Unintended Consequences

Terminator Responses

2009-05-26T11:40:00.000-07:00

Via Instapundit, H+ magazine has a bunch of interesting responses to the latest Terminator movie.

I'd never heard of Josh Halls before and am now very interested in his book Beyond AI: Creating the Conscience of the Machine.

Chris Frith on the Brain Science Podcast

2009-05-26T10:30:00.000-07:00

I enjoyed another episode of the Brain Science Podcast this morning.

I loved this line in particular (emphasis mine):

Perception is not a passive process because, as I said, the only way
you can find out about the world is through the errors in your model. And the
very best way to do that is to actually act upon the world, so that you say, given
my model this is what ought to happen if I do this, and then you can find out
whether it does happen and adjust your model accordingly

And if you, like me, are getting confused by all the different things you hear about dopamine then this exchange quite helpful:

GC: And in the imaging work we see, is that one of the ways we know the
dopamine systems are involved here? That when we correctly predict what’s
going to happen, that gives us a good positive dopamine signal?
CF: Oh, no, no. Quite the other way around. The dopamine signal is a
prediction error. So, basically if something unexpectedly nice happens, then you
get a shot of dopamine; and so, the dopamine neurons become more active. And
if you expect something nice to happen and it does happen, there’s no response;
because there’s not an error. If we expect it to happen and it doesn’t happen,
then the activity goes down. So, that’s a negative error.

Chris Frith's Books is Making Up the Mind: How the Brain Creates Our Mental World.

Another interesting point that came up in the podcast is that many imaging studies are validating results that psychologists "knew" from more introspective experimentation, and making old news into news again.

Semantic Web

2009-05-14T08:42:00.000-07:00

Via ReadWriteWeb I find a good three-part read on the semantic web making the rounds from Semantics Incorporated:

Part 1: Tying Web 3.0, the Semantic Web and Linked Data Together --- Part 1/3: Web 3.0 Will Not Solve Information Overload
Part 2: Tying Web 3.0, the Semantic Web and Linked Data Together - Part 2/3: Linked Data is a Medium
Part 3:Tying Web 3.0, the Semantic Web and Linked Data Together - Part 3/3: Structuring Chaos

My favorite paragraph occurs in the third installment:

What’s the endgame for the Semantic Web? I’d propose that it is a web where any information you input is immediately cleaned up, pre-structured and pre-connected to the rest. There is a variant of this vision that would see any information input remaining in its raw format until one needs it, at which point it is structured and connected on the fly, using the perspective of the person who queried to shape the structure and the connections. The problem with this vision obviously is that unless you have scouting agents that can query the whole web instantaneously for every query, and structure and link data on the fly -- and I think we can safely say that’s not going to happen anytime soon -- you need some pre-defined structure and connections so you know what the information is about and where it is. We need to meet those agents half-way.

Neologism Watch: Group Formation

2009-03-10T11:29:00.000-07:00

A key idea in Clay Shirky's Here Comes Everybody: The Power of Organizing Without Organizations is that the cost of group formation has fallen precipitously, so we have an explosion of new groups.

Sooner or later a group identity starts to define itself in opposition to other groups. So if there are more and and more groups, and each individual identifies with more and more groups, sooner or later more and more people are going to be in groups that are mutually opposed to each other.

And we're going to need a word for this.

I've been waiting to post this until I had a good example. Well this morning I was reading Life of Pi and found what I've been looking for. As a teenager, the main character Pi is fascinated by religions and spends a lot of time in a church, a temple, and a mosque. The leaders of each house of worship think he is devoted to their religion, and his parents know nothing about any of it. Then one day the boy, his parents, and all three religious leaders run in to each other on the street and it all comes out in the open. All manner of inter-group animosity comes rushing to the surface. The poor boy wants to focus on the unity of all the religions but not one of the groups is so happy with him anymore.

'Cognitive dissonance' is a phrase that would describe the individual, internal nature of this confusion but what's a good word for the group dynamic?

Practice, Practice, Practice

2009-01-03T07:41:00.000-08:00

There's an interesting discussion of "practice" among O'Reilly authors on the O'Reilly Radar.

My favorite quote from the whole long discussion is by Brett McLaughlin:

Practice must, over time, simulate deeper understanding. Good practice, over time, connects ideas. Bad practice, over time, only creates muscle/mind memory. Practice structured correctly will eventually create a fluidity with the mechanical components of [insert discipline here], and free the mind up to consider the bigger picture.

More good follow-up on the topic from Simon St. Laurent including:

Frankly, I think that "mastery" is usually the wrong goal, a strange habit in our culture of setting ourselves up to fail. Mastery happens, but we need to remember - and value - the intermediate steps. Most of us will never be the Outliers Malcolm Gladwell describes as "The Story of Success".

The opposite of the "success" he describes, though, isn't failure. It includes a wide range of competencies, of people getting things done, without necessarily hitting "the big-time". You don't have to become an outlier to be a success - and you shouldn't let that get in the way of practicing.

Personally I find "practice" vaguely humiliating. It exposes the gap between understanding a concept and being able to fully express it. I'll read something and think I understand it perfectly only to fail the first practical test of applying a concept. But that doesn't mean there is no part of me that understands the concept, just that there are more parts that do not. Instead of thinking rather simply that "I know something" it is more appropriate to think of the percentage of my being that has now embodied a concept. After a lot of practice over the last month, "I know how to play the first Goldberg Variation". But do I? I know the notes, I've got good fingerings worked out, and if I start at a reasonably tempo and am relaxed I can probably get through it. But if I'm a little nervous or distracted, I'll screw it up bigtime. Furthermore, if I pose myself the challenge to play the left hand and improvise with the right hand I'll crash and burn. So maybe I should say that after tons of practice I've embodied about a 20% knowledge of the first Goldberg Variation. Lots more practice to go.

Megan McCardle on Villains

2008-11-26T14:26:00.000-08:00

Megan McCardle is still one of my very favorite bloggers. Here are a few choice quotes from her today on the search for the villains behind the financial crisis:

Nature is not a novelist. Reality does not come packaged in narrative form, and rarely gifts us with either true heroes, or true villains.

And later:

A failure this massive can only occur if massive numbers of people had their hands in it somehow. If you want to find a villain, there's probably one handy at the nearest reflective surface.

Long Now Seminar: Peter Diamandis

2008-11-18T10:43:00.000-08:00

The Long Now Seminar's Podcasts are almost all worth a long careful listen. The latest I heard (MP3) features the founder of the X-Prize Foundation talking about the amazing art of using prizes as leverage to encourage audacious innovations. Here is a thorough written summary.

Towards the end there is a long list of potential future prizes and the audience is asked to pick their three favorites.

It's an interesting list. I'm going to pick some relatively mundane, though not unambitious ones:

* Develop a teaching system that allows an increase learning rates by an order of magnitude.
* Create an AI that can engage and educate children to their highest potential
* Eradicate poverty for > 90% of the human population

The first two are extremely non-mutually-exclusive, but I personally am more interested in pursuing the "teaching system" as it is agnostic about how it is to be implemented. Maybe it's an AI, maybe AAI (Artificial Artifical Intelligence), maybe something else entirely.

With 2 young children, I have a ~~low cost~~ experimental platform for testing. Time to get to work.

More on the X-Prize Foundation's thoughts on education here.

Doctors Without Borders in Myanmar

2008-05-08T13:42:00.000-07:00

Doctors Without Borders is on the ground in Myanmar. They can be supported here.

They're Back!

2008-04-24T23:04:00.000-07:00

My dear friends Danny and Juliann are back from a long journey studying meditation and they've come back with a whole lot to show for it!

Knowledge Chunks and Mastery

2008-03-25T22:44:00.000-07:00

Many of us software developers will enjoy this excellent blog post by Jeff Moser about becoming a Grandmaster Developer (
via Scott Hanselman). It's another good source in the genre of Introspectively-Improving-Your-Thinking-For-Software-Oriented-Type-People such as this post by Ed Boyden and the previously mentioned book Refactor Your Wetware.

But I am most interested in it for the discussion it kicks off about thinking and learning in general. Two aspects really caught my attention: the pain on the way to mastery, and the concept of named "Chunks".

Pain
Most of us have experienced a painful stage of the learning process at some point or another. But why is it painful? What precise neurological events are occurring during such learning that are experienced as pain?

A couple of ways that understanding this pain could guide the learning process are:

The Deal-With-It TheoryLearning in depth will always require some kind of neurophysiological changes that will be painful

The Work-Around-It Theory There are painful and non painful ways to learn, and if you find YOUR optimal learning process you can attain mastery without pain.

Are there any world-class pianists that do not practice scales and tedious exercises far beyond any point of pleasure? Or did many of them attain that level of mastery because they truly learned to enjoy even that arduous digging for the last mile of perfection?

Are there any world-class developers who do not force themselves to read through painful, tedious specifications or tortuous bodies of source?

Someone who enjoys such process do so by training themself to embrace the pain like a long-distance runner addicted to endorphins, or do they find learning mechanisms that avoid whatever process causes everyone else pain to being with?

When you are struggling through the painful part of learning something, is the pain an inevitable waste product of a successful engine of learning, or is it actually a sign that you're approaching everything wrong?

Perhaps it can be useful to evaluate specific learning situations through the lens of such dichotomies.

I've been learning, rather pleasurably, a lot from blogs and it is useful; but perhaps in so doing I am just avoiding deeper but more painful learning opportunities such as studying text books and the formal academic peer-reviewed literature.

Chunks
Reducing complex patterns to simple names is all the rage, isn't it? Software examples are well-known (Design Patterns, Refactorings, Micropatterns), but I wonder about other domains. In medicine each disease is essentially a complex pattern simply named. In music I know only the basic patterns "chord", "key", but I could not tell you from memory what something like Sonata form. I suspect one could accellate one's mastery of any new domain by explicitly embarking on a process of cataloging the complex patterns. That's the good news. The bad news is that to master something, according to the article that started this line of thought, you may need 50,000-100,000 entries in your instantly-available mental catalog of problems. That is going to take a while, indeed!

Medical Tourism

2008-03-22T10:08:00.000-07:00

Great article at The Futurist about the trend towards medical tourism and how that could produce a healthy competitive pressure on the U.S. health care system. Blue Cross of South Carolina is already looking to cover it!

Economics Witticism of The Day

2008-03-16T23:06:00.000-07:00

Perhaps Helicopter Ben should start pumping anti-depressants into the Wall Street water supply. --Megan McArdle

Exciting Brain Simulation Project

2008-03-14T07:51:00.000-07:00

Overcoming Bias reminds me to pay attention to this amazing neocortex simulation. It models 10,000 neocortical neurons of a 2 week old rat, and each is "single simulated neuron is really the sum of 400 independent simulations".
http://www.blogger.com/img/gl.link.gif
The referenced Seed article is well worth the read. Here are some of my favorite parts:

A Whole New Kind of Neuroscience

According to Markram, however, this scientific approach has exhausted itself. "I think that reductionism peaked five years ago," he says. "This doesn't mean we've completed the reductionist project, far from it. There is still so much that we don't know about the brain. But now we have a different, and perhaps even harder, problem. We're literally drowning in data. We have lots of scientists who spend their life working out important details, but we have virtually no idea how all these details connect together. Blue Brain is about showing people the whole."
In other words, the Blue Brain project isn't just a model of a neural circuit. Markram hopes that it represents a whole new kind of neuroscience.

Tangibilizing

Once the team is able to model a complete rat brain—that should happen in the next two years—Markram will download the simulation into a robotic rat, so that the brain has a body. He's already talking to a Japanese company about constructing the mechanical animal.

Proving the project has absolutely no lack of ambition, they hint at the possibility of reconstructing the rats conscious experience from the model!

Rendering cells is easy, at least for the supercomputer. It's the transformation of those cells into experience that's so hard. Still, Markram insists that it's not impossible. The first step, he says, will be to decipher the connection between the sensations entering the robotic rat and the flickering voltages of its brain cells. Once that problem is solved—and that's just a matter of massive correlation—the supercomputer should be able to reverse the process. It should be able to take its map of the cortex and generate a movie of experience, a first person view of reality rooted in the details of the brain.

Info Trail:
Overcoming Bias

Seed
IBM Research
Whitepaper

Eagerly Awaiting Twine

2008-03-13T16:14:00.001-07:00

I, like many others, am eagerly awaiting the rollout of Twine; apparently you can accelerate your journey through the wait list by blogging your enthusiasm.

I've been consuming a lot more information than my little brain can easily access and am seeking better ways to organize it. Blogging, text files, and Mind Mapping (FreeMind and MindMeister are my tools of choice) have all been nice and useful tools of this process. But Sometimes I use simple text files, and more often I've been using Mind Mapping. (FreeMind and MindMeister are my tools of choice). In practice though I don't find myself going back to my archives very often. Twine is exciting to me because it promises to be a more intelligently organized information repository, and assist with automated discovery of new relevant information.

It is especially important to me to have programmatic access to my personal knowledge repository, so I can keep dreaming up new ways to play with the data. Twine plans to expose it's data in RDF via an API and I can't wait to start playing around with it.

Let the Computers Drive

2008-03-11T13:34:00.000-07:00

Here's one more reason why we need computers doing the driving for us:

I always try to drive with a bigger space around me than most people on the road seem to think is necessary. I think of this as a "probability buffer" that I protect myself with, and it is proportional to the number of children in the car. To maintain such a buffer I probably slow down more than necessary sometimes. Perhaps I thereby occasionally help cause such waves, but more often than not I suspect that the buffer around me limits the number of times that I need to suddenly slow down and the net effect of my habit is good for the traffic flow surrounding me. I hope so, anyway!

Trail to source:

Cognitive Daily

Slashdot

New Scientist

New Journal Of Physics

Alan Kay on Teaching

2008-03-10T23:31:00.001-07:00

It's hard to keep up with all the great talks coming out of Ted.com but this one by Alan Kay is particularly worth watching:

Notable moments include an intuitive method of teaching Pythagorean theorem and building them up to understanding diiffferential equations through shape play!

It also inspired me to finally start playing with eToy's on my kids' OLPC. What a great way to introduce programming concepts.

On the Rightness of Wrong

2008-03-03T23:00:00.001-08:00

I'm a big fan of Paul Graham's writings and this new essay does not disappoint. A feeling that something is wrong is an expected consequence of a good idea!

People look at Reddit and think the founders were lucky. Like all such things, it was harder than it looked. The Reddits pushed so hard against the current that they reversed it; now it looks like they're merely floating downstream.

Worth it For the Title Alone

2008-02-20T10:24:00.001-08:00

How can I NOT buy a book called Refactor Your Wetware?!

It reminds me of a discussion on 10outof10.

Mindfulness and Short Term Memory

2007-12-07T11:13:00.000-08:00

This morning, after forgetting to put my son's lunch in the car about 20 seconds after seeing the lunch and making a mental note to remember it, I started wondering about the link between Mindfulness practice and short term memory.

Perhaps when one is not "Mindful" one keeps feeding a stream of new data to the short term memory faster than it can keep up.

I suspect there's been a fair amount of work on this subject, but all I could find so far are this article
and this very interesting sounding book Attentional Processing: The Brain's Art of Mindfulness. I'm tempted to rush out and buy the book now but it's 12 years old which is a long, long time in field of neuroscience.

Microlending to the homeless?

2007-12-06T23:22:00.000-08:00

The Economist had an interesting story about homelessness back in October that got me thinking.

Despite the perception that it consists mostly of single men with drug, alcohol or mental-health problems, the majority are families, singles or young people who simply cannot afford housing, says Nan Roman at the National Alliance to End Homelessness. The group reckons that 600,000 families with 1.35m children experience homelessness each year, accounting for about half of the national total. Many are in dire need for a relatively short time, living in shelters for just days or weeks.

Could you imagine a direct donor-to-recipient service to buy hotel rooms for people that need them in a pinch. Like the wonderful Kiva but for providing emergency housing to the homeless here in the US.

It is easy to see an endless list of problems with this approach, so I will just focus on the hypothetical positive side.

Presumably on cold winter nights many shelters are full and forced to turn people away. What if there was a site that aggregated weather data, hotel availability, homeless shelter availability, and counts of people in urgent need that night? An individual or social service organization could register someone's urgent need such that warm, comfortable, charitably-minded people could have a chance to type a few keystrokes and paypal somebody off the streets for the night?

Something about this idea seems kind of horrifying. Can you imagine carrying your kids around, having a shelter sign you up on some website, and sit around for a couple hours to see if someone sends you money? But that is less horrifying than if there was no hope for the night at all.

Perhaps there is nowhere near enough money conceivably available from the private sector. Maybe such a real-time emergency housing portal could even be an efficient way for governments to spend their homelessness dollars.

The same Economist article credits the work of Dennis Culhane at the University of Pennsylvania with inspiring governments to help the homeless more proactively. On the homeless of New York:

Each of them used up an average of $40,000 a year in public services, such as hospital care and jail time. When half of the group was offered public housing (coupled with services such as counseling) that group's time in hospital and prison fell dramatically. The net result was a big improvement in the problem at little extra net cost.

Semantic Web

2007-11-19T23:16:00.000-08:00

Notes from a nice article in the December Scientific American.

Update: Full Screen View on MindMeister.

Understatement of the Day

2007-02-10T12:50:00.000-08:00

"They take good care of us, and having my child and husband with me makes me very happy," said [a convicted murderer]. "But this is not the best place to bring up a child. In some ways they are imprisoned too."

On raising her child in a Spanish family style prison.

Dictionary for the Future

2007-01-12T07:33:00.000-08:00

Well, really more for the present.

Daly New Year's Resolution

2007-01-02T23:03:00.000-08:00

Enough about trying to get in better shape or read more books: how about stating New Year's resolutions in terms of Disability-Adjusted-Life-Years?

I will strive to save 100 Disability-Adjusted-Life-Years.

I don't know if this is difficult or easy. If I figure it out I can work up from there.

WHO defines the Disability-Adjusted-Life-Year (DALY) as follows:

The Disability Adjusted Life Year or DALY is a health gap measure that extends the concept of potential years of life lost due to premature death (PYLL) to include equivalent years of ‘healthy’ life lost by virtue of being in states of poor health or disability (1). The DALY combines in one measure the time lived with disability and the time lost due to premature mortality. One DALY can be thought of as one lost year of ‘healthy’ life and the burden of disease as a measurement of the gap between current health status and an ideal situation where everyone lives into old age free of disease and disability.

The first step is to be clear on what this New Year's Resolution even means. Since DALY itself is a statistical abstraction, it seems fair to count relatively abstract contributions. I don't have to stand on the street waiting to find a child to save who was about to get hit by a car. If I find a project that is spending $1,000,000 to save 100,000 DALYs and I give it $1000 then perhaps it's fair to consider my resolution resolved.

And if it's that easy then I'm aiming too low.

There is a counterargument that is worth mentioning: in a large project it is unlikely that my paltry contribution made the difference in whether the project was fully funded or not, and any multifaceted organization asking for my money for a particular project very well many be creating a kind of accounting fiction. If it is a high value AIDS project they were going to fund it one way or another, so my marginal contribution may in the end actually have more influence on whether they buy a new copy machine this year or next. I don't have a good answer to this counterargument, but I'm just going to plain old overlook it in my own accounting for now.

The cost of "buying" a DALY is of course widely variable. My first task will be to figure out more about this. In my initial searching, it seems that numbers in the $5-$100 per DALY in very poor countries are plausible. I wonder if I call my favorite charities if they will think I'm crazy to ask for the approximate cost/DALY of their services. I would love to see or build a website that let's you "shop" for DALY and other public health/economic development outcome measurements and match you to charities or other organizations able to report their effectiveness in that category. Maybe I'll have to build it but surely LazyWeb will save me the trouble?

        Buy Now!

       10 DALY in Country X
            Buy Now from NGO A for $293
            Buy Now from NGO B for $317
            !Best Value! Buy Now from NGO C for $360 $180 (Sale! Limited Time Matching Grant. Buy now for just $180)


        Donors Who Bought this Also Bought:
                                  Clean Drinking Water for 100 in Country Y

What if down the road I work up to a goal of 10,000 or even 100,000 DALYs saved, what is the best way to go about accomplishing this?

1) Change careers to specialize in an area most likely to save DALY?

Public Health
Medicine (It's way too late for me here. I know very well how long that path is.)
Biology
Medical Informatics

2) Deploy my software development skills in some way, like:

Generate new fund raising tools for non-profits
Prostate myself to some existing non-profit and volunteer or work cheaply to do whatever they want done.
Create a magic new information processing system that magically saves lives.
Build software to help existing organizations meet their goals more efficiently.

3) Ignore direct DALY conservation in professional work but try to make as much money as possible in whatever legal/ethical means are available so as to have more to donate to the most efficient providers of the final "product". It looks the least noble while it's happening but in truth would probably be the most effective.

4) Politics. Focus money expenditures on trying to effect some policy change to promote economic growth in developing countries. Ending rich world farm subsidies is an obvious example.   Probably this is where the most good could come from but where I personally have the least possible chance of making any impact.

5) Get a job as a venture capitalist funding bio-tech companies. Ultimately more good is likely to come from new technologies than trying to rearrange the distribution of those that we have, and many of the new technologies are likely to have some venture capital involved in their road to fruition. But there's not much comparative advantage for me in this space either.

6) Economics. I know hardly any economics but it fascinates me. If economists truly have so many wonderful unimplemented ideas about more efficient ways to promote economic growth, I could learn them and do whatever I can to publicize them.

7) Education: help raise a bunch of smart kids to grow up studying all aspects of the problem and run off to the world to save DALY.

A tragically small portion of humanity is currently reaping the benefits of the technology we have. Economic/Policy "technology" is just as important as the more tangible kind. Market economies seem to trend towards prosperity with great velocity; other systems much less so. The problems are apparently deeper than anyone can quite figure out, but it's obvious that things could be arranged better.

For a small reality check, here's a few sources that quote cost-effectiveness per DALY:

	Office of Health, Infectious Diseases, and Nutrition (PDF) To provide USAID and the Bureau for Global Health with assistance to decrease the burden of 13 diseases, recognized as 'the Neglected Tropical Diseases' (NTDs) that inflict economic, psychosocial, and physical damage on the poorest populations in the developing world. These NTDs account for over 500,000 deaths, 50 million disability adjusted life years, and hundreds of millions of dollars lost in decreased productivity each year.
	Cost-effectiveness of malaria control in sub-Saharan Africa. FINDINGS: In a very-low-income country, for insecticide treatment of existing nets, the cost-effectiveness range was US$4-10 per DALY averted; for provision of nets and insecticide treatment $19-85; for residual spraying (two rounds per year) $32-58; for chemoprophylaxis for children $3-12 (assuming an existing delivery system); for intermittent treatment of pregnant women $4-29; and for improvement in case management $1-8. Although some interventions are inexpensive, achieving high coverage with an intervention to prevent childhood malaria would use a high proportion of current health-care expenditure.
	Cost-effectiveness assessments important for HIV response in Africa Cost-effectiveness varied greatly between interventions. A case of HIV/AIDS could be prevented for $11, and a DALY gained for $1, by selective blood safety measures, and by targeted condom distribution with treatment of sexually transmitted diseases. Single-dose nevirapine and short-course zidovudine for prevention of mother-to-child transmission, voluntary counselling and testing, and tuberculosis treatment cost under $75 per DALY gained. Other interventions (eg. formula feeding for infants, home care programmes, and antiretroviral therapy for adults) cost several thousand dollars per infection prevented, or several hundreds of dollars per DALY gained.
	Gates Foundation Global Mobilization for HIV Prevention (PDF) Available hiv prevention strategies are extraordinarily cost-effective. Basic hiv prevention interventions cost substantially less per disability-adjusted life year (daly) saved than the accepted $50 per daly cost-effectiveness threshold for health interventions in resource-poor settings—$1 per daly for condom distribution to women with multiple partners in sub-Saharan Africa, $5 for implementation of basic blood safety practices, $4 to 7 for peer education targeting sex workers, $5 to 12 for nevirapine to prevent mtct, and $12 for std control. Indeed, existing tools to reduce hiv transmission appear to be among the more cost-effective standard interventions currently available for health conditions that primarily affect resource-poor countries.

Optimism

2007-01-02T12:07:00.000-08:00

Of an interesting variety.