School as Research
A reply to Riley Drake, then a young biology researcher, on the question of whether new tools (computer-based learning platforms like Khan Academy and Coursera, decentralized clinical research, and so on) can replace traditional academic institutions. Alyssa argues that the question is wrongly framed: the school system is not a standalone tool but a load-bearing component in a much larger, post-WWII social structure, and you can’t simply “swap out” one piece without the rest collapsing.
The modern American educational system is best understood not in isolation, but rather as a component within a much larger system (set up during and shortly after WWII, and now gradually breaking down). For instance, in the photography world thirty years ago, it might make sense to ask “can the digital camera really replace the film camera?”. But going back to 1400, it wouldn’t make so much sense to talk about “can the gun really replace the bow and arrow”; “replace” isn’t really the right word. All of military strategy for thousands of years had been based on the limited effectiveness of ranged weaponry. You can’t just swap out “gun” for “bow”; if you pull out that one brick, the entire structure comes tumbling down around it.
People right now are asking questions like “can computer programs teach kids algebra like schools?”. But once you step outside the system like that, you have to ask questions like “what is the most effective way for people to learn algebra?”, which then leads to “what do people actually use algebra for?”, which leads to “what should people learn in order to achieve their goals?”, which leads to “what goals do people actually have?”, and at that point you’re pretty much re-designing everything from the ground up. E.g., the ultimate reason almost everyone learns algebra is some combination of “because they’re told to” and “because employers require it”. Once you are in C code rather than a government-funded classroom, there’s no longer anyone telling you what to do, or anyone requiring anything. Initially, what you get is a cargo cult, which you see a lot of in places like Coursera and Khan Academy — people keep on learning algebra because it’s what they’ve always done. But fairly soon that breaks down (Paul Graham talks about this in “Hiring is Obsolete” , under “The Open Cage”), and then you have nothing supporting you but a yawning black hole. And it’s honestly a little scary to realize how deep that hole goes.
E.g., if you look at history, you can go to Amazon, and pull up books about any culture or historical period you like. Read the reviews. Almost all of them will be from people who purchased the book in school, because they were told they had to to graduate, because they were told they had to “to get a good job”, ultimately because big companies are still riding on sheer inertia from 1952. And so we can infer that almost no one is actually learning history — almost all the history books are bought by people who forget them two weeks after the test. The Dutch Republic of 1600, the British suppression of the Thuggee, the Guns of August, the Opium Wars, the Taiping Rebellion… they have all been forgotten, almost as thoroughly as if a nuclear firestorm had burned down the libraries. Countries like Egypt write constitutions in imitation of an imitation of an imitation of a form of government that worked pretty well in 19th-century America with 19th-century technology, because the skill of designing governments based on historical data about what is likely to actually work has been lost. E.g., in 1000, the Chinese Empire was arguably the most powerful state in the world. In 1900 it was in ruins. Why? It seems like a pretty important question, and we have copious records detailing every step of every year, and yet I have no idea.
See e.g. Paul Lockhart’s A Mathematician’s Lament for version 0.1 of what new math might look like, and e.g.
CGP Grey for version 0.1 of what new history might look like.
On clinical trials, obviously a huge amount of that is just the FDA/DEA being evil, but of course lots of stuff won’t have immediate applications anyway (e.g. quantum mechanics when first discovered in the 1910s).
Dario is great; he’s on MetaMed’s advisory board.
Sarah Constantin’s blog is an excellent source for stuff like looking at biological data analysis, how it isn’t being done now, and how we can apply it to useful problems.
Everyone in the startup industry realizes how important employee quality is, and yet no one (as far as I know) has a systematic way of evaluating it that’s newer than “outsource the problem to a bunch of college admissions people who don’t know what a cosine is”. What an almighty bang that’s going to make when it comes crashing down.