence,” while stopping short of reversing or altering Nazir’s win. A further update on the review’s progress followed in June. What hasn’t happened, months on, is a resolution: no reversal, no confirmation, no public timeline for when the review concludes.

Why the silence that matters is narrower than it looks

The Foundation hasn’t stayed quiet, but it has stayed inconclusive, which is a different and in some ways more telling position. Defending the judging process while simultaneously opening a review of “all available evidence” isn’t a contradiction so much as an organization trying to hold two things at once: confidence in a system built on trusting its writers, and an acknowledgment that the trust itself is now the thing under public examination. A prize built on the assumption that a shortlisted writer’s word is sufficient has run directly into a case where a growing number of outside observers no longer consider that assumption sufficient, and the Foundation’s own statements show it hasn’t decided which side of that gap it’s going to stand on.

The same trust problem showing up across publishing this year

The Commonwealth Foundation’s position, that it trusts writers’ word over automated detection because detection tools aren’t infallible, echoes a justification that has surfaced elsewhere in publishing this year: trade bodies offering “human authored” labels that check membership rather than manuscripts, and retailers proposing disclosure standards that check a cover’s wording rather than a book’s origin.

Each institution has independently arrived at the same practical position, that verifying authorship at the level of the actual text is not currently something they’re equipped to do at scale, and each has substituted some form of self-report or institutional trust in its place.

The Commonwealth Foundation’s version of that substitution is now the one attracting the most public scrutiny, because unlike a retailer’s shelf or a trade body’s membership roll, a literary prize’s entire value depends on the credibility of exactly the claim currently in dispute.

What stays unresolved

Months after the accusation, Nazir’s name is still attached to the win, the review is still open, and no public evidence beyond a detection tool’s probabilistic flag and readers’ stylistic impressions has surfaced to settle the question either way. That’s less a failure of process than an accurate reflection of how little any institution, prize committees included, currently has to work with when an authorship accusation lands after a decision has already been made public.

Whether the review changes anything will likely say more about how literary institutions plan to handle the next version of this dispute than it will resolve the current one. A finding against Nazir would be the first time a major prize reversed a win over an AI-authorship dispute; a finding that clears him would do little to settle the broader question, since the tools used to raise the accusation remain exactly as unproven as the Foundation itself has said they are.

For now, the only settled fact is the one the Foundation itself keeps repeating: it doesn’t yet know, and neither does anyone reading the story from the outside, relying on the same probabilistic tools and stylistic hunches that started the dispute in the first place.

That distinction is worth keeping straight too, months into a dispute that still hasn’t produced anything firmer than suspicion on one side and denial on the other.

The post A Commonwealth Short Story Prize winner was accused of using AI to write his entry, and months into a review, the prize committee still hasn’t said whether that changes the win appeared first on The Blog Herald.

A text that stays still while the reader doesn’t

Literary scholar Patricia Meyer Spacks spent a year systematically rereading books from across her life for her book On Rereading, and found her reactions split in two directions. Some books, she found, “remained absolutely and completely the same” on a second or third pass, as reported in a Washington Post review of the book. Others reversed her original verdict entirely, a novel she’d once dismissed reading as newly essential, one she’d loved reading as suddenly thin. Spacks described rereading as “a way to evoke memories (not only of the text but of one’s life and of past selves).” The distinction that matters is between the two kinds of reread: the one that confirms the text hasn’t changed, and the one that reveals the reader has.

Why this isn’t the nostalgia it looks like

Psychologists Constantine Sedikides, Erica Hong and Tim Wildschut, the researchers most closely associated with the empirical study of nostalgia, are also explicit in their own work that nostalgia is only one of several distinct mechanisms people use to maintain what they call self-continuity, the felt sense of connection between a past self and a present one.

In a 2023 review, they separate nostalgia, a warm, social, emotionally self-contained state, from what they call autobiographical reasoning, the more deliberate, comparative act of relating a specific past experience to a person’s present sense of who they are. Rereading a book from years ago sits closer to the second category than the first. The reread isn’t primarily about feeling warmly toward the past; it’s an active check on the distance between then and now.

What happens when the reread finds an unwelcome answer

Not every reread confirms growth in a direction the reader wanted. Spacks herself found books she’d once loved thinning out on a later pass, and that outcome is data too, just less comfortable data.

A reader who discovers a once-favorite novel now reads as shallow isn’t discovering the book was always overrated. They’re discovering that the version of themselves who loved it responded to something they no longer respond to in the same way, which is a harder thing to sit with than simple nostalgia would ever require, and part of why the practice can’t be fully explained as comfort-seeking. Comfort-seeking wouldn’t risk finding out the thing you loved doesn’t hold up.

The narrative work a reread is actually doing

That checking function is close to what psychologists Dan McAdams and Kate McLean describe as narrative identity, the ongoing, revised life story a person builds by integrating past episodes into a coherent account of who they currently are. Their 2013 paper describes narrative identity as “a person’s internalized and evolving life story, integrating the reconstructed past and imagined future to provide life with some degree of unity and purpose.” A book reread at intervals functions as a kind of checkpoint inside that ongoing story, a stable text against which the same reader can measure how their interpretation, and by extension their own thinking, has moved.

What deliberate rereading actually produces

Consumer researchers Cristel Russell and Sidney Levy studied this behavior directly, rather than inferring it from adjacent theory. Their 2012 study of people who deliberately reread books, rewatched films, and revisited places found that the practice “facilitate[s] existential understanding, as the linkages across past, present, and future experiences enable an active synthesis of time and promote self-reflexivity.” Participants in their research described gaining “an enhanced awareness of their own growth in understanding and appreciation through the lens of the reconsumption object.” That’s a specific, documented outcome, not a side effect: people who reread on purpose are using the fixed object to notice their own change, and reporting that noticing as the actual value of the practice.

What the shelf is actually for

None of this means comfort-seeking never plays a role; some rereads really are just about wanting something familiar on a hard week. But the specific pattern of returning to the same book every few years, tracking a different reaction each time, describes something more structured than comfort. The book stays the same so that the difference in the reader’s response has somewhere stable to show up. Read once, a book is an experience. Reread every few years across a life, it becomes something closer to an instrument, a fixed text a reader keeps returning to specifically because it’s the one thing in the equation guaranteed not to have changed.

The next reread won’t tell that reader whether the book got better or worse. It will tell them something about where they’ve moved to since the last time they stood in front of it.

That risk is precisely what separates rereading from other, safer forms of revisiting the past, a photo album, a familiar song. Those formats mostly confirm what a person already remembers feeling. A book long enough and complex enough to reward close attention can return a different verdict on a second or third pass, which is what makes it useful as a measuring instrument in the first place. An instrument that always confirms the same reading wouldn’t be measuring anything.

Which is also why the practice tends to survive scrutiny that comfort-seeking wouldn’t: a reader who keeps returning to a book that sometimes disappoints them isn’t chasing a feeling. They’re keeping an appointment with whoever they turn out to be by the time they get there.

That’s a small distinction, but it’s the one worth keeping straight.

The post People who reread the same book every few years aren’t stuck in nostalgia, they’re checking in with the version of themselves who read it the first time appeared first on The Blog Herald.

What the data actually shows

Economists Imke Reimers and Joel Waldfogel, in a National Bureau of Economic Research working paper first circulated in January 2026 and revised in May, used Amazon release data to track monthly new ebook volume. Their finding, summarized in the NBER Digest: monthly new ebook releases averaged around 100,000 in the 2020-2022 period and rose to more than 300,000 a month by late 2025. That’s a monthly rate that roughly tripled over several years, not a single before-and-after cliff the day ChatGPT launched, but the timing and shape of the increase track closely enough with the availability of consumer AI writing tools that the researchers treat generative AI as the primary driver.

What tripling actually means for a storefront

Three hundred thousand new titles a month is not a marginal increase for a retailer to absorb. It’s a volume that changes what “self-published” can mean as a category, from a slower stream of individually written manuscripts to a flow that includes, alongside genuine single-author work, output assembled at a pace no unassisted writer can match. Amazon’s own catalogue infrastructure, built for the earlier, slower rate, was not designed around the assumption that any single account might attempt to publish dozens of titles in a short window.

What Amazon actually changed

The company’s concrete response, confirmed by Publishers Weekly, was a cap introduced in September 2023 limiting authors to three new title uploads per day through Kindle Direct Publishing. That’s a throttle on the rate of publication, not a check on how any individual manuscript was produced, and it was adopted specifically in response to a wave of low-effort AI-generated content flooding the store faster than any review process could keep up with.

The identity check that exists, and what it doesn’t check

Separately, KDP’s own help documentation confirms the company does require identity verification, government ID and a selfie, for some accounts, though Amazon does not publish the specific criteria that trigger it. A widely repeated claim that this kicks in automatically the moment an account crosses a specific number of published titles could not be confirmed anywhere in Amazon’s own documentation or in trade coverage from Publishers Weekly or the Authors Guild, and is best treated as an unconfirmed figure circulating alongside the real policy rather than the real policy itself. What is confirmed is narrower: Amazon checks who a person is in some cases, and throttles how fast anyone can publish. Neither check reads a single word of what’s actually inside the books being uploaded.

The same gap as everywhere else in this story

That’s a familiar shape by now. A trade body’s authorship label checks membership, not manuscripts. A retailer’s disclosure policy checks a cover’s wording, not whether the wording is true. A storefront’s identity check confirms a person exists behind an account, not what that person actually wrote. Every mechanism publishing has produced this year to address AI-generated content checks something adjacent to the actual question, was this specific text written by a person, because that is the one thing none of these systems are built to answer at the volume they now have to operate at.

What the numbers actually settle

The tripling in monthly ebook volume doesn’t prove that any specific book was AI-written, and neither Amazon’s upload cap nor its identity verification was designed to make that determination. What the numbers settle is narrower and more useful: the rate of production changed enough, and fast enough, that the systems built to manage a slower self-publishing pipeline are now running well behind the volume actually moving through it, and the fixes deployed so far manage the rate and the applicant, not the writing itself.

What readers are actually left with

None of this gives a reader browsing a self-publishing storefront any reliable way to tell which of the three hundred thousand monthly new titles were written by a person sitting with the material for months and which were assembled at a pace that rules that out. Amazon’s cap slows the rate; its identity check confirms an applicant; neither one functions as a quality or authorship filter, and neither was built to. That leaves readers doing the same thing the honour-code label and the disclosure debate elsewhere in publishing this year have left them doing: relying on reviews, sample chapters, and an author’s existing reputation to do the filtering no system-level check currently performs at this volume.

That’s not a criticism of Amazon specifically. No retailer operating at this scale has built a system that reads manuscripts before listing them, and none of the fixes proposed anywhere in publishing this year, an honour-code label, a disclosure standard, an identity check, a daily cap, actually does that either. The volume increase didn’t just add more books to the shelf. It made clear how much of what already sat there was never being checked in the way readers had quietly assumed it was.

For a reader trying to decide what to trust on that basis, the honest answer is that the volume of new titles has already outpaced any system built to vouch for them, and no cap or identity check currently being deployed changes that math.

Whatever comes next, whether that’s better detection tools, stricter caps, or some version of the honour-code labelling now being tried elsewhere in publishing, it’s likely to arrive after the volume problem it’s responding to, not ahead of it, which has been the pattern in every corner of this story so far.

Until then, the gap between how many books are being published and how many are being meaningfully checked is the actual story, and it’s a gap that existing policy tools were not built to close.

The post Self-publishing produced roughly one hundred thousand ebooks a month before ChatGPT and more than three hundred thousand a month by late 2025, and Amazon has responded with upload caps and identity checks rather than checking any book’s content appeared first on The Blog Herald.

The usual explanation offered is perfectionism, the sense that it still isn’t good enough. The more accurate explanation, according to research on how people relate to things they’ve made, is closer to reluctance to lose something rather than doubt about its quality.

What psychological ownership actually names

Organizational psychologists Jon Pierce, Tatiana Kostova and Kurt Dirks developed the concept of psychological ownership specifically to describe a felt sense of possession that doesn’t require legal or formal claim to exist. In their foundational 2001 paper, they define it as the state in which a person feels a target, an object, an idea, a piece of work, is “theirs,” full stop, independent of who legally owns it.

Their 2003 follow-up notes that “psychological ownership can exist in the absence of legal ownership, and vice versa” — a manuscript is legally the author’s from the moment it’s written, but the felt sense of ownership is a separate, psychological fact that can be stronger or weaker independent of the legal one, and it’s that felt sense, not the legal fact, that governs how reluctant someone is to hand the thing over.

Why the self is part of the object

One of the three roots Pierce, Kostova and Dirks identify for psychological ownership is self-identity: ownership functions, in their words, as “a symbolic representation of the self.” That framing changes what’s actually being risked when a piece gets shared. It isn’t only the work being judged. It’s a thing that has been serving, privately, as a stand-in for the writer’s own identity, and showing it to someone else converts it from a private extension of the self into a public object other people now have opinions about. That conversion is the actual event a writer is stalling in front of, not the editing.

What changes at the moment of handoff

A 2024 study by Marianne Sætre and colleagues, examining how psychological ownership affects the handoff of creative and technical work between people inside organizations, found that ownership specifically, not competence anxiety, was what drove resistance at the exact point a piece changed hands. The resistance clustered at the handoff itself, not earlier in the process when the work was still being actively developed and evaluated. That timing detail matters: if the holdup were really about quality, it would show up throughout drafting, whenever the writer judged the work harshly. Instead it concentrates at the specific moment the work stops being exclusively theirs, which points toward ownership as the mechanism rather than ongoing quality doubt.

A different engine than perfectionism

This is a meaningfully different mechanism from perfectionism, and the distinction holds up under direct comparison. A 2023 study on perfectionism-driven hesitation found its engine runs through fear of failure, typically seeded by internalized external expectations, a supervisor’s standards, a parent’s, an imagined critic’s. That’s a judgment-based mechanism: the fear is about being found lacking by someone else’s measure. Psychological ownership doesn’t require an external judge at all. Its closest economic cousin is the endowment effect, documented by Daniel Kahneman, Jack Knetsch and Richard Thaler in a classic 1990 experiment, which found that mere possession inflates how much someone values a thing, independent of its actual quality. People asked to sell an object they’d just been given demanded far more than people asked to pay for the same object were willing to offer. Possession itself does the inflating. Nothing about the object’s quality needs to be in question for that effect to hold.

What the delay is actually protecting

None of this means the delay is irrational or that ownership should simply be overridden. It means the thing being protected isn’t the piece’s reputation. It’s the writer’s exclusive relationship to it, the fact that, for as long as it goes unshared, the work still belongs to exactly one person’s private sense of who they are. The moment it’s sent, that stops being true, whether or not a single word gets criticized. Calling that perfectionism blames the writer for a quality judgment they may not even be making. What’s actually happening is smaller and more specific: they are holding onto the last version of the piece that was still only theirs.

Why the size of the audience changes the calculation

Psychological ownership theory also predicts something the perfectionism framing doesn’t: that the reluctance should scale with how public the release is, not with how finished the work is. A writer will often show a trusted early reader a rougher draft with less hesitation than they’ll show a polished one to an open audience, even though the polished draft is, by any quality measure, the safer thing to share. That pattern makes no sense if the holdup is fear of criticism, since the trusted reader is just as capable of criticizing as the public is, sometimes more bluntly. It makes complete sense if the holdup is about how many people are about to start having their own relationship to something that used to belong only to the writer. A small, trusted audience barely dilutes that exclusivity. A public one ends it outright.

That distinction has a practical use for any writer trying to figure out why a particular piece is stuck. If a trusted reader could see it tomorrow with no hesitation but a public post is stalled indefinitely, the holdup was never really about whether the piece was good enough. It was about exactly how many people were about to stop treating it as private.

Recognizing that doesn’t make the piece any easier to send. It just means the writer isn’t waiting to become more sure the work is good. They’re waiting to become ready to stop being its only reader.

That’s a small distinction with a real effect on how the delay should be treated, by editors, collaborators, and the writer’s own inner critic alike: not as a flaw to be corrected, but as a cost that comes due the moment the work stops being private, and is worth naming honestly rather than mistaking for a lingering doubt about quality.

The post Writers who can’t let a piece go the moment it’s good enough to show someone aren’t perfectionists, they’re holding onto the last place the work still belongs only to them appeared first on The Blog Herald.

What the survey actually asked

The Generative AI and News Report 2025, produced by Felix Simon, Rasmus Kleis Nielsen and Richard Fletcher and published on October 7, 2025, surveyed 12,217 people across Argentina, Denmark, France, Japan, the UK and the US, fielded by YouGov between June 5 and July 15, 2025.

It’s a companion piece to the Institute’s better-known annual Digital News Report, built specifically to track public perception of AI’s role in journalism rather than news consumption habits generally.

The numbers

Fifty-one percent of respondents said they believe generative AI is used “always” or “often” by news media to produce content, according to the report’s full data. That’s a belief about the industry as a whole, not a claim any single respondent is making about a specific article they read that morning, but it puts a majority of the surveyed public in the position of assuming AI involvement is now the default rather than the exception. Asked whether that involvement has made their experience of news better, only 26% said yes, 8% “much better” and 18% “somewhat better.” News media was one of only three sectors in the entire survey, alongside government and politicians, where more people expected AI to make things worse than expected it to make things better.

What “better” was actually being asked about

The report frames this as a comfort gap rather than a knowledge gap: people aren’t confused about what generative AI can do for a newsroom, faster summarization, automated aggregation, translation at scale, they’re specifically unconvinced that any of it translates into a better experience for them as readers. Nieman Lab’s coverage of the same report, written by co-author Felix Simon, notes that this skepticism holds even as personal use of tools like ChatGPT has roughly doubled year over year among the same population. People are adopting the technology individually while remaining unconvinced that its adoption by the institutions producing their news is doing them any favors, which is a more specific and less resolvable finding than generic AI anxiety.

Why news sits with government and politicians

The company news media keeps in this ranking is worth sitting with. The other two sectors where pessimism outweighs optimism are government and politicians, institutions people already have structural reasons to distrust independent of AI. News media landing in that group, rather than with sectors like healthcare or education where the survey found more balanced or positive expectations, suggests the skepticism isn’t a generic technology worry. It’s specific to institutions the public already suspects of not fully acting in their interest, and AI adoption is being read through that pre-existing lens rather than evaluated on its own terms.

What the gap actually represents

A newsroom can adopt generative AI for entirely defensible reasons, cost pressure, speed, coverage of stories that would otherwise go unreported, and still run directly into a readership that has already decided the outcome will be worse regardless of the reasoning. That’s a harder problem than a disclosure label solves, because the report’s own data shows the belief that AI is already pervasive is not paired with any expectation that pervasiveness is good news.

Newsrooms weighing how visibly to adopt these tools are not choosing between an informed public and an uninformed one. They’re choosing how to operate in front of a public that already assumes the adoption happened and has already made up its mind about what that means.

A perception that may already be running ahead of the practice

Nothing in the report claims that generative AI actually is used “always or often” across the news industry, only that a slim majority of the public believes it is. Actual adoption rates vary enormously by newsroom, market, and beat, and plenty of reporting involves no generative AI at all.

That gap between perceived and actual usage is its own finding: distrust of a technology’s role in an institution can outpace the institution’s real adoption of it, particularly when the institution in question, like government and politicians, was already a low-trust category before the technology arrived.

Correcting the record on actual usage rates might do less to move opinion here than newsrooms would like, because the report suggests the skepticism isn’t purely a response to what AI is doing, but to who is doing it.

What this means for how disclosure gets read

That has a direct bearing on disclosure policy. A newsroom that clearly labels its AI-assisted content is answering the question “did AI touch this,” which the data suggests readers have mostly already decided the answer to regardless of the label. What the label doesn’t answer, and what the survey suggests actually drives the negative sentiment, is whether the reader trusts the institution’s judgment about when and how to use the tool at all. Disclosure changes what readers know. It doesn’t obviously change the more structural distrust the Reuters Institute data is picking up.

The one number in the report that complicates any tidy conclusion is the adoption rate among the same respondents: roughly double what it was the previous year. People are not opting out of generative AI in their own lives while condemning it in the news. They’re doing both at once, which suggests the objection was never to the technology in the abstract, only to trusting an institution’s judgment about where and how to deploy it on their behalf.

Whether that changes as the technology matures, or hardens further, is likely to depend less on newsroom disclosure policy than on whether the institutions using it can rebuild the baseline trust the survey suggests was already thin before AI entered the picture.

The post More than half the people surveyed by the Reuters Institute now believe AI is routinely used to produce the news, and only about a quarter of them think that has made it any better appeared first on The Blog Herald.

What the working-memory model actually separates

Writing researcher Ronald Kellogg’s model of composition, most recently updated in a 2013 progress report co-authored with Alison Whiteford, Justin Turner, Erin Cahill and Piers Mertens, maps writing onto Alan Baddeley’s architecture of working memory rather than treating it as a single undifferentiated skill. Planning and reviewing draw on what Baddeley called the central executive, the system that manages attention and coordinates other resources. But translating an idea into an actual sentence, the report argues, draws specifically on verbal working memory, the phonological loop, the same system that holds a phone number in mind long enough to dial it. Composing a sentence and monitoring whether it sounds right are not the same operation running twice. They are two different systems, doing two different jobs, at the same time.

Saying it sharpens it, rather than distracting from it

That two-system account gets direct experimental support from research on self-directed speech more broadly. In a 2012 study, psychologists Gary Lupyan and Daniel Swingley found that people who said an object’s name out loud while searching for it visually found it faster than people who searched in silence. Speaking the target’s name aloud wasn’t a lapse in concentration, competing with the visual task. It sharpened the visual task by giving the brain an additional verbal cue to search against. The same logic extends naturally to drafting: saying a sentence aloud gives the writer an auditory version of the sentence to check against the version still forming, which is a resource, not a leak of attention.

What happens when the speech gets blocked

The clearest evidence that this verbal channel is actually load-bearing, rather than an optional habit some writers happen to have, comes from experiments that block it. Durham University psychologist Charles Fernyhough, describing his lab’s work in a Psychology Today piece, reports that articulatory suppression, forcing a person to repeat an irrelevant sound while they try to plan, measurably impairs their planning performance, whether the blocked speech would otherwise have been spoken aloud or silent.

“We do our planning at least partly in language,” Fernyhough writes, “whether it is overt (social speech), semi-internalized (private speech), or the kind that only we ourselves can hear.” Remove access to that channel entirely, in either its spoken or silent form, and the planning gets worse. That’s a hard result to square with the idea that talking through a sentence is idle noise competing with the real cognitive work. The talking, in some functional sense, is part of the real cognitive work.

Speech that builds the sentence as much as it checks it

Writing in Psyche, researcher Nana Ariel extends Lev Vygotsky’s theory of private speech, originally developed to explain why young children narrate their own actions, into adult thinking and writing specifically. Her point goes a step further than monitoring: speaking a half-formed sentence out loud doesn’t just let a writer hear whether it’s right, it actively generates material that wasn’t fully there before the words were spoken.

“Not only does speech retrieve pre-existing ideas,” Ariel writes, “it also creates new information in the retrieval process, just as in the process of writing.” Her practical advice for writers stuck on a blank page is not to sit quietly and wait for the sentence to arrive fully formed, but to keep speaking and writing, on the premise that the generative process needs the verbal channel active to produce anything at all.

Two jobs, one sentence

Put together, the research describes something more specific than the idea that some writers simply think out loud. It describes a sentence being built by one process and checked by another, with speech as the interface between them, sometimes generating the next phrase, sometimes catching that the last one didn’t land.

A writer muttering a line before typing it isn’t failing to concentrate on the page. They’re running the page through an additional system that silent drafting doesn’t have access to, and reaching for exactly the resource the research says composing under pressure actually needs.

Why reading a draft aloud catches what silent proofreading misses

This is also the mechanism behind a piece of editing advice that gets repeated without much explanation: read your draft out loud before you consider it finished. The advice works for the same reason talking through a sentence while composing it works. Silent reading lets the eye skip over a clunky phrase because the brain already knows what the sentence is supposed to say and fills the gap automatically. Reading aloud forces the phonological loop to actually process the sentence as spoken language, in real time, which is where an awkward rhythm or a missing word tends to surface. Writers who dismiss the read-aloud pass as an unnecessary extra step are skipping the one part of editing that uses a distinct cognitive channel than the drafting that came before it.

None of this means every writer needs to start talking through every sentence, or that silence is somehow the inferior mode. Kellogg’s model and Fernyhough’s experiments both describe a verbal channel that can operate covertly, in inner speech, as well as aloud. What the research rules out is the assumption that audible drafting is a lesser, less disciplined version of “real” writing. It’s a visible instance of a process every writer runs, whether or not anyone else in the room can hear it happening.

None of this requires a writer to change how they work. It’s mostly useful as a reason to stop apologizing for it in shared offices and co-working spaces, where muttering a sentence before typing it can look, to someone unfamiliar with what’s actually happening, like a lapse rather than a method.

The post Writers who talk out loud while drafting a sentence aren’t losing focus, they’re running two minds at once, one composing and one already listening back appeared first on The Blog Herald.

What Daunt actually said

Daunt’s full answer, reported by Today.com, was: “I have actually no problem selling any book, as long as it doesn’t masquerade or pretend to be something that it isn’t, and that it has an essential quality to it, and that the customer, the reader, wants it… So as long as an AI-written book says it’s an AI-written book and doesn’t pretend to be something else and isn’t ripping off somebody else, as long as that’s clearly stated and the customer wants to buy it, then we will stock them.” He added that Barnes & Noble carries roughly 300,000 titles, as NBC News also reported, and that “the chances are” some of them are already AI-written without the company’s knowledge, a detail that undercut the tidiness of the disclosure standard he had just proposed. If the company can’t currently tell which books in its own stock were AI-written, a rule requiring future disclosure does nothing about the books already on the shelf.

The reaction disclosure didn’t anticipate

Fortune reported that Daunt’s comments “drew swift ire from social media, including intentions to boycott the company until it changed its AI policy.” The anger wasn’t really a dispute over labelling mechanics, whether a sticker should go on the front or the back, or whether the wording should say “AI-assisted” or “AI-written.” It was a rejection of the premise that an AI-written book belongs on the same shelf as a human-written one at all, provided the packaging is honest about which is which. Daunt’s standard treated AI authorship as a preference to be disclosed, like a spice level or a runtime.

A meaningful share of the reaction treated it as a defect to be kept out of the store altogether. The specific reactions circulating in the days after the interview ran toward calls to boycott the retailer entirely rather than requests for a clearer labelling policy, which is itself telling: people angry about mislabeling generally ask for better labels; people angry about presence ask for removal.

Why disclosure is always the first offer

Executives in Daunt’s position reach for disclosure requirements before outright exclusion for a predictable reason: disclosure is procedurally simple to promise and hard to enforce, while exclusion is procedurally difficult to promise and, once adopted, comparatively straightforward to check against. Telling a supplier “label it” costs a retailer almost nothing to say.

Telling a supplier “we won’t carry it at all” requires the retailer to actually be able to tell the difference at the point of intake, across hundreds of thousands of titles from thousands of publishers and self-published authors, which is a much harder operational problem than printing a sticker.

Daunt’s original answer optimized for what was easy to commit to in a live interview. His revised answer, days later, admitted that the easy version wasn’t going to satisfy anyone.

What disclosure can’t do

This is the gap disclosure-based standards run into whenever they’re proposed as a solution to a values disagreement rather than an information gap. A label works when the objection is “I didn’t know” — tell the customer, and the transaction becomes informed. It doesn’t work when the objection is “I don’t want this sold here regardless of whether I know,” which is closer to what the boycott threats were actually expressing.

No amount of clearer disclosure resolves a disagreement about whether the product should exist on the shelf in the first place.

The scale problem neither answer solves

Even the exclusion standard Daunt described to Fortune, active measures to keep AI-generated books out of the catalogue, runs into the same practical limit he flagged himself in the original interview: with 300,000 titles in circulation, and no reliable way to detect AI-generated prose at intake with certainty, “we don’t knowingly stock them” is a statement about intent, not a guarantee about the shelves.

That’s a narrower gap than the one between disclosure and exclusion, but it’s the same shape of problem: a retailer can commit to a policy without being able to fully verify that the policy is being met, for much the same detection-at-scale reason that led the Society of Authors to settle for an honour code rather than a manuscript review process for its own AI-authorship labelling scheme.

Daunt’s own retreat from his answer

Daunt appears to have registered the difference between what he’d said and what people heard. Days after the original interview, he told Fortune directly: “Our position is that we do not sell AI books, as far as we are aware; we take active measures to exclude all AI-generated books from our online catalogue and never knowingly order any for stocking in our stores… This is a straightforward rejection of AI books.” That is a different policy than the one he described on Today. The first version was a disclosure standard: label it and we’ll sell it. The second is an exclusion standard: we try not to sell it at all. The distance between those two statements, made roughly a week apart by the same executive, is a reasonably precise measure of how quickly the disclosure framing collapsed under public reaction.

What readers were asking for, when they threatened to boycott a bookstore over a disclosure policy, was never going to be satisfiable by better disclosure. They wanted the books gone, not labelled. Daunt’s revised answer suggests he understood that faster than his own initial framing did.

The episode is a useful marker for anyone watching how retailers and publishers respond to AI-authorship questions generally: the first public answer tends to be the procedurally convenient one, and the real position tends to surface only after the convenient answer meets an audience that already knows what it actually wants.

The post Barnes and Noble’s chief executive said he has no problem stocking an AI written book as long as the cover says so, and the backlash that followed suggests most readers wanted something disclosure was never going to give them appeared first on The Blog Herald.

What the Times said it found

The Times’ reporting, as described by CBC News and TechCrunch, centered on textual evidence suggesting substantial portions of Shy Girl were generated by AI rather than written by its credited author, Mia Ballard. Neither outlet had access to the Times’ full methodology independent of the Times’ own account, so the precise basis of the finding is best treated as the Times’ claim rather than an independently re-verified one — but it was specific enough, and public enough, that Hachette acted within 48 hours rather than opening an internal review first.

What “evidence” means when the accuser won’t show its work

Neither CBC nor TechCrunch reports that Hachette was shown the Times’ full analysis before deciding to act, only that the Times contacted the publisher with its findings. That distinction matters for how much weight the accusation can bear on its own. A newspaper’s internal investigation is not the same evidentiary standard as a court finding or a forensic audit commissioned by the publisher itself, and Hachette’s decision to pull the book within 48 hours reads less like the conclusion of a verification process and more like a risk calculation made under public pressure, with the Times’ reporting as the trigger rather than the final word. That doesn’t mean the underlying accusation is wrong. It means the public record, as it stands, contains an allegation and a swift corporate response to it, but not an independently confirmed account of what actually happened during the writing of Shy Girl.

What Ballard actually said

Ballard’s response, quoted by CBC, was narrower than a flat denial of AI involvement in the book altogether: she “denied using AI to write the novel but said it was possible an editor she’d worked with on the self-published version might have.”

In further comments reported by TechCrunch, she said she was pursuing legal action against that person, and described the personal cost of the fallout directly, saying her mental health was “at an all time low” and that her name was “ruined for something I didn’t even personally do.”

The defense she offered was not “this isn’t AI-written.” It was “I didn’t do it, and if it happened, someone I hired did it without my knowledge.”

The distance an editor’s involvement creates

That distinction matters more than it might first appear. A self-published novel typically passes through at least one other set of hands before it reaches readers, and each of those hands can, in principle, touch the actual prose. Ballard’s defense relies on that structure: if an editor she hired ran passages through an AI system during an edit, the resulting text could carry AI’s fingerprints without the credited author ever opening a chatbot herself.

Freelance developmental editors are common in self-publishing specifically because self-published authors don’t have access to an in-house editorial team the way traditionally published authors typically do, which means the self-published-to-traditional pipeline runs through exactly the kind of outside labor Ballard’s defense points to.

That doesn’t make her account true. It makes it structurally plausible in a way it wouldn’t be for a book edited entirely in-house at a traditional imprint from the start. Whether it is what happened here is not something either CBC’s or TechCrunch’s reporting resolves — Ballard has not named the editor publicly, and no independent confirmation of the claim has surfaced.

One case among several this year

Shy Girl is not an isolated dispute. Within the same publishing cycle, the Society of Authors and the US Authors Guild rolled out a “Human Authored” label explicitly framed around this exact problem, a scheme that runs on the same honour-code logic that makes cases like Ballard’s hard to resolve cleanly: it can confirm who signed a declaration, not what actually went into the manuscript.

Publishing’s AI-authorship disputes this year have consistently landed in that same gap between what can be claimed and what can be checked.

What publishers are likely to change because of this

Cases like this tend to produce policy rather than resolution. Whatever did or didn’t happen with Shy Girl’s manuscript, publishers acquiring self-published titles now have a visible example of what happens when a manuscript’s origin becomes a public question after the deal is signed rather than before.

The more durable effect of the Ballard case may not be anything that happens to Ballard herself, but a shift toward publishers asking harder questions about editorial history before acquiring backlist self-published work at all, precisely because the alternative, cancelling a book already announced to readers, is expensive and public in a way that upfront diligence isn’t.

What the fallout leaves unresolved

Hachette’s decision to pull the book rather than investigate further suggests the publisher decided the reputational exposure of keeping it on shelves outweighed the cost of losing it, regardless of where the truth ultimately lands.

That is a defensible business decision, but it isn’t a finding of fact, and it leaves the actual question — did Ballard write this novel herself — unresolved in public the way accusations like this usually are: with the accused’s name attached to the story permanently, and the underlying claim never quite fully proven or disproven either way.

For readers deciding whether to trust the next disputed byline, the Ballard case offers a template rather than an answer: an accusation with real consequences, a defense that shifts responsibility without fully closing the question, and a publisher that chose distance over investigation. That template is likely to repeat, given how many hands now touch a manuscript before it reaches a shelf.

The post Hachette pulled a horror novel from release after the New York Times accused its author of using AI to write it, and she never admitted to it, she only said an editor she’d worked with might have appeared first on The Blog Herald.

The instinct people have about this reaction is that it reveals something unflattering about the earlier writer, that they didn’t know what they were doing. The instinct has the direction backward.

What the cringe is measuring

In their original 1999 study on self-assessment, psychologists Justin Kruger and David Dunning found that people who scored poorly on tests of grammar, logic, and humor consistently overestimated their own performance — the finding usually summarized as the Dunning-Kruger effect. Participants who scored in the bottom quartile on those tests estimated, on average, that they had performed in the 62nd percentile, a gap of roughly fifty percentile points between where they thought they stood and where they actually did. Less often cited is what happened in the study’s fourth experiment, where researchers trained the lowest-scoring participants in logical reasoning. Their scores went up, and so did their ability to judge their own answers accurately.

As Kruger and Dunning wrote, “the incompetent had become experts”: once trained, the formerly low-scoring group could evaluate their own work as accurately as people who had scored well from the start. The same paper found the reverse effect held throughout all four studies: participants who tested in the top quartile consistently underestimated how well they had done relative to others. Skill and self-criticism move together, not apart. The more accurately someone can do a thing, the more accurately they can see where an earlier attempt at it fell short.

Why the gap widens instead of closing

That gap between skill and self-perception isn’t static, and its direction is the important part. As competence increases, self-assessment tends to get more accurate, which means the size of the gap between someone’s current skill and their memory of a much earlier draft grows precisely because the ability to see the gap is what’s improving.

A writer five years into a body of work has a sharper instrument for detecting flaws than they did five years ago, and that instrument is what turns on old drafts that never changed at all. The manuscript didn’t get worse sitting in the folder. The reader coming back to it got more capable of seeing it clearly.

The reader a writer builds without noticing

Composition researcher Nancy Sommers found something structurally similar when she compared how student writers and experienced writers revise their own drafts. In her 1980 study of twenty student writers and twenty experienced adult writers — journalists, editors, academics — she found that experienced writers had internalized what she called an abstracted reader, a sense of audience expectations specific enough to produce, in her words, “dissonance when the writer recognizes incongruities between intention and execution.”

That internal reader gives experienced writers, as Sommers put it, “new eyes to ‘re-view’ their work,” a capacity her study found the student writers largely lacked. They could tell something was off in a way their younger selves structurally could not, because the critical apparatus that notices hadn’t been built yet when the younger draft was written.

What this looks like for anyone keeping an archive

Bloggers experience a version of this that most professions don’t, because blogging is one of the few forms of writing that leaves a public, dated, searchable archive of a person’s early work sitting next to their current work, permanently. A novelist’s unpublished first drafts usually stay in a drawer. A blogger’s first eighteen months are often still indexed, still linked, sometimes still the top search result for their own name. That visibility doesn’t create the cringe reflex, but it does make it harder to avoid, and it means the same mechanism Sommers and Kruger and Dunning documented in controlled studies plays out in public, on a timeline anyone can scroll back through.

The common response to that discomfort is to quietly unpublish or rewrite the old material, treating the cringe as a problem to fix. That misreads what the reaction is telling the writer. The old post isn’t retroactively wrong because a more capable reader now finds it lacking; it was written by someone with a less developed critical apparatus, doing the best they could with the instrument they had at the time.

Deleting it doesn’t correct the record so much as erase the evidence of the distance travelled between then and now, which is generally the more interesting part of a writer’s archive, not the part worth hiding.

The gap between taste and output

Radio producer Ira Glass gave this experience its most quoted description, in a 2009 interview later transcribed and widely circulated: “your taste is good enough that you can tell that what you’re making is kind of a disappointment to you.” Glass wasn’t describing a research finding, but he was naming the same asymmetry Kruger, Dunning, and Sommers had already documented from different angles: the capacity to judge good work and the capacity to produce it develop on different timelines, and the first one, in most working writers, gets ahead of the second and stays there. Glass’s fuller point, in that same interview, was that the way to close the gap isn’t to stop noticing it, but to keep producing work anyway until output catches up with taste — a separate, practical claim about what to do with the cringe, not just a description of why it happens.

None of this means every old draft was secretly fine and the cringe is some kind of miscalibration running the other way. Some old work really was weaker, by any reasonable standard. The point is what the wince is evidence of: a critical faculty — Sommers’ internalized reader, Kruger and Dunning’s trained metacognition — that the writer rereading the page now has, and the writer who wrote it then didn’t yet. Cringing at an old paragraph isn’t the discovery that a past self was a bad writer so much as the only available proof that a present self is a better one.

The post Writers who reread their own old work and physically cringe aren’t bad judges of their own quality, they’re the only kind of writer who has actually gotten better appeared first on The Blog Herald.

Readers who pick up a book carrying it are not being shown evidence. They are being shown a promise, made by the person whose name is on the cover, that no generative AI system wrote the words inside.

What the label actually checks

The Human Authored scheme, launched by the UK’s Society of Authors in partnership with the US Authors Guild at the London Book Fair in March 2026, works through self-declaration. An author who wants the logo signs a licensing agreement stating the text was written without generative AI, though the scheme permits assistive tools such as spellcheck, brainstorming aids, and outlining software. It is free to SoA members, limited to text-based work for now, and open to backlist titles published from 2020 onward.

The Society does not read the manuscript before granting the label. Its own FAQ is direct about this: “The SoA cannot give individual bespoke advice on whether a work meets the criteria. This is an assessment that needs to be undertaken by you individually as the author.” What the Society verifies, within 24 to 48 hours, is that the applicant is a paying member. The claim about the manuscript itself is taken on trust.

The part that gets enforced, and the part that doesn’t

The Society owns the trademark on the logo, and that ownership is where its enforcement power actually lives. If an author or publisher prints the Human Authored mark without registering for the scheme, that is a trademark violation the Society can pursue. If an author registers, signs the declaration, and the manuscript turns out to have been substantially AI-generated, there is no equivalent mechanism. The honesty of any individual label rests entirely on the honesty of the person who applied for it, and the scheme appears to have been built that way deliberately: checking manuscripts at scale for a membership body of thousands was never going to be operationally realistic, so the tradeoff was to protect the mark rather than police the content underneath it.

Why authors wanted it anyway

The scheme exists against a backdrop of numbers the Society has cited from its own membership: 86% of authors say generative AI has already reduced their earnings, and 57% say they no longer consider a writing career sustainable. Against that, a label that can’t verify content but can signal intent has still been treated by many members as worth having. Novelist Tracy Chevalier presented the scheme at its launch, and children’s author Malorie Blackman framed the motivation in terms of craft rather than compliance: “The Human Authored scheme seeks to highlight the imagination, commitment, craft and care taken to produce stories and books which can be enjoyed by everyone.”

The timing turned out to be more relevant than its organizers could have planned for. Ten days after the scheme launched, Hachette pulled Shy Girl, a horror novel, from release in the UK and cancelled its US publication after AI-authorship allegations — a case that made the honour-code nature of any authorship claim hard to treat as hypothetical. A label introduced into that environment was, among other things, an early answer to a question readers were about to start asking anyway: how would anyone actually know?

A narrower version, on the other side of the Atlantic

The US Authors Guild runs a parallel version of the scheme, and the difference between the two is instructive. The American program verifies the identity of the person applying, using an identity-verification service, but like its UK counterpart it does not verify the content of the manuscript. Two national trade bodies, working independently, arrived at the same boundary: confirm who is asking, not what they wrote. That convergence suggests the limitation isn’t an oversight specific to one organization’s resources, but a structural fact about what a membership body can actually check versus what it can only ask its members to state.

What staking your name on it actually costs

Self-declaration schemes get dismissed quickly as toothless, and in the narrow sense that no one is checking the manuscript, they are. But the label isn’t only a technical claim about a text; it’s a reputational one, made under the name of a professional body whose credibility depends on members not making a habit of lying. An author who registers a book that later turns out to be substantially AI-generated isn’t just risking a correction. They’re risking becoming the case study everyone cites the next time someone argues the scheme doesn’t work — a cost the Society has an informal interest in making expensive, through exactly the kind of reputational fallout Shy Girl generated for its own author within days of the accusation surfacing. The label doesn’t need a verification system to carry a penalty. Public attention supplies one on its own, as Ballard’s case demonstrated well before any trade body had to intervene.

That doesn’t make the honour code equivalent to verification. It makes it a bet that public scrutiny will do the enforcement a membership organization can’t afford to build. Whether that bet holds depends entirely on how often anyone actually checks, and so far, the checking has been coming from outside reporters, not from the scheme itself.

What the label is actually for

None of this makes the label meaningless, but it does mean it measures something different from what its name implies. It doesn’t measure whether AI wrote a book so much as whether an author was willing to sign a document saying it didn’t. For most authors, in most cases, that distinction won’t matter, because most authors aren’t lying about how they wrote their own books. What the label actually formalizes is something publishing has always run on without saying so out loud: the presumption that a name on a cover corresponds to a person who wrote what’s inside it.

The Human Authored mark just puts that presumption into print, on the record, in a case where some readers apparently needed it stated rather than assumed.

The post The Society of Authors just launched a label that goes on the back of a book jacket reading “Human Authored,” and it runs entirely on an honour code, which means the only thing standing between a reader and the truth is a writer’s word appeared first on The Blog Herald.

The post New York’s legislature has passed a bill requiring both AI disclosure and human review of AI-assisted news — the first time a government has tried to legally define what a human editor’s presence requires appeared first on The Blog Herald.

The pattern is consistent enough, and has repeated itself enough times, that naming it is no longer controversial. What is still underappreciated is how reliably the warnings were dismissed at each iteration — not by people who were naive, but by people who had good reasons, in the moment, to believe that this platform would be different. The early evidence always supported their optimism. The pattern only became visible in retrospect, which is the mechanism that makes it so durable.

Twitter launched in 2006 as a communication tool: a simple, open publishing layer that bloggers, journalists, and independent writers adopted with enthusiasm, partly because the platform’s early design actively supported the ecosystem that grew on top of it. Third-party clients were welcomed; the API was open; the culture of the platform was shaped by the writers and thinkers who arrived early and set its tone. By 2012, Twitter had begun restricting that ecosystem, capping third-party app users and introducing rules that made it clear the platform intended to control how its content was distributed. By 2023, it had blocked third-party clients without warning, ended free API access, and effectively dismantled the developer ecosystem that its growth had depended on. Jack Dorsey later acknowledged that cutting off the API in the early years was one of the company’s biggest mistakes.

Writers who raised platform-dependency concerns in 2012 were routinely waved off as unnecessarily paranoid.

The Medium arc

Medium launched in 2012 with a pitch that was explicitly about quality writing. Ev Williams, one of Twitter’s co-founders, described it as a place for ideas and perspectives — a reaction against the incentives of the attention economy and a genuine attempt to build a home for writing that rewarded substance over virality. The early platform attracted serious writers, and the design reflected their needs: clean, distraction-free, generous with formatting, built around the reading experience rather than the advertising surface.

The pivot sequence that followed has become a case study in platform incentive drift. In 2017, Medium abandoned its advertising model, laid off staff, and launched a subscription product, creating a Partner Program that would share revenue with writers based on reader engagement time. The Partner Program was a genuine attempt to align platform and creator incentives, and it attracted more than 200,000 enrollees and paid out approximately $28 million to writers over its life. 

Then, as the platform’s economics shifted, the revenue share changed, then changed again — from a claps-based system, to reading-time-based payouts, to a referral-share model that rewards writers for driving traffic from outside Medium. Each change redistributed who earns what, typically favoring writers already skilled at growing referral traffic over those who had built their strategy around the platform’s earlier terms. The platform that had been built on the promise that writing could be its own revenue model had, through a series of individually defensible pivots, arrived at a system where earning a living from it required constantly re-learning how the platform wanted to be played.

The Facebook Instant Articles lesson

Facebook’s relationship with publishers and writers follows a different version of the same arc. The platform built its early content ecosystem by being genuinely useful to writers and publishers: organic reach in the 2010-2013 period was high enough that a Facebook page was a meaningful distribution channel, and the platform’s social graph produced real referral traffic to the sites that content linked to. Publishers invested significantly in Facebook audiences, trained their readers to follow them there, and built editorial and distribution strategies around what Facebook sent them.

By 2014, organic reach for Facebook pages had fallen to roughly 6% of followers — down from 16% two years earlier — as the platform began reducing distribution of unpaid content to create demand for paid promotion. In 2015, Facebook launched Instant Articles, inviting publishers to host their content natively on Facebook’s infrastructure in exchange for faster load times and the suggestion that native content would receive preferential distribution. The publishers who adopted it most fully were the ones who had built the deepest dependence on Facebook traffic. When Facebook wound down the Instant Articles program in 2023 — determining it was no longer central to its strategy — those publishers found themselves with content hosted on a platform whose interest in distributing it had evaporated.

The writers and editors who had argued against building on Facebook’s infrastructure — who had insisted that a platform whose business model depended on keeping users on Facebook would never consistently prioritize sending them elsewhere — were, in the period of high organic reach, described as missing the distribution opportunity. The argument that seemed paranoid in 2012 was well-supported by 2018.

The mechanics of extraction

The pattern repeats because the underlying mechanics are consistent. Platforms need content to grow. In the early phase, attracting writers and creators is the primary growth challenge, and the incentives are structured to solve it: open APIs, favorable revenue splits, organic distribution, tools built around creator needs. The creators who arrive early bring their audiences, establish the platform’s cultural tone, and provide the content that attracts the next wave of users. The platform and the creator, in this phase, are genuinely aligned. The early evidence of good faith is real, not manufactured.

The divergence begins when the platform has reached sufficient scale that the creator is no longer the scarce resource. At that point, the creator’s audience is on the platform regardless of the creator’s preferences — the followers are already there, already accustomed to consuming content in that format, already trained to expect it. The creator who wants to reach their audience must go through the platform, whether or not the platform continues to distribute their work on favorable terms. This is the moment at which the leverage shifts, and platforms have consistently used the shift in the same direction: reducing organic reach, restricting external links, adjusting revenue shares, and introducing features that keep the audience engaged with the platform rather than directed to the creator’s off-platform presence.

The creator who objects is in a weak position. Their audience is on the platform. Leaving means leaving the audience. Building an alternative distribution channel — an email list, a direct subscription, a different platform — takes time and resources, and the audience is usually smaller and less engaged off-platform than it was on it. The rational response for most creators is to stay, adapt, and work within whatever terms the platform currently offers. Which is precisely what the platform has calculated.

Why the warnings get dismissed

The persistence of platform optimism in the face of repeating evidence is not irrational. The writers who moved to Medium in 2015 made a reasonable calculation: the platform was growing, the tools were good, the audience was engaged, and the revenue share was better than most alternatives available at the time. The abstract argument — “platforms eventually extract more than they give” — was less compelling than the concrete reality of reach and revenue in front of them. Being right about the long-term trajectory at the cost of forgoing the short-term opportunity is not obviously a better choice.

The “paranoid” label also does real work in platform ecosystems. Platforms in the growth phase depend on creator enthusiasm, and creator enthusiasm depends on the social narrative that the platform is a good partner. Writers who raise structural concerns about dependency and extraction are not just skeptics — they are challenges to the shared story that makes the platform attractive to other creators. The social pressure to dismiss them is real, and it comes from other creators as much as from the platform itself.

What changes the calculus, for the individual writer, is the accumulation of evidence across platforms rather than within any single one. A writer who watched the Facebook organic reach collapse, then watched the Medium partner program wind down, then watched the Twitter API close, is reading the same pattern across three data points. Each data point was predictable from the one before, which is why the people who identified it earliest were not prophets. They were reading the incentive structure honestly and not letting the early-phase generosity obscure what it was in service of.

What the pattern predicts about current platforms

The question the pattern raises for any platform currently in its writer-friendly early phase is not whether the extraction will happen — the mechanics suggest it will, unless the platform has found a structural reason why its interests won’t eventually diverge from its creators’. The question is when, how quickly, and how complete.

Substack is the current case most worth watching. It has maintained a revenue split more favorable to creators than most predecessors, and its architecture — email delivery, portable subscriber lists — provides more exit optionality than platforms that trap audiences in social graphs. But it is also a platform with its own distribution logic, its own content policies, and its own strategic interests, and it has already made product decisions — the follow feature, the recommendation engine — that serve platform growth in ways that don’t necessarily align with individual creator interests. Writers who’ve flagged early signs of that divergence have often been met with the same response: that they’re worrying unnecessarily.

The pattern doesn’t guarantee any particular outcome. But the writers likely to navigate the next phase of platform evolution best are the ones who understand these mechanics clearly enough to make informed decisions about where to build — treating platform generosity as a phase, not a promise.

The post The platforms that built their early growth on being tools for writers and creators eventually became systems that extracted value from that writing and returned less of it over time — and the writers who noticed this earliest were the ones who got called paranoid appeared first on The Blog Herald.

Cannes Lions’ response to the DM9 scandal rests on a specific theory of institutional failure, and it is worth naming before evaluating whether the response is likely to work.

The theory goes like this: the problem was that the wrong people were accountable. Junior teams made submissions that senior leaders had not scrutinized. If the CEO and CMO are required to put their names on every entry, they will scrutinize what they’re signing, and fraudulent submissions will be caught before they’re submitted. The rule addresses a gap in oversight. Close the gap, solve the problem.

This theory is coherent and, in limited respects, accurate. What it is not is sufficient. The DM9 submission — a campaign for the Consul brand that used AI-manipulated footage from a CNN Brasil broadcast to simulate a campaign that had not run as presented — did not happen because no one senior was paying attention. It happened because someone decided that winning was worth the risk of fabricating evidence for it. Adding a signature requirement to that decision does not make the decision less likely. It makes the decision more expensive if discovered, which is a different thing.

What the rule actually does

The new Cannes Lions integrity standards, announced after the 2025 festival, require that every entry be personally attested by the business leader of the submitting agency and a senior marketer from the brand. The festival has also introduced AI-detection tools, an Integrity Council for escalated cases, and the possibility of three-year bans for agencies found to have submitted deliberately misleading work.

The CEO and CMO sign-off is the most discussed element, and it does change the incentive structure in a specific and real way. Under the previous system, the person who decided to fabricate case study materials was unlikely to be the person who suffered the most severe consequences if the fabrication was discovered. The sign-off requirement moves liability upward, making agency leadership personally accountable for the integrity of every submission bearing their name. This is a form of accountability, and it is meaningfully different from what existed before.

What it is not is a test of whether the work actually ran. An agency CEO who is willing to authorize fabricated campaign materials — who has decided that a Grand Prix is worth the risk of deception — will sign the form confirming the materials are authentic with approximately the same ease as an account director. The signature is not a verification mechanism. It is a liability assignment. And liability assignment changes behavior most reliably in people who were not planning to commit fraud; for people who were, it primarily changes the distribution of consequences if they’re caught.

The fraud economy the rule leaves intact

The question the CEO/CMO sign-off requirement does not answer — and that the broader package of new rules only partially addresses — is why the fraud happened in the first place. The incentive is well understood: Cannes Lions awards carry substantial commercial value in the form of new business pitches won, talent recruited, and client confidence maintained. A Grand Prix is not just a trophy. It is evidence, in the market for agency services, that a firm’s creative output meets one of the highest available standards of creative excellence.

This evidence is used in competitive pitches, in fee negotiations, and in conversations with the clients whose spending sustains the agency. The award is worth money, directly and repeatably, well beyond the festival at which it is won.

Ghost campaigns — entries that represent work that was never run, or was run in a form significantly different from what the submission presents — are not a new phenomenon. Industry insiders have discussed the practice informally for years, and the DM9 case is better understood as the most visible recent instance of a persistent pattern than as an aberration. What AI did was lower the production cost of plausible fabricated evidence to near zero. Before generative tools, constructing a convincing case study for a campaign that hadn’t run required significant effort: staged photographs, constructed media schedules, invented metrics that required detailed crafting to withstand scrutiny. The effort was a friction that deterred some fraction of those who might otherwise have attempted it. AI removed that friction without removing the incentive.

The CEO/CMO sign-off requirement addresses neither of these things. It doesn’t reduce the commercial value of a fraudulently obtained Grand Prix. It doesn’t make fabricated evidence harder to construct. It adds a procedural step to the submission process and assigns personal liability to the people whose names appear on it. This is governance, and governance has a role. But governance is a response to the symptom — a submission was fraudulent — rather than to the cause — an industry decided that awards were worth fabricating.

What honest accountability would require

The harder question, which the new rules do not attempt to answer, is whether the industry is willing to examine the awards culture that made the fraud economy possible. Cannes Lions is not simply a celebration of creative work. For many agencies, particularly those outside the major English-language markets, it is a commercial lifeline — the primary mechanism through which creative quality is credentialed in a global market that has no other common standard. The pressure to win is proportionate to the commercial stakes attached to winning, and those stakes are set by the clients and procurement teams that use Cannes results as a proxy for creative quality when evaluating agencies.

An industry that genuinely wanted to address the fraud economy would look at that dynamic and ask whether it should change. Should Cannes Lions results carry the commercial weight they currently do in pitch processes? Should clients and procurement teams treat award credentials as less determinative, given that they’ve now established a public record of being falsified? Should the festival itself reduce the number of awards categories, limiting the supply of credentialing hardware and therefore reducing the pressure associated with any single entry?

These are questions about culture and incentive, and they are substantially harder than requiring a signature. They require the industry to decide that the problem is not a governance gap but a values problem — that the fraud happened not because no one senior was watching but because the rewards for winning had grown disproportionate to the integrity required to win honestly. That conclusion requires more discomfort than a form.

The paperwork vs. the practice

The agencies quoted in the trade press describing the new rules as “burdensome but necessary” are being honest about both adjectives. The rules are burdensome: gathering senior sign-offs across agencies and client organizations, meeting new documentation requirements, navigating fact-checking processes, is real work added to an already elaborate submission process. And they are necessary, in the sense that some response to a public fraud of this magnitude was inevitable and doing nothing would have been more damaging than doing something.

But “burdensome but necessary” is a framing that locates the solution in compliance — in doing the required paperwork — rather than in the practice the paperwork is meant to enforce. The agencies that submitted fabricated work in 2025 did not do so because the submission process was too light. They did so because they concluded that the risk of fraud was acceptable relative to the reward of winning. New documentation requirements do not obviously change that calculation for the minority of agencies willing to make it.

What might change it, over time, is a consistent pattern of detection and enforcement — the three-year ban prospect being taken seriously, the Integrity Council operating with genuine independence, the AI-detection tools improving in capability and coverage. If the expected cost of fraud rises substantially and demonstrably, the risk-reward calculation changes. Rules alone don’t produce that outcome. Rules enforced, consistently and publicly, might.

Cannes Lions has introduced the rules. The question the industry should be watching is whether the enforcement that gives them meaning materializes. Bureaucracy without consequence is a form of theater — a visible response to a problem that leaves the problem’s underlying structure intact. The CEO and CMO who sign every entry will, for the most part, be signing work that was real. The small minority for whom the signature is a lie will not be stopped by the form. They’ll be stopped, if they’re stopped at all, by whether the festival has actually built the capacity to find out.

The post An advertising agency won a Grand Prix at Cannes Lions last year with fabricated case study evidence, and the festival’s response was to introduce a rule requiring the CEO and CMO to personally sign every future entry, which assumes the problem was insufficient bureaucracy rather than insufficient honesty appeared first on The Blog Herald.

The Pew Research survey of 5,119 US adults conducted in February 2026 describes a different story. Forty-nine percent of respondents now use AI chatbots — up from roughly a third in 2024, a substantial increase in a short period. Forty percent say AI will make society worse. Sixty-three percent say AI is advancing too fast. Fifty-nine percent say they have little to no confidence in the companies developing AI to do so responsibly. Sixty-seven percent say they have little to no confidence in the government to regulate it.

What the survey describes is not a technology that people have adopted because they trust it. It is a technology that half the country is using, despite expecting it to harm the world they live in. That is a condition without a clear precedent in the history of mass consumer software adoption

What past adoption waves looked like

The internet reached 50% US household penetration around 2001, by which point it had been widely characterized as a transformative force for human communication, commerce, and access to information. The prevailing public sentiment was optimistic, sometimes extravagantly so. The smartphone reached 50% US adult ownership around 2012-2013, during a period of broad cultural enthusiasm for mobile computing and the connected life it enabled. Social media platforms grew through the 2010s on the explicit promise of connection — keeping in touch with people you cared about, finding communities organized around shared interests, having a voice in public conversation. The early criticism of these technologies existed and was occasionally prescient, but it was a minority position during the period of rapid adoption. Most people who adopted them did so because they expected them to make their lives better.

This is not to romanticize those adoption waves. The internet’s optimism produced the dot-com crash. The smartphone’s promise came with surveillance infrastructure that most early adopters didn’t understand they were accepting. Social media’s connection promise arrived attached to engagement algorithms that turned out to optimize for outrage rather than community. The retrospective picture of each technology is substantially more complicated than the enthusiasm that drove its adoption. But the enthusiasm was real, and it was the engine of adoption.

AI chatbots in 2026 are being adopted at the same rate without the enthusiasm. The majority of Americans using them expect society to be worse for it. This is a genuinely new configuration of the technology-adoption relationship, and the question it raises is not whether people will keep using AI — the data suggests they will — but what it means to build an industry on adoption that is driven by something other than belief.

Why people use things they don’t trust

The mechanism driving AI adoption without trust is not difficult to identify. It is competitive pressure, or more precisely, the fear of falling behind a standard that AI is setting. About a quarter of Americans now report using AI chatbots daily; 12% use them several times a day. For this group, AI has become a tool that makes specific tasks faster or more tractable — writing, research, coding, summarizing, generating options to consider. The individual who uses an AI chatbot to draft an email faster is not making a statement about AI’s societal impact. They are making a local, rational decision about productivity. The fact that they expect the technology to make society worse is a separate belief, held in a separate register.

This separation — between what I use and what I think is good — is not unique to AI. People use tobacco while knowing it is harmful to their health. People use social media platforms whose design they find manipulative. People shop at retailers whose labor practices they would not endorse. The consistency between personal behavior and social values is an aspiration, not a steady state, and the gap between the two is especially wide when the cost of consistency is high — when not using AI means being slower, less capable, or less competitive than the people around you who are using it.

What AI has done faster than most previous technologies is produce a felt competitive necessity. The adoption of social media was partly social — not being on Facebook or Instagram had social costs in terms of connection and visibility. The adoption of AI is largely productive — not using it carries efficiency costs in contexts where it is increasingly the baseline. The feeling that you have to use it is not driven by peer pressure in the social sense. It is driven by the sense that the alternative is falling behind a standard that is being reset by people who aren’t waiting for the broader social questions to be resolved.

What the distrust consists of

The 40% who say AI will make society worse are not a monolith, and understanding what the distrust consists of matters for understanding what, if anything, might change it. The Pew data gives some texture. About half of Americans who get news from AI chatbots say they sometimes encounter information they think is inaccurate. Distrust of accuracy is one component — a belief that the technology produces unreliable outputs — and it coexists with continued use because inaccurate-but-fast is still useful for many applications.

A second component is distrust of governance. Sixty-seven percent of Americans have little to no confidence in the government’s ability to regulate AI, and 59% have little to no confidence in the companies developing it to do so responsibly. These are not small numbers, and they are not primarily about AI’s technical capabilities. They are about whether the institutions that control AI’s development and deployment can be trusted to make decisions in the public interest. The answer, across a substantial majority of the public, is no — and this is the dimension of distrust that matters most for understanding the long-term social contract around the technology.

A third component is pace. Sixty-three percent say AI is advancing too fast. This is not the same as saying AI is bad. It is saying that the speed of deployment has outrun the capacity of society — its regulatory frameworks, its ethical norms, its labor markets, its educational systems — to adapt. The concern is not that AI won’t work. It is that it will work faster than the human institutions that are supposed to manage its consequences can keep up with it.

What it means to build on adoption without belief

The companies building the AI economy — OpenAI, Google, Anthropic, Microsoft, Meta — are doing so against a backdrop of public sentiment that is substantially more skeptical than the sentiment that attended the build-out of previous technology platforms. This matters for reasons that go beyond public relations. Technology industries that develop in an environment of public trust tend to benefit from regulatory tolerance, from consumer goodwill that absorbs early failures, and from a cultural narrative that frames the industry’s growth as progress. Technology industries that develop in an environment of public skepticism face the inverse conditions: tighter regulatory scrutiny, less consumer goodwill to absorb failures, and a cultural narrative that frames growth as a problem to be managed.

The current AI industry is operating in the second environment while applying the assumptions of the first. The product launches, the capability announcements, the deployment timelines, and the investment levels are all calibrated to a world in which the public is, if not enthusiastic, at least broadly receptive. The Pew data suggests the public is broadly using and broadly skeptical — a combination that is more fragile than either enthusiasm or resistance alone, because it is held in place by felt necessity rather than genuine buy-in. Felt necessity can evaporate when alternatives appear, when the competitive advantage of AI tools narrows, or when a sufficiently visible failure makes the cost-benefit calculation shift.

The 49% adoption rate is, by the industry’s own metrics, a success story. Half of American adults using the technology within a few years of its mainstream introduction is an adoption curve that most new technologies don’t achieve. A 2025 YouGov survey found the same pattern a year earlier: widespread use, persistent distrust.

What this adoption rate actually represents — belief in the technology, or resignation to it — is a question the headline number doesn’t answer. The Pew data does, and the answer is instructive. We have reached the point where people are using a thing they distrust at the rate they once used a thing they believed in. Whether those two conditions produce the same outcomes, over time, is the question the industry has not yet had to answer.

The post Half of American adults now use AI chatbots, but 40% of Americans expect the technology to make society worse — which means we have reached the point where people are using a thing they distrust at roughly the same rate they once used a thing they believed in appeared first on The Blog Herald.

I should be straight with you about the line at the top of this piece. The deep version of this insight is not my personal tally; it comes from the researchers who have actually logged the hours, sitting with long-married couples in a lab and tracking, over years, who lasts and who does not. The most famous of them is John Gottman, and one finding from that work reframes the whole question of what marriage even is. By his research, “sixty-nine percent of relationship conflict is about perpetual problems. All couples have them.” Roughly two-thirds of what you fight about, you will fight about forever.

What we get wrong about lasting love

The cultural story says a great marriage is sustained by passion, and that when problems show up, good communication will resolve them. Both halves of that are shaky. Passion, the breathless kind, is not built to run for four decades at full volume, and couples who treat its fading as a verdict tend to keep leaving in search of it. And the idea that talking it through will solve your problems runs straight into Gottman’s number: most of your conflicts are not solvable. They are rooted in permanent differences in personality and in what each of you needs from a life. You do not fix those. You carry them.

Which means the skill that actually predicts survival is not resolution. It is what Gottman’s team describes as learning to establish a dialogue about the unsolvable things — ideally one that “communicates acceptance of your partner with humor, affection, and even amusement.” Translated out of the lab, that is almost exactly the unglamorous thing long-married couples keep circling back to: you accept that your person will always be a little too slow, or too loud, or too cautious, or bad at the one thing you wish they were good at, and you decide, again and again, that this is the cost of admission and you are paying it.

The tolerance that does the work

This is what “a tolerance for mild disappointment” really means. Not numbness, and not settling. It is the daily, slightly unsexy act of letting your partner be a whole separate human who will inevitably let you down in minor ways — and choosing to stay in the room anyway, with some humor about it. The couples who last are not the ones with less to be annoyed by. They are the ones who stopped treating every annoyance as evidence that they picked wrong.

I have been married long enough to feel the early version of this, if not the forty-year one. In the first stretch I quietly believed that a good enough conversation could iron out any difference between my husband and me — that disagreement was a problem to be solved and then filed away for good. Some of those differences are simply never going to close, and the marriage got easier the moment I stopped trying to win them and started, on the good days, finding them a little funny. That is a small preview of what the long-married seem to know in their bones.

You can hear the difference in how the long-married describe their spouse’s flaws. They do not tell you about a partner who finally turned into someone easy; they describe the same stubbornness or messiness or moodiness they clocked at thirty, only now it is narrated with a shrug and a half-smile instead of a grievance. The trait did not disappear. The charge drained out of it. What looks from the outside like rare compatibility is often just two people who, somewhere along the way, stopped filing each other’s permanent quirks as fixable offenses and started treating them more like weather — irritating on a given day, but not a reason to leave the country.

The line this advice must not cross

Now the essential caveat, because “tolerate disappointment and stay anyway” is dangerous advice if it is left unqualified. There is a bright line between mild, perpetual friction and corrosion. Gottman’s own research is just as famous for naming what actually destroys couples: the Four Horsemen of “criticism, contempt, stonewalling, and defensiveness,” with contempt the single strongest predictor of divorce. That is the opposite of mild disappointment. Tolerance is for the dishwasher loaded wrong and the story told too many times. It is emphatically not for contempt, cruelty, fear, or abuse. If what you are being asked to tolerate is being demeaned or made afraid, “staying anyway” is not wisdom, and the right move is the door, not more patience.

I am not a therapist, so take this as a reader’s synthesis of other people’s research rather than counsel for your specific marriage. If your relationship is in real trouble — not the ordinary friction kind, but the kind that has you walking on eggshells — a good couples therapist, or in the case of any abuse, a domestic-violence resource, will help far more than any article can. Knowing the difference between a problem worth tolerating and a pattern worth leaving is exactly the thing worth getting help to see clearly.

What to actually take from this

So if you are trying to build something that lasts, the work is less romantic than the movies promised and more durable. Pick your perpetual problems on purpose — every partner comes with a permanent set, so choose ones you can live beside with affection. Expect to be mildly disappointed on a regular basis, and let most of it go. Guard hard against contempt, in both directions, because that is the thing that actually ends marriages. And notice that “staying anyway,” when it is healthy, is not grim endurance. It is the quiet, repeated decision that this particular flawed person is still the one you want to be mildly disappointed beside — which, after forty years, starts to look a great deal like love.

The post Gottman’s most overlooked finding: 69% of what couples fight about never gets resolved — and the ones who last have made peace with that appeared first on The Blog Herald.