Idle Time

Nestle vs. Wakefield: Comparing two versions of the Toll House chocolate chip cookie recipe

2025-12-23T06:48:45Z

For many years I’ve used Michael Chu’s telling of the chocolate chip cookie recipe printed on every Nestlé Toll House chocolate chips bag. Chu’s version is the recipe that radicalized me into using a scale to measure flour; to this day, I use 360 grams of flour for a full batch, or (more often) 180 grams for a half batch, and I pay no attention at all to how many cups that is.

But part of Chu’s telling has always stuck in my mind:

… Nestlé’s recipe only states: “2 ¹⁄₄ cups all-purpose flour”. Is this flour sifted (as all flour should be before measuring), unsifted, or settled for one year and then packed down to fit as much as possible in a cup? I tested the whole range of flour density options starting with the USDA standard 125 g per cup (sifted) up to the maximum I could push into a leveled cup, 160 g per cup.

Recipes (should) always use sifted measurements when providing volumes of flour because of repeatability. If a recipe used unsifted flour, it would be nearly impossible to replicate the exact same quantity of flour using measuring cups because it’s impossible to tell how much the flour has settled. (See Kitchen Notes: Wheat Flour for more commentary on measuring flour.)

I can testify to that—back when I followed the recipe directly from the bag, I once tried making a half-batch and it was a mess. They came out completely wrong.

The linked article says, in the section on measuring, “Whenever possible, flour should be measured using a scale.” Since I learned that, I’ve made numerous half-batches with the flour scaled by weight, and they’ve come out perfectly every time.

So I wouldn’t even say you should measure flour by sifting it—you should weigh it. But, of course, that requires knowing what the correct weight is. Back in the recipe, Chu continues:

Unfortunately, in the case of this recipe, it was clearly not written with 125 g per cup in mind. After testing a whole range of flour measurements, it seems that 160 g per cup (or a total of 360 g) of flour was the intended quantity. For those of you who do not use kitchen scales and wish to dry measure this amount – it’s a little more than 2-³⁄₄ cup sifted flour.

Chu bases his conclusion that “360 g… of flour was the intended quantity” on, as far as I can tell, what he expects a chocolate chip cookie to be. He finds 280 grams of flour—roughly the weight of the prescribed 2+¹⁄₄ cups of flour if that flour has been sifted—to produce “thin and chewy” cookies that lack the body he expects, and finds his expectation fulfilled only at the upper end of the scale, 360 grams.

To be fair, his expectation is corroborated: Nestlé also make pre-made dough, in both break-and-bake and scoopable-tub forms, and the cookies so produced are more similar to what I’ve gotten from the 360-gram version of the recipe. An interpretation of Nestlé’s recipe that doesn’t get at least similar results to Nestlé’s pre-made dough rightly should be questioned.

At the same time… Nestlé employ professional bakers and recipe writers, don’t they? They should know full well that flour should be sifted when measuring by volume. Surely if they intended the recipe to use 360 grams of flour, but wanted to specify that in cups, they would have given a number of cups that equates to 360 grams of sifted flour?

For a long time I let that matter rest. Then I remembered something:

Nestlé didn’t create this recipe. They’re quite open about the fact that Ruth Wakefield, the proprietor of the real Toll House inn from whom they bought the recipe and the trademarks, created the original recipe back in the early 20th century.

So maybe there’s an earlier version of the recipe? What did Ruth Wakefield herself have to say about it?

It turns out there is. Nestlé themselves published it in a handout of some sort published in 1939, which is available today from Project Gutenberg, and Wakefield published it in her own cookbook titled “Toll House Tried and True Recipes”.

The Nestlé pamphlet version, from Project Gutenberg.

They sure did write recipes differently 90 years ago. Today, you always see ingredients and steps listed separately, but here they’re mixed together.

I dug this up a year ago, but only recently got around to actually making this version of the recipe (minus the nuts—I’ve never included them and I’m not starting now). I ended up largely following the same path as Max Miller, who covered it on Tasting History about a year before my own investigation, about two years ago now.

Here are the differences I observed between Wakefield’s original recipe and the more recent versions:

Flour: Sifted

We’ll start with the question I started out with, which Wakefield answers conclusively, in both the Nestlé pamphlet version as well as in her own cookbook: 2+¹⁄₄ cups flour, sifted.

Can’t get any clearer than that. Contrary to Chu’s inference, Wakefield did in fact intend for bakers to use sifted flour, which works out to 280 grams in Chu’s conversion.

The “thin and chewy” aspect that Chu found in his full-size cookies is resolved by another difference:

Drop size/recipe yield

The last step is “drop by half teaspoons”. Half-teaspoons?!?!

Wakefield’s recipe also says it “Makes 100 cookies”. That’s a lot!

The Nestlé version we’re all familiar with says to drop the dough by “rounded tablespoon”, and that their recipe “Makes about 5 dozen cookies”.

I normally use a 22.5-ml (1.5-tbsp) disher. I pack the dough into the bowl with a table knife and slice off the excess so every single cookie is pretty much exactly 22.5 ml worth of dough. With this, I get 24 cookies per half-batch (consistently, due to consistent measurements and my pack-and-slice dropping method). If I do a full batch, that’s 48 cookies, or 4 dozen.

So let’s work backwards from that and check the other recipes’ math:

22.5 ml × 48 = 1,080 ml of dough per full batch

Yeah, I can believe that the final volume of dough is a little over a liter. There is a little more dough than that left stuck to the mixer paddle, the mixer bowl, and the scoop and the knife. But as far as what goes into actual cookies, that’s close enough for some further napkin math.

Wakefield’s recipe says it makes 100 cookies. That works out to 10.8 ml (or so) of dough per cookie. Half a teaspoon is 2.5 ml, so this doesn’t add up—even a fully-rounded half-teaspoon would at most be 5 ml, and this is twice that. Inversely, if we divide 1,080 ml by level 2.5-ml scoops, that’d be more than 400 cookies! I don’t have any ideas on how to figure out whether her drop size or her yield estimate (or both) were wrong.

Nestlé’s version says it makes “about 5 dozen cookies”. 1,080 / 60 = 18 ml of dough per cookie. This is 4 teaspoons or 1+¹⁄₃ tablespoons—maybe “heaping tablespoon” is more accurate than their wording of “rounded tablespoon”.

I will say that half-teaspoons did produce some very cute mini-cookies, but it takes forever. I gave up on that quickly and switched to my 15-ml (1-tbsp) disher, two-thirds the size of what I normally use but six times the amount prescribed by Wakefield. I didn’t bother counting the cookies after that since they were different sizes.

L: Mini-cookies produced from 2.5 ml of dough each.
R: Somewhat more normal-sized cookies from 15 ml of dough each.

Applying the same math as above, if I started a new Wakefield batch and used my 15-ml disher for all of it, I should get 72 cookies.

If you try a drop size in that smaller range, whether it’s Wakefield’s 2.5 ml or the 15 ml I ended up using or anything in between, you’ll need to be aware that smaller cookies bake faster. That brings me to…

Baking time

This is not a difference in the recipe-as-written, but in the recipe-as-executed.

Nestlé’s bag says “9 to 11 minutes” at 375°F (190°C). Chu says basically the same: “ten minutes (give or take a minute depending on your oven)”. I always do 12. Wakefield says “10 to 12 minutes”.

Folks, I tried baking half-teaspoons of cookie dough for 10 minutes and they were this close to becoming cinders. They were still technically edible, but not very good and I would never serve them to another person.

Miller had the same finding. He went down to 8 minutes. I settled on 6 for the half-teaspoons, though I went back up to 8 when I switched to my 15-ml disher.

Whether I did 2.5 ml for 6 minutes or 15 ml for 8 minutes, the cookies came out thin and crunchy, not thin and chewy.

Baking sheet: Greased

The Nestlé recipe says to drop the tablespoonfuls “onto ungreased baking sheets”.

Chu doesn’t mention this instruction and instead says “I prefer to bake the cookies on either a silicone baking mat or on parchment paper”. I settled on a silicone mat, which slides around less than parchment paper does.

Wakefield says to drop the half-teaspoonfuls “on a greased cookie sheet”. Interesting!

I tried this—using the leftover butter on the inside of the wrapper, and forgoing the silicone mat—and I wasn’t impressed. They did release with some encouragement, but that ran into the crunchier cookies being easier to break. Finding no other noticeable difference, I went back to my silicone mats pretty quickly for the remainder of the batch.

Baking soda: Dissolve in hot water

That’s a new one on me. Wouldn’t this use up some of the fizz of the baking soda?

I did try it. If there was any significant difference one way or another in rise, it was swallowed by other factors. I don’t intend to try this step again.

Chocolate chips: Method

Back in the 1930s, Nestlé hadn’t yet introduced pre-made chocolate “morsels”. You had to buy a whole chocolate bar or two and chop it up yourself, into “pieces the size of a pea” as prescribed by Wakefield. The “morsels” came later, just after WW2 by Miller’s telling.

I do think Miller got this step wrong when he tried the recipe. Neither he nor I were interested in chopping up chocolate bars, so we both used pre-made chocolate chips, but he used full-size chocolate chips, then complained that this was too big to fit in his half-teaspoon.

I used mini chocolate chips and those fit my half-teaspoon fine. So that’s what I’d recommend if you try this recipe, especially if you try the half-teaspoon as written.

Chocolate chips: Quantity

The other difference on the chocolate chip front is the amount.

Every chocolate chip bag in the grocery store, unless it’s either a cheap dollar-store brand or a larger bulk package, contains 340 grams (12 ounces) of chocolate chips. The recipe on the bag generally calls for you to use one entire bag’s worth.

(The actual net weight in the bag can vary. I weighed out the contents of a Ghirardelli bag once and that came in a little under 340. Haven’t bought Ghirardelli chocolate chips ever since. Trader Joe’s consistently weigh in at 343 grams.)

The two “economy size” chocolate bars that Wakefield called for back in the 1930s were “7 ounces” (200 grams) each. She calls for two bars, so that’s 400 grams of chocolate—not 340.

This tracks with a problem I’ve been having: After using up the dough that contains the normal concentration of chocolate chips, I tend to have some dough left over at the end, enough for a few extra cookies, with few to no chocolate chips left. (Those cookies are basically butter cookies.)

Wakefield’s recipe offers an answer: The rest of the ingredients remain the same amounts, but those amounts were calibrated for 400 grams of chocolate chips, not 340. One bagful literally isn’t enough!

Last night, I made a half-batch of my normal chocolate chip cookies (based on Chu’s version of the recipe) but used 200 grams of chocolate chips instead of 170 grams. I still had a couple of butter cookies at the end, but fewer than previously. I feel like I could put even more chocolate chips in, honestly.

(A couple decades ago, I actually made a full batch with two whole bags of chocolate chips—so 680 grams—and I dimly remember that working fine, but that was before I was more precise about everything else and I haven’t attempted that again since.)

Eggs: Beaten whole

No other version that I’ve seen specifies to beat the egg(s) before adding to the mixer bowl, instead implying that one should basically crack the egg and dump the contents straight in, then let the mixer effectively beat the egg while incorporating into the dough.

In the half-batch that I did last night, I added this step. I think this made the dough less sticky somehow—the dough released from both the mixer paddle and my disher much more easily than I’m used to.

I’ll definitely be trying this again and paying attention to this aspect the next few times I make cookies.

What worked, what didn’t

First the parts that I’m not adopting:

dissolving the baking soda in hot water: This seems pointless.
half-teaspoons: Waaaay too much work. Maybe try this if you have two or three pairs of helping hands to share the labor. I don’t so I’ll just keep using my 22.5-ml disher.
less flour than Chu thought: Nah. I like the fuller-bodied cookies from Chu’s interpretation. I’ll keep using 180 grams per half batch or 360 grams per full batch.

What I am adopting:

more chocolate chips: At least 200 grams for a half batch or 400 grams for a full batch. I may try even more.
beating the egg: If I’m right, this is helping the dough release easier, which is a tremendous win.

It’s not you—Betty Crocker’s directions are wrong

2025-12-09T05:06:13Z

Sometimes I bake from scratch, and sometimes I work from a box mix. Either way, having an accurate recipe to follow is vital for consistently correct results, whether it’s from a cookbook or from the back of the box.

I’ve found errors in the package directions for a couple of Betty Crocker products: their “Batchables” cookie mix, and their boxed chocolate chip muffin mix.

If you’ve ever tried the different variations on a Betty Crocker package and gotten wildly different results than you expected, it might not be that you screwed something up—you were following bad directions.

Batchables: Supposedly scalable, but good luck doing that by the directions

Betty Crocker sells a line of baking mixes called “Batchables”, so named because they come in resealable plastic tubs that you can make small batches from, rather than the usual pouch or box that you’re expected to use all at once.

I’m looking at the chocolate-chip cookie mix. The ingredient measurements in the directions are as follows:

Batch size	Mix	Butter	Water
4	²⁄₃ cup	2 tbsp	1 tbsp
6	³⁄₃ cup	3 tbsp	1 tbsp
8	⁴⁄₃ cup	¹⁄₄ cup	2 tbsp
12	1+³⁄₄ cup	¹⁄₃ cup	2 tbsp
24	3 cup	¹⁄₂ cup	4 tbsp

(They say “¹⁄₄ cup” of water rather than 4 tbsp, but I didn’t want to have one cell in that column with a different unit.)

Right away you can see that the ratios are inconsistent. Look at the water: It takes the same amount of water to make 12 cookies at it does 8? That doesn’t make sense.

Long section in which I explain the problems and work toward a better recipe

Black Friday deals I recommend

2025-11-28T17:26:54Z

I know Black Friday—which these days lasts about a week, if not all of November—gets pushback for the crass commercialism of it, and I don’t disagree.

At the same time, I’m keenly aware that some folks need to save money wherever they can, and also that people who are in that position deserve nice things when they can get them.

Also, the pushback should be directed at big corporations that drive down costs by outsourcing their labor to workers in foreign countries whom they can pay pennies on the dollar, and then have to lower their prices and offer sales in order to sell anything to now-impoverished people at home. Small businesses don’t have the same culpability; they’re just trying to scrape by in a market where everybody’s belts are tightened and the big boys hold the marketing megaphones.

So, here’s my little attempt to help: A collection of Black Friday sales that I think you should check out, because I think the product is worthwhile. These are all by small businesses; some are only one or a few people, while others might be a bit larger but still aren’t humongous mega-corporations.

Nothing in this post is a paid advert. With only one exception, these are products I’ve bought before with my own money and like enough to recommend. The exception is my own cookbook, so I hope you’ll indulge me on that.

Unless otherwise noted, all of these deals are available all day today, if not longer. There are a couple that end before then!

Software for macOS

Acorn

What: Image editor (competitor to Affinity, Photoshop, etc.)
How much: $20, which is half-off
How: No effort needed; price shown is with discount
When: For one week (so, through Thursday)
Kaleidoscope

What: Diffing tool (like FileMerge but better)
How much: $57.60, which is 40% off, each year for two years (it is a yearly subscription), and you get an extra two months free
How: Coupon code KS23BF25
When: Until Wednesday, December 3

I admit that I don’t use the current version; I use version 3, which was the last version with a perpetual license. But they don’t sell that version anymore, so if you want to start using Kaleidoscope for your diffing, and you can afford the yearly cost, this is the best deal going.

Reading material

I Need Food

What: My cookbook!
How much: $2.50, which is half-off (you can also pay more if you like)
How: No effort needed; price shown is with discount
When: Through Wednesday, December 3

This is a cookbook for people who aren’t cooks. If you’re low on energy, or don’t want to stock ten million different ingredients, or just want easier-to-follow instructions, my cookbook is for you.
Iron Circus comics

What: Graphic novels and anthologies
How much: Half-off sitewide
How: Add to cart; discount shows in cart automatically
When: Until 9 PM PT today

Iron Circus publishes a wide variety of stories by independent, mostly queer authors. One recent title that I recommend is “Motherlover”, about two moms who meet when one of them moves into the neighborhood, and how their lives change as a result. You can also read it for free online, but reading webcomics from PDFs is generally a nicer experience than waiting for each and every page to load.

Note: Some titles Iron Circus publishes are meant for adults. These are clearly marked.
Wizard Zines

What: Educational zines about various (mostly computer) topics, the newest one being “The Secret Rules of the Terminal”
How much: Half-off sitewide
How: Add to cart; discount shows in cart automatically
When: Until 9 PM PT today

Julia Evans combines solid explanations that promote understanding with a cheerful, encouraging style. No matter the topic, you’ll come away feeling like you can do it.

Respiratory health/disease prevention

Flo Mask

What: High-quality reusable mask to block smoke, viruses, and more
How much: 25% off sitewide
How: Coupon code FLO25OFF
When: Through December 15

I know most people have at this point resigned themselves to getting covid over and over, but for those of us who still insist on staying healthy, a good mask is vital. The Flo Mask is my preferred mask as a glasses-wearer, and produces less plastic waste (because you only have to throw away the filter) compared to N95s.
SIP

What: Drinking valve for N95 masks
How much: Today, buy two, get one free; from today until 9 PM PT on December 1, buy four, get two free
How: Add three/six items to cart; discount shows in cart automatically

One of the challenges of wearing a mask for extended periods is staying hydrated. Doffing the mask to take a drink is a little bit of a hassle and wears on the elastics. This thing enables you to basically just drink through your mask, while still maintaining an airtight seal whenever you’re not drinking.

I have tried it with my Flo Mask and found that the two don’t combine very well; I’d mounted it in the middle of the filter, where the largest gap in the frame is, and that proved to be too high up. At some point I might try installing it lower. But if you use N95s or other disposable masks that don’t have a plastic frame behind them, this may work a lot better for you.
Aranet4 Home

What: Carbon dioxide monitor
How much: $143, 46% off
How: No effort needed; price shown is with discount
When: Ends Tuesday, if I’m doing the math correctly

Carbon dioxide is an indoor pollutant that tends to accumulate in enclosed spaces such as unventilated rooms, cars, buses, etc. The more people in a space, the more people building up CO² there. If you’ve ever gotten sleepy on a long drive or a shuttle bus ride, CO² is probably why. I did a long drive with an Aranet4 some years ago and it quickly identified I was accumulating CO² inside my car; I turned my car’s fan up higher and that fixed it. I’ve kept it set that way, and I’ve never gotten sleepy on a long drive ever since.

This is in the covid-prevention section because carbon dioxide accumulation in an enclosed space can also be a rough proxy for germ transmission risk, since if CO² is piling up, germs might be, too. I say “can be” and “might be” because filtration can remove germs from the air but not carbon dioxide, so if there are visible air purifiers or if filters have been installed in the HVAC system (and maintained), the risk might not be as elevated as the CO² levels.

Still, if CO² levels are elevated, you might want to (weather permitting) open a window—for multiple reasons.

Other

Wicked Cushions

What: Replacement headphone earpads
How much: 25% off sitewide
How: Coupon code BFCM2025
When: Through Monday, I think

I’ve written about these before. If the earpads on your headphones are falling apart but the rest of the headphones are still perfectly good, you might be able to replace the earpads alone.
LD Products

What: Toner cartridges (and ink, but I haven’t tried their ink)
How much: 20% off their own-brand cartridges; 5% off brand-name cartridges
How: Coupon code BLKFDY20
When: Until 23:59 today PT

Black toner is one of those things that’s basically a commodity; there’s not a lot of difference between brand-name or off-brand toner cartridges other than cost. I did find the LD Products toner cartridge for my laser printer produced somewhat darker prints; maybe their toner particles are slightly bigger than the OEM toner. But it’s within my tolerance.

They also sell ink, but my inkjet printer is one of those that takes its refills from bottles rather than having ink cartridges, so the name-brand ink for mine isn’t obscenely expensive and I’m not particularly motivated to try an off-brand. I’m even more disinclined to try their color ink, because the one or two times I tried off-brand color inks (in my old Color StyleWriter 2400), the color balance never came out right. That was many years ago with a different brand, but regardless, I haven’t tried LD Products’ ink and I’m not planning to.

Viewing old hintbooks without a red gel viewer

2025-11-25T07:35:37Z

I’ve recently been re-playing “Indiana Jones and the Last Crusade”, and it’s been long enough since the last time that I’ve needed to re-consult the hintbook a couple times.

Lemme back up.

Adventure games had hintbooks?

First, by “adventure game”, I specifically mean point-and-click graphical adventure games like “Last Crusade”.

Hi, I’m Indiana Jones. Welcome to my game.

Early adventure games had a raft of verbs for specifying what you wanted your character to do—early Lucasfilm/LucasArts games like “Last Crusade” had over a dozen, and early Sierra games had command-line input with too many verbs to list. As the form matured, games dispensed with the specific verbs in favor of contextual “use” and “examine” actions on the left and right mouse button, respectively, but the essence remains the same: exploration and puzzle-solving to progress a narrative.

And some of those puzzles could be tricky, so one thing that publishers like LucasArts either included in the box or sold as an add-on was hintbooks.

Photo “borrowed” in true adventure-game-protagonist style from this eBay listing. All this can be yours for only $385!

Typically, a hintbook included a series of questions you might ask, like “how do I get ye flask”, with one or more hints after each question, each one more spoilery than the last, and each one veiled in some way so you could get a little closer to figuring out the solution without having it just dropped on you right away.

How hintbooks worked

The exact mechanism varied. Infocom was famous for their “InvisiClues” which involved disappearing ink that reappeared with a special marker pen. But Lucasfilm/LucasArts—and one or two other publishers, including Sierra—used a different system involving a red gel viewer.

“Gel” in this case is the photographic term: a sheet of translucent plastic that filters out all colors except one. A red gel passes red light but blocks all other colors. If you’ve ever seen a pair of red-and-blue 3D glasses, those are red and blue gels—and, in fact, you could use the red side of such glasses if you lost the viewer that came with the hintbook.

The text in the books was printed in color. Question headings and the actual hints were printed in cyan, but the veil over the hints was made of red text, or a noise pattern in red and orange.

When you applied the red gel, the red veil and the white paper appeared to be the same color, while the cyan text—reflecting nearly no light that could pass through the red gel—appeared dark, or at least darker. (The contrast wasn’t great.)

This worked well enough… as long as you had a red gel viewer to enable you to read the text. If you lost it, and didn’t have the viewer from a different game or a pair of 3D glasses, you were kind of stuck.

Viewing preserved hintbooks

Many if not all of these hintbooks have been preserved online, thanks to tenacious collectors, but reading them can be a challenge. Some of have had the hints unveiled for you, which is for the best for the invisible-ink ones (and I’m sure was no small feat).

For red-gel hintbooks, it actually is possible to replicate the red gel viewer in software, given sufficiently high-quality scans of the pages.

I use Affinity Photo (a pre-Canva version). You could use Photoshop or maybe even Acorn.

At minimum, you need a Color Matrix* filter (called something else in some apps—Affinity Photo calls it Channel Mixer) and something that can convert the fairly murky result of that into a more-readable black-and-white, such as a Threshold filter. Ideally, you could mask these filters to a particular shape; Affinity has this ability (you can put the filters in a group, with the scanned page outside of and below the group, and apply a rectangular mask to the group).

For the “Last Crusade” hintbook, I had success with red=red × 100%, green=red × 100% + green × 100%, blue=red × 100%, and a threshold of 92%. No need to mess with the alpha.

With a document so constructed, you can drop in a scanned page, layer it underneath the group, and then drag the mask shape around to use as your “viewer”.

Now you can read those hintbooks just like in the 1990s, except you’ll never have to worry about losing the hintbook viewer gizmo again.

I’m here to chew bubblegum and punch Nazis. And I’m all out of bubblegum.

The tricky bit: Getting usable scans

I was going to include a list of links to hintbooks you can try this with, but apparently I got lucky with the Indiana Jones and the Last Crusade hintbook being a usable scan. The others I looked at all have the colors messed with (probably on some blanket preset that makes sense for B&W documents) in such a way that the trick no longer works.

Please, if you scan in a hintbook that requires a red gel viewer, publish the scans without color-correction.

Apparently, later LucasArts games—including the other Indy adventure, “Fate of Atlantis”—ditched the red gel trick entirely in their hintbooks, and those are of course perfectly readable today.

Bonus link

While looking for a photo of the sort of viewer I’m talking about, I came across this tutorial for making your own red gel viewer. I haven’t tried it, but if you really want to do things the old-school way, and don’t want to try your luck finding a vintage one on eBay, that might be an option.

* I’ve written about Color Matrix before. ↶

Fixing my foaming soap dispenser

2025-10-13T04:36:26Z

Mostly I wash my dishes in my dishwasher, but when I hand-wash the things that need hand-washing, I use a foaming soap dispenser.

The contents of a foaming soap dispenser are watered down relative to “regular” soap. When the manufacturer does it, this is a ripoff (you’re paying them for water replacing 80% of the soap), but when you’re the one doing the watering-down, it becomes a tremendous cost advantage. I save so much money on hand-wash dish detergent, because I barely use any of it; I buy a new bottle every few years. I don’t even buy the big jugs like I used to.

To make your own foaming soap refill with dish soap, add four parts water to one part dish soap (e.g., 240 ml water + 60 ml soap), pour that into an empty foaming soap dispenser, and mix. (Some soaps might require a different ratio.) I use this soap in this dispenser that screws onto a mason jar.

If I were buying the jar today, I’d get a plain-looking one from Dollar Tree or Daiso.

Recently, my dispenser stopped dispensing. It wasn’t stuck, which is a common failure mode for this type of soap dispenser (the pump gets stuck down and doesn’t pop back up); in this case, I could still pump it like normal, but all it did was squirt air into the jar rather than squirt foamy soap onto my hand/sponge/thing-that-needed-cleaning. Or if I did get any soap foam, it was an unusably-small amount, probably back-fed from the foam that accumulated in the jar as the pump squirted more and more air through the jarful of diluted soap.

I fixed it by filling a bowl with warm water, dipping the whole upper half (but most especially the hole where the foam is supposed to come out) upside-down in the water, and pointing the dip tube away from me, then pumping away until the dip tube started shooting squirts of plain water.

Basically, if your pump gets into a state where it only pumps air backward into the jar, lean into that and pump water backward through it instead. That should rinse out whatever dried soap (I guess) is fouling the mechanism.

Once I did that, I was able to put the pump back on the jar, then do a few more pumps to re-prime it with soapy water, and now it gives the same excellent foam it used to.

Let the user help solve their own problem

2025-02-04T22:14:15Z

I wish we had a maps app like Apple Maps or Google Maps that let you order up a travel itinerary using public transit between two points, and explicitly pick the transit routes involved. Or, ideally, multiple sets of routes, for comparison.

Like, let’s say I’m in the Mission and I wanna get to the Metreon. One of the existing apps might suggest the 14R rapid bus, which arrives in X minutes and takes Y minutes, or BART, which arrives in Z minutes and takes W minutes. But it might leave out alternatives like taking the 49 bus to Van Ness Station and then taking the Muni subway from there.

Sometimes all the app’s recommendations are reasonable, but sometimes there’s one or more options that might be preferable—and I don’t know how preferable if the app isn’t showing me when the next 49 arrives, so I can compare to the 7 minutes for a 14R or 9 minutes (including a short walk) for BART.

What I want is the ability to add a specific set of routes to include in consideration, or even to force an itinerary using those routes. Let me say “49, KLMN” and have it include that series of routes among my options for comparison.

This is one instance of a general problem, which is products having only algorithmic solutions to the user’s needs, with no opportunity for the user to contribute to the solution.

The algorithmic-only model admits only one remedy: Improve the algorithm. But because no algorithm will ever be perfect, you’ll be playing this game of whac-a-mole forever.

This goes for the developer of each product, as well as for the user. I could try the same query in Apple Maps and Google Maps and the Transit app and whatever else, but as long as they all work this way, all I’m doing is holding up different slices of Swiss cheese next to each other and comparing their holes.

When the user can contribute to the solution, then there’s a chance that they’ll have a better idea of how to meet their own needs.

And these aren’t mutually exclusive. You could use the user’s input to improve the algorithm’s suggestions.

Email spam filters have had this for decades. The “Report Spam” and “Not Spam” buttons help train the filter. And we still have them because the filter will never be perfect (not just because spam is always evolving).

For the transit routing example, it’s a more complex problem (not a simple ham-or-spam dichotomy) and there are privacy considerations. Even so, helping improve the routing algorithm for everyone could be something people could opt into.

If they decline, maybe the algorithm could use training supplements kept locally. The user who provided a suggestion could still benefit, even if they’ve declined to share that (potentially personal/identifying) data with others.

And even just the ability to add a route combination to the list, even if I have to do it every time, would be an improvement over not having that and being limited to whatever options the algorithm picks for me.

Don’t assume your algorithm has to solve everything all on its own. Let the user help. They’ll be happier with a solution they helped create, partly because it may be better for their specific needs, and partly because they got to be involved.

Stuff I made this year

2024-12-30T05:30:46Z

It’s been a hell of a year, but it’s almost over.

Since social media has gotten a lot more fragmented since Twitter died, probably a lot of folks haven’t seen the stuff I’ve only posted on my Mastodon or on my Cohost page (RIP).

So here’s a round-up of most of what I made this year. I’ll likely import select posts from my Cohost archive (download yours today!—seriously, the site’s getting deleted in a couple days!) at some future time, but for now, they’ll stay on Cohost (and the Internet Archive’s mirror thereof).

If you’re wondering why March is so heavy in this list, it’s because of #MARCHintosh. Which is coming up again soon!

March: I upgraded my iBook G4 to have an SSD, and took some notes on the process.
March: Improvements to impluse, my command-line tool for accessing the contents of vintage HFS volumes. For some highlights, see this toot.
March: A YouTube version of the Guided Tour of Macintosh Plus, demonstrating the audio cassette and software program that were meant to be used together to learn how to use a GUI and a file-system.
March: Tutorial on extracting icon resources to modern image formats using emulated Classic Mac OS
March: I discovered that you can move windows on a Mac using any edge.
March–April: Command-line tools for working with CSV files
July: A chocolate milkshake recipe equivalent to, if not better than, what’s sold in Ghirardelli chocolate shops (that Cohost link will break soon; the recipe is also in my cookbook, listed below)
September: Overhauled my webpage on political postcarding. It’s a bit of a seasonal item, only relevant in the months leading up to an election, but in those months, feel free to use this in your own postcarding efforts or send it to friends who are interested in easy ways to get out the vote.
September: Made an educational shitpost about trigonometry.
October: Another election-season item, a button to advertise that you voted early. Reusable, unlike a sticker!
October: Instructions on sewing a handle holder for a cast-iron skillet
November: As a sodium-conscious home cook, I looked up the sodium content of various brands of a couple of things I eat, and published those notes: Sodium content in frozen chicken tenderloins and Sodium content in lemon-pepper seasonings
November: I wrote my first cookbook!
December: I wrote a command-line tool to extract resources from old Mac files (including the System file, applications, many types of plug-ins or data files, and more).
December: I wrote a utility to sort the output of du -h by file size. Then somebody pointed out that sort has a -h option that does the same thing. A thousand times had I wished for it; 999 times had I looked and not found it. Now it exists—hooray! I thanked them in my reply, and have since archived this repo and will simply use sort -h henceforth.
December: Earlier in the year, I came up with an effective solution for storing my two kitchen knives (big knife and little knife) without damaging their edge. (This was motivated by my learning to sharpen them using whetstones.) This month, I improved that solution.

Sodium content in lemon-pepper seasonings

2024-11-14T23:03:54Z

While I was writing the previous post about sodium in frozen chicken, it occurred to me that I should also include info on sodium in the lemon-pepper seasoning I often add to it, and eventually I decided to split that off into its own post.

I have high blood pressure. Have had for years. I take medication, but I’m also trying to reduce the sodium in my diet to limit how much it contributes.

It turns out that lemon-pepper seasonings, which I use frequently, are a significant contributor of sodium, as they all include salt (presumably sodium chloride) as one of their top two ingredients. Sigh.

There are salt-free options and I’m going to have to try some of them at some point. In the meantime, here are the sodium contents of all of the lemon-pepper seasoning products listed on Safeway’s website (minus the explicitly salt-free ones), plus the one currently in my kitchen, along with where “salt” is listed in the ingredients list.

These are all based on 1/4 teaspoon of seasoning, which ranges from 0.7 to 1 gram depending on brand. Most of these have either salt or pepper as the first ingredient; I’ve noted the exceptions.

Brand	Sodium content	Salt is listed…
Lawry’s The one I was using before my current jar	90 mg	2nd
Pacific Organic The one I’m currently using	140 mg	1st
Sunny Select (Save Mart)	140 mg	1st
Signature Select (Safeway)	95 mg	2nd
Kinder’s	105 mg	2nd (sugar is 1st, pepper 3rd)
McCormick	180 mg	1st
Scott’s	210 mg	1st (“spices” is 4th)

Sodium content in frozen chicken tenderloins

2024-11-12T14:03:49Z

[Content note: This post is about meat used in cooking. If you don’t eat meat, this will be at best academic. If you’re opposed to the eating of meat, feel free to skip this post entirely.]

I have high blood pressure. Have had for years. I take medication, but I’m also trying to reduce the sodium in my diet to limit how much it contributes.

I also do a lot of home cooking, and many of my dinners involve a frozen chicken breast tenderloin. For one person, namely me, this is the perfect serving size of meat protein. Generally I grill it on my George Foreman grill with a dusting of lemon-pepper seasoning.

Frozen chicken typically has some sort of brine solution added to it. I assume this helps the freezing process, or something. One consequence of this is that the sodium content of frozen chicken is greater than refrigerated chicken (which might be a reason for me to switch to refrigerated… hmmm) and also varies widely between brands.

I’m only going to be looking at tenderloins, since that’s what I buy. I’d guess that the ratios would be similar for a different cut—but if you buy a different cut and you care about sodium content, you should probably research the sodium content of your options for that cut yourself.

For most of these, I’ll link to the product online. One of them I don’t have an online product link for; the other, I’ll elaborate on after the table.

All of these are based on the nutrition facts label, which all of them give in terms of one 112-gram (4-ounce) tenderloin.

Brand	Sodium content
Signature Farms (Safeway)	190 mg
Trader Joe’s	75 mg
Foster Farms	300 mg Note: Their website says 280 mg but I’m going by a bag I own.
Kirkland Signature	200 mg
Good & Gather (Target)	280 mg
Tyson	190 mg
Kroger	180 mg
Great Value (Walmart)	190 mg
Perdue	260 mg

The least was Trader Joe’s at 75 mg, and the greatest was Foster Farms at 300 mg (with Target just behind it at 280). Quite a range!

One thing I noticed is that nutrition facts published online may disagree with what’s printed on the bag. One example is noted above; another, bigger discrepancy is Safeway’s listing for store-brand frozen chicken tenderloins, which says that those contain 1,090 milligrams of sodium—five times the sodium content listed on the (differently-styled) Safeway-brand bag in my freezer now. That seems like it might be the wrong product’s nutrition facts.

Skillet handle holder

2024-10-27T20:12:03Z

I cook with a couple of cast iron skillets—a 6-inch one and a 10-inch one—and, after each cooking session is complete, I wash the skillet so the fats and food bits left behind don’t set. In order to do this, I need to hold onto the handle—but by that point, the handle is very hot. Even with cast iron’s notoriously slow internal distribution of heat, the handle will still be 200°F or more.

My kitchen towel is often wet from hand-washing by that point, so I can’t very well use that as an insulator—the water will conduct the heat very efficiently. And I’m not stocking the kitchen with half a dozen towels at a time.

Pot-holders work, but it’s easy for a pot-holder to get misaligned and me to end up touching hot metal anyway. What I need is something I can slip onto the handle for that final cleaning step.

I’ve been using the Ove Glove, but wanted something easier to wash.*

Lodge sells silicone pot handle covers, but I’d rather not spend $10 a pop for more plastic.

What I realized is that I can modify a pot-holder by folding it over and sewing it closed. So I bought a two-pack of black all-cotton pot-holders from Dollar Tree and got started.

The first step was to seam-rip all of the original bias tape away from the edge, exposing the edges of the pot-holder body. This binding also includes the loop that can be used to hang the pot-holder, and secures the label with the care instructions. I want to keep both of these in the new design.

I cut the original bias tape (from the end opposite the loop) to a length that I could put back on one edge, including both adjacent corners. This edge becomes the perimeter of the opening.

Then I made some fresh bias tape from black cotton quilting fabric. I measured the length around one and a half sides of the pot-holder, which is the length of binding that will run from the open end down one side and along the closed end. The width is three inches, producing 1½-inch (36 mm) bias tape.

I don’t have a bias tape maker in that width (my widest one is 24 mm), so I had to apply the folds manually using pins and my ironing board. All-metal tailoring pins are helpful here because you can iron directly over them.

I cut off the diagonal ends to leave a right-angle end before sewing.

The reason for the jumbo bias tape is that twice the thickness of the pot-holder is _thick_. I used a 100-diameter denim needle and reduced my presser foot pressure by two full turns, and my Singer Heavy Duty machine still struggled at times. And you can really tell where the cotton batting is being compressed by the lockstitch.

I decided to put the care label at the closed end. One thing I’d do differently: I think I’d prefer the label on the underside. The side I put the label on ends up facing up and tickling my palm when I’m holding the skillet with my left hand.

The loop also goes at the closed end, at the very end of the new bias tape. I cut it off from the original bias tape and stuck one end in under the new bias tape on each side.

The finished product is very much a universal skillet handle holder, longer than either of my skillets’ handles; theoretically I could have trimmed it down to fit one of my skillets more exactly, and done the same with the other pot-holder to make one tailored for the other skillet. But I’m happy with this for now.

* I had misremembered the Ove Glove as not being machine-washable, but I just looked up the care instructions and it is. Then I checked on the care label for the pot-holders and apparently those, despite being all-cotton, are “hand wash only”. Oops. (Neither can be tumble-dried.) When the time comes, I’ll probably machine-wash the pot-holder handle holder anyway and see what happens. ↶

I upgraded my iBook G4 to have an SSD

2024-12-30T05:56:00Z

I did take some notes which I’ll present below, but this isn’t a full how-to. I used iFixIt’s guide plus occasional reference to the official Apple Service Source repair guide (those are not strictly public but can be had from your favorite abandonware site).

For this year’s #MARCHintosh, I decided to replace my iBook G4’s 30 GB spinning-rust hard drive with an SSD.

This was my second SSD upgrade, as I’d previously replaced my G4 Cube’s hard drive with an SSD. (The pictures on that page show a hard drive because, before the SSD upgrade, I’d replaced the Cube’s hard drive with another hard drive, and that was what I originally documented on that page. Then, after that, when I decided to upgrade to an SSD, I used my own tutorial. iFixIt didn’t exist yet.)

I rather despise working on laptops, though this wasn’t as bad as I’d worried it would be. (Upgrading the memory in my Mac mini was harder. I pointedly did that as soon as the machine arrived so that it would be done and I’d never need to open the machine back up for the rest of its life.)

The thing that motivated me to go forward with it was that the iBook was absolutely filthy. It had been Mom’s, and she was a smoker in her life; she would routinely be smoking a cigarette and working on the computer, and getting so absorbed in the latter that ash would fall from her cigarette onto and into the computer. So I resolved to clean the disassembled parts as well as upgrade the storage.

For the cleaning, I mostly used paper towels wetted with diluted all-purpose cleaner. A couple small spots of deposited nail polish were resolved with cotton pads soaked with nail polish remover. It worked fine, at least so far—if I’ve started some chemical process of plastic deterioration, I don’t know it yet.

Before cleaning.

After cleaning.

Once I decided the project was go, I also added in a memory upgrade, because it was less than $20 and I was already buying stuff from OWC for the operation anyway. The machine had 512 MB of RAM; now it has 1 GB. (Plus the 128 MB on the logic board.)

One key difference from the Cube upgrade: The iBook, being a laptop, doesn’t have as much space for the upgraded drive. The Cube had some wiggle room taken up by brackets; the iBook has basically none. In the Cube, I installed a standard-size SATA SSD plus a SATA-to-PATA adapter; in the iBook, that wouldn’t have fit.

So my first thought was an M.2 SSD, that being the form factor that today’s computers generally use. I ran into a problem: There are like three different signaling protocols that all run over the M.2 form factor, and M.2 correspondingly has three different keying combinations to guard against protocol mismatches (an incompatible SSD won’t physically fit, though an SSD that fits isn’t necessarily compatible). I noped out of trying to sort that out.

What I went with instead was mSATA. This form factor is kind of dying off as M.2 takes over, but Kingston still sells mSATA SSDs directly from their own website, and I found a suitable adapter on Amazon. (I buy from alternatives like Micro Center or direct from manufacturers whenever possible, but it wasn’t in this case. The manufacturer’s website links to their Amazon store.)

The old spinning-rust drive is 30 GB; the new SSD is 256 GB.

What’d I put in there?

Here’s the bill of materials I ended up with:

Kingston 256GB mSATA SSD (you could go bigger, but I cheaped out)
Ableconn mSATA to IDE adapter
OWC 1 GB PC2700 RAM SO-DIMM (the original Apple DIMM was PC2100, so this was a speed upgrade as well as a size upgrade)
OWC external FireWire SSD to reinstall the OS from, since this machine can’t boot from USB and I don’t trust the internal optical drive to not eat the next disc I put in it

… and as I write this up now, I notice that OWC has now discontinued that last product. Did I buy the last one or something?

(It feels like OWC is moving away from selling products for older Macs, having now discontinued their last FireWire product AFAIK, and I’m bummed. At a time when the retro Mac community is resurgent, we can use every source for quality gear for our old Macs that we can get.)

For tools, I mostly used one of iFixIt’s toolkits, plus the screw extractor set they sell. I also found a lighted magnifier extremely helpful.

I also used iPartition to set up the external SSD with the Tiger and Leopard boot DVDs and various other things (including a copy I’d already made of the HDD).

Hiccups

Hiccup #1: The case of the borked screw

One of the three M2 (i.e., 2mm, i.e., really tiny) screws on the right side of the keyboard tray was visibly rusted and seemingly damaged. Trying to unscrew it (including a couple of attempts with the wrong size driver) damaged it further.

This was after I’d already worked over that third screw a bunch. It was even more rust-covered at first.

I looked at a bunch of options and asked for ideas on Mastodon. A couple of people recommended precision screw extractors (much smaller and a different type than the kind you drill into a screw). I bought the set that iFixIt sells, which it turns out is OEM’d by Moody Tools.

The second extractor bit I tried, one of the smaller ones, worked perfectly, and drove the screw out as easily as the normal drivers did the undamaged screws. Thus I was unblocked and able to continue toward the hard drive.

I did encounter a second rusted screw near the latch mechanism. This I was able to clean with a few things, including a nylon-brush bit on a Dremel, without damaging the head. That screw came out and went back in with no further fuss.

Hiccup #2: The case of the stuck installer

After I got everything back together, the Tiger DVD wouldn’t boot. After either the verbose spew or the spinner, the OS switches to the normal GUI with a blue background, and should then show the Installer on that background—but it would get stuck at the blue background, with no window and no cursor.

On a whim, I tried Leopard (not a supported OS on this machine), and was pleasantly surprised to find that the iBook booted that DVD happily. Moreover, that it worked.

Leopard’s installer refused to install, of course, but it let me cancel and access the rest of the utilities—including Terminal, from which I was able to run system_profiler. Here, I figured, I could at least see whether any of my new hardware was showing up.

And the answer was no. Well, that was a let-down! The RAM slot read as empty (it definitely wasn’t), and querying the PATA bus hung.

I decided to try switching back to the original Apple DIMM, since it’s easier to get to the RAM than the hard drive. When I put the original DIMM back in, I noticed that it sat farther into the slot than the upgrade DIMM had been.

“HMMMMMM.”

So I popped it back out and re-upgraded the RAM. Sure enough, with a bit more attention to this detail, the new DIMM went into the slot exactly as far as the Apple DIMM had.

And when I booted the machine up again… everything worked. The new RAM showed up. PATA no longer hung.

“Hmm. PATA doesn’t hang anymore. Was that where Tiger was hanging?”

Yup! Or similar enough, anyway. It fixed the problem; Tiger’s installer was now unblocked.

Hiccup #3: The case of the 512 MB partition map

One of my goals for this machine is to try to cajole it into booting Mac OS 9 even though that isn’t supported, so I wanted to make sure to install the disk drivers that Mac OS 9 needs onto the SSD. This is an option in Disk Utility of that era… except it wasn’t showing up, presumably because this is a machine that (officially) can’t boot Mac OS 9 anyway, so why would it need that?

(The original hard drive had Mac OS 9 drivers installed. Maybe Panther was more permissive about that.)

Now, there is a second way to get Mac OS 9 drivers installed: hdiutil create has an option, -layout 'UNIVERSAL HD', that sets up the image with a partition map containing a full set of Mac OS 9 drivers for every then-imaginable device interface, from SCSI to ATA to FireWire. (There is also 'UNIVERSAL CD', which gives a different set of drivers.)

So my plan was:

Create a RAM disk using hdiutil create -nomount ram://262144 (that creates a 128 MB RAM disk), since I had nothing writable that I wasn’t about to obliterate (besides thumb drives), and I had plenty of RAM.
On the RAM disk, create a sparse disk image with the “UNIVERSAL HD” format and sized to the number of blocks diskutil info reported for the SSD.
Attach that disk image with -nomount so I can munge its contents directly.
Use pdisk to edit its partition map, specifically to create all the partitions I wanted.
Use newfs_hfs to format each of the partitions as HFS+J (or plain HFS+ in the case of the OS 9 partition).

Problem: Whenever I attached the sparse disk image, pdisk showed that the partition map was set for a 512 MB storage device, not a 256 GB storage device. It doesn’t offer any way to change that, even in the advanced options.

Solution: This was apparently a Tiger bug, because when I rebooted back into the Leopard DVD, the same series of steps worked perfectly.

So I had to do my partitioning and formatting in Leopard, then reboot back to the Tiger DVD to install the OS.

Results

Well, the first sign of success was that the Tiger installation—from the FireWire SSD to the IDE SSD—only took 15 minutes. If you’ve ever installed old Mac OS X from a CD or DVD to a spinning-rust drive, you’ll remember it definitely takes longer than that.

My informal test of the HDD’s speed before the operation showed about 30 MB/sec write speed (dd if=/dev/zero bs=10485760 of=test, followed by math to convert its output to MB/sec because Tiger’s dd doesn’t have status=progress). After the upgrade, I found out about Xbench, and ran that, and it showed about 90 MB/sec write speed in most cases, which is basically saturating the UDMA100 interface. So, with the caveat that the methodologies of the two tests aren’t the same, I feel comfortable saying I tripled my throughput.

Tiger certainly feels way faster than ever before on this machine. The progress-bar portion of the boot process is now a tiny fraction of it, as the progress bar goes from 0 to 100% in about a second.

I’m very happy with the upgrade. I plan to see if I can convince the machine to boot from a Mac OS 9 partition (not a supported configuration—the machine came with a special version of Panther) or install Leopard (also not supported; this machine officially only supports Panther and Tiger plus Classic).

More photos

How dirty this thing was

Note #1: Drive bay shock bumpers

Note the rubber bumpers inside the drive bay, inside the front of the machine. The heads of the screws on that side of the drive bracket (or the IDE adapter replacing it) rest inside these bumpers. There’s also a bracket with matching bumpers on the other side, which should be transferred from the old drive to the new one.

Note #2: Reinstalling the feet

The iBook’s feet have a design that ensures there’s only one way to reinstall them. You’ll need to pay attention to it when putting the feet back in, to ensure you don’t put them in crooked.

The lip in this foot ring mates to the ridge in the recess. If the foot ring isn’t properly aligned, it won’t sit in the recess evenly. Don’t force it! Pull it out, look for the keying, and turn it so it’s correctly positioned.

More keying. As before, pay attention and make sure you’re lined up, and back off and try again rather than trying to force it in.

Just for fun

High-resolution Creative Commons badge

2024-04-07T05:07:45Z

I noticed that the pointers tutorial‘s Creative Commons badge had gone missing. I guess they got tired of people hotlinking it.

So I grabbed it from the Wayback Machine, and when I did, I noticed that it’s in the classic 88×31 format used by so many miniature promotional images. If the phrase “Netscape Now!” means anything to you, you know what I’m talking about.

There’s been a growing trend of making new 88×31 images, some in higher resolutions for modern hi-DPI displays. So I thought I’d do one: redrawing the classic Creative Commons bug as a vector image that could be exported as SVG and as high-res PNG.

Here you go:

The double-resolution PNG.

The SVG.

I think the original image falls under the CC-BY (Attribution) 2.0 license that was then current. I’m happy to place these new images under the same license or the newer 4.0 version.

The third option: Novavax’s covid vaccine

2024-04-07T05:07:45Z

I’ve had four mRNA-based covid shots so far: Pfizer, Pfizer, Moderna, Pfizer.

They’re great protection, of course, but I get harsh side effects from them—two or three days of alternating fever and chills. Not fun. Some folks take ’em just fine, and if you’ve never had an mRNA-based vaccine, I’d encourage you to try it at least once—my experience isn’t universal. But I always have a rough time.

I wanted to get this year’s covid shot back in September, which was the anniversary of my previous one, but I had to wait for availability to settle a bit since shots were hard to come by for several weeks, and I know folks who had appointments and then found out they’d been canceled on the day of due to shortage.

The longer that went on, the more I wanted to get my shot ASAP since we’ve already been in a covid surge for months by this point (per wastewater data) and we aren’t even to the holiday season yet.

Pictured here: Everyone I know, and everyone they know, who’s been coming down with covid over the last few months.

Then I heard that Costco has Novavax. (I later found out that Rite Aid also carries it. CVS might have it but apparently you can’t just book it through the website, you have to ask—weird.)

I went on Costco’s website (which offers both Moderna and Novavax), made an appointment for what is now this past Saturday, and stocked up on my usual post-vax supplies: Gatorade, water in the fridge, clean laundry, and a few low-effort meals and snacks.

Saturday, I got the shot. No difficulty, and my insurance covered the cost—I paid $0.00.

Sunday, I spent the entire day feeling like I had a mild cold. No fever, no chills, just lots and lots of sleeping. Drank lots of water, some Gatorade, and even ate on my usual schedule.

Monday… I was fine. By Tuesday, I was back to 100%.

I spent one whole day with the symptoms of a mild cold.

This is a night and day difference from my experience with the Pfizer and Moderna vaccines. It’s more like my experience with a flu shot: sleep like a cat in a sunbeam for a day, then right back to normal.

The efficacy of Novavax is comparable to the mRNA vaccines—it might be a little lower, but close enough that the difference in post-vax experience makes it well worth the tradeoff. Doubly so if you’ve already had, or might get next time, an mRNA-based shot and want that “all of the above” protection.

If you also have a rough time with mRNA-based vaccines, try Novavax.

If you don’t have RSI, ergonomics are for you

2024-04-07T05:07:45Z

This was originally posted as a tweet thread back in February 2022. For this Director’s Cut Extended Remix, I’ve added the photos and applied styling.

I used my laptop as a laptop for about an hour yesterday and my wrists still hurt. It’s fading but slowly.

So I guess I don’t get to do that anymore. Split keyboard+vertical mouse or nothing.

2023 note: I’ve since replaced this keyboard with an ErgoDox EZ.

Maybe a split keyboard and vertical mouse seem like luxuries, because that’s their market position (the high end), but I promise you there’s nothing luxurious about this.

Really it’s more that there are keyboards and pointing devices that hurt people, and those that don’t.

But I use a one-piece keyboard and a regular mouse/trackball/trackpad and I’m fine!

Well, maybe. Or you’re not injured yet. Or not enough yet to notice without trying ergonomic hardware for a week and gauging the difference.

Ergo hardware can help you stay uninjured.

I cannot emphasize enough how important prevention is. How important it is to protect your hands before they’re injured.

You can’t un-injure them. You can only avoid injury… or not.

Minting a trillion-dollar coin is harder than you think

2024-04-07T05:07:45Z

The trillion-dollar coin is an idea for one way that Democrats could get around Republican threats to throw the United States into default. (There are other options, including standing on the 14th Amendment clause that says “the public debt… shall not be questioned”.)

The basic idea is this: The Department of the Treasury has the authority to direct the US Mint to produce a platinum coin at any denomination they see fit. The value of such a coin is its face value—that is to say, whatever it says it is. So when the Republicans start throwing around threats like “you need to cut off services to these groups of people or else the government is going to run out of money!!!”, as they have been, one option is to simply literally make more money—in enough of a quantity that it will pay for all the US Government’s expenditures for the next year or so and take a lot of the hostage-takers’ leverage away.

I’m not qualified to debate the policy or economics of it, and most likely, neither are you. But what I can do is think of any number of ways that someone—intentionally or otherwise—could fuck it up.

Mis-striking the coin

The last step of the process of producing a coin is called “striking” it. That’s the part where the design gets pressed into the faces of the (hitherto) blank.

Normally coin production is a mass-production process; the country’s mints produce up to tens of thousands of coins per minute. In this case, we’re talking about a one-off, so I don’t know whether they’d do the process differently or just run the machine for a very, very short amount of time.

Either way, it’s certainly possible for the coin to be mis-struck, or otherwise produced in a way that it is obviously defective. This has happened in a variety of ways to nearly every type of coin, and normally, mis-struck or otherwise defective coins are worth significantly more than face value.

What does that look like when the face value is $1,000,000,000,000?

Most likely, the mis-struck coin wouldn’t stay that way—they’d melt it down and try again. (After all, platinum ain’t cheap.) We might never know that there was a mis-struck trillion-dollar coin in existence for some short amount of time.

Composition

The specific section of the US Code that authorizes this stunt says:

The Secretary [of the Treasury] may mint and issue platinum bullion coins and proof platinum coins in accordance with such specifications, designs, varieties, quantities, denominations, and inscriptions as the Secretary, in the Secretary’s discretion, may prescribe from time to time.

The trillion-dollar coin wouldn’t be a bullion coin, which is a coin defined by its amount of some precious metal—that would be a “this much platinum” coin, not a coin with a dollar denomination. So the trillion-dollar coin would be a proof coin.

The Mint regularly issues platinum proof coins, such as this coin for this year. That coin contains 1 ounce of platinum, and has a face value of $100, but is sold at a price dependent on the value of its platinum content, which is somewhere between $1,000 and $2,000—ten to twenty times its face value.

So, for a trillion-dollar coin, how much platinum would they need? Does there need to be a particular ratio, or could they make a zinc coaster with 1 oz of platinum mixed in?

Or does it even matter? Could they make a 1-oz platinum coin, not much different from the ones they’re already making, and just add ten more zeroes across the back of it?

Speaking of which…

The design

There’s at least one artistic rendering of what such a coin could look like, but it’s just one artist’s conception and not an official rendering from the US Mint.

Presumably they’re not going to just type in “ONE TRILLION DOLLARS” in Impact and call it good. This is The Coin! It’s got to look like something.

On the flip side, this whole idea is an emergency measure. They’re not going to have time to go through the usual processes for coming up with new coin designs—not when the US Government could reach the statutory debt limit in… not even a couple of weeks at this point.

Hopefully they’re coming up with a design now that they can have ready to go if and when it’s needed. (And make absolutely sure it has no typos in it.)

Of course, having the design ready to use at a moment’s notice gets into issues of…

Operational security

It’s easy to say “the Mint should do this” or “Treasury should do that” but it’s worth remembering that these are granfalloons, in the sense in which Don Lancaster used the term (slightly different from Kurt Vonnegut’s original meaning):

tactics secret—beware the granfalloon, my son

A granfalloon is any large bureaucratic figment of people’s imagination. For instance, there’s really no such thing as the Feds or the General Veeblefeltzer Corporation. There are a bunch of people out there that relate to each other, and there’s some structures, and some paper. In fact, there’s lots and lots of paper. The people sit in the structures and pass paper back and forth to each other and charge you to do so.

All these people, structures, and paper are real. But, nowhere can you point to the larger concept of “government” or “corporation” and say, “There it is, kiddies!” The monolithic, big “they” is all in your mind.

If the Mint produces a trillion-dollar coin, it’s because people designed it and people fabricated it. If Treasury deposits the coin into the Federal Reserve, it’s because someone from Treasury personally visited the mint where the coin was struck, took possession of it, carried it to the Federal Reserve, and deposited it.

There are so many ways that could go wrong.

Every person involved in this would need to be vetted sixteen ways to Sunday. No foreign allegiances, no debts, not even a whiff of past criminal activity.

And you’d need a significant number of people. Nobody gets to go alone; you’d need multiple people monitoring each other, all with bodyguards, while also trying to remain as inconspicuous as possible and not look like they’re carrying the most valuable single object in the country.

Assuming, of course, that it remains a single object. I mentioned above that the production run would be a one-off. It would be supposed to be, at least—but as soon as the die exists, it’s theoretically possible to strike a second blank and make another trillion-dollar coin. Either a counterfeit, if it doesn’t actually contain the platinum, or a duplicate if it does.

You could argue that theft or counterfeiting are not actually as big of a concern with this project as they might be with, say, one-dollar coins. Supposing you stole the trillion-dollar coin, or struck a duplicate—what could you even do with it? Nobody will accept it as tender. No commercial bank or credit union will accept it in deposit; they’d immediately phone up the Secret Service and be like “yeah we found your coin”. What could you do, put it on eBay?

Small things, easily lost

Even barring any acts of malice or greed, what if the Custodian simply… lost it?

Coulda sworn it was in that pocket.

Did it fall out when I was paying for lunch?

Hope it didn’t roll into a storm drain…

But let’s say none of that happens and the Custodian makes it to the Federal Reserve with the solution to the debt ceiling crisis safely on their person.

Then what?

When you deposit hard currency—including coin—at your local banking institution, they put it in their drawer and mark up your account. From that point, the physical coins you left behind are then eligible to hand out to any other customer in service of a withdrawal, or to be transferred between tellers or between branches. They are fungible; the bank has hundreds or thousands of them and there is no particular reason to care about the location of any single one of them.

The trillion-dollar coin would be an extremely different situation.

To be fair, it is a (mostly) solved problem. The New York Federal Reserve stores gold and other reserves on behalf of various governments, including the US. It may also be that other Federal Reserve Banks around the country offer similar services. The trillion-dollar coin would likely end up at any of those locations.

Buuuut there are some differences.

First, it’s not a stack of gold bars. It’s a coin. Its value wouldn’t derive from its scrap metal value (as noted above, somewhere in the 1- to 2-kilobuck range) but from its denomination.

A stack of gold bars is hard to exfiltrate. Maybe a thief could remove a bar or two (if they somehow got past all the security) without anybody noticing. It’d be an extremely high-stakes game of Jenga. (Don’t ask me for tips; everything I know about this sort of crime I learned from heist movies, and I haven’t watched many heist movies.)

A coin is, well, a coin. People regularly carry dozens of them on their person without anyone noticing. When you go through a metal detector, you dump your coins into a pile in a little plastic tray and nobody looks at it.

Where would they even keep The Coin? Do they have little safety deposit boxes at the New York Fed?

That leads to the other problem: Keeping track of it.

Somewhere there needs to be a record of where, in the New York Fed or wherever else, the coin is kept. It needs to be in a place where (theoretically) someone from Treasury could retrieve it if there were ever a need to do so, not to mention a place that could be checked if there were suspicion of theft. Of course, that would also be sensitive information; you wouldn’t want anyone in the whole organization to be able to look up where the USG’s trillion dollars is.

Some of this is, again, solved problems or otherwise not worth worrying about. It’s a bank; not a normal bank but still a bank that (one hopes) has a means to keep even something as small as a single coin in a safe place, remember where that is, and guard access to both that location and the knowledge of it. And, as I mentioned above, theft is of limited concern for a coin that there is (or should be) only one of and that no place will accept.

On a more serious note

None of this is to say that they shouldn’t do it; that’s more of a policy and economics question. Government works on hard problems all the time, and usually does better than we give it credit for. (Especially better than libertarians give it credit for.) Success is the expectation, and I’d argue it is actually the norm, but we don’t notice it and don’t appreciate it. Failure stands out, and certain actors are ideologically motivated to spotlight it. I’m more interested in anticipating failure as a means to ensuring success.

A lot of the difficulties I’ve outlined arise from the unique nature of this particular minting job. It’s a singular coin of exceptionally high face value. Processes that are normally routine become high-stakes; hazards that are normally negligible become serious concerns.

I really hope the folks at Treasury have thought about this more than the couple of hours I put into this blog post. Because with less than a couple weeks left of “extraordinary measures”, if the Republicans keep trying to hold the country hostage by threatening to throw it into default, we might need this to go from “wild idea” to “thing we are actually doing” in a hot second.