• Posts
  • RSS
  • ◂◂RSS
  • Contact

  • The Value of Learning Names

    November 5th, 2010
    lesswrong  [html]
    I just read "fake explanations" on less wrong. Eliezer yudkowsky presents children who have learned that in response to questions like "why is this side of the object warmer than the other" you should guess at the teacher's 'password' and respond with "heat conduction". He goes on to argue that we should not mistake this for real learning because the children have no more knowledge than if they had learned to respond "because of magic". The children have not actually learned enough about the theory of how heat moves to use it to make predictions. They've just learned a 'password'.

    I disagree: I think there's value in teaching children that "stuff falls because of gravity", "magnets stick to the fridge because of magnetism", and that "the prime minister is in charge because they were elected", but it's not value of the kind that immediately improves our ability to predict things. Human knowledge and even the subset that is science are both so large that we need a good index. Learning that "gravity" is the response to "why does stuff fall?" is helping me build my index. I can't yet predict correctly that a feather will fall just as quickly as a bag of rocks if we drop them in a room with no air, but I know where to go for a better theory when I need it.

    Sometimes we do this on purpose, such as when in chemistry class we learned the ideal gas equation (in enough detail to make predictions) and then were told that there existed "van der waals corrections" that made the equation more complex but also made it more accurate as gas molecules got heavier and you got farther from standard temperature and pressure. This didn't let us improve our model of the world, but it told us the name of a better model and the conditions under which we might need to seek it out.

    I agree that it's important not to think that because someone (or you) knows a fact like "matter is made of waves" they understand more about the world. I learned that fact in school, but I couldn't make any predictions based on it. As a component of my index-of-science, though, it might come in useful some day.

    Recent posts on blogs I like:

    What should we do about network-effect monopolies?

    Many large companies today are software monopolies that give their product away for free to get monopoly status, then do horrible things. Can we do anything about this?

    via benkuhn.net July 5, 2020

    More on the Deutschlandtakt

    The Deutschlandtakt plans are out now. They cover investment through 2040, but even beforehand, there’s a plan for something like a national integrated timetable by 2030, with trains connecting the major cities every 30 minutes rather than hourly. But the…

    via Pedestrian Observations July 1, 2020

    How do cars fare in crash tests they're not specifically optimized for?

    Any time you have a benchmark that gets taken seriously, some people will start gaming the benchmark. Some famous examples in computing are the CPU benchmark specfp and video game benchmarks. With specfp, Sun managed to increase its score on 179.art (a su…

    via Posts on Dan Luu June 30, 2020

    more     (via openring)


  • Posts
  • RSS
  • ◂◂RSS
  • Contact