Feedback dominates ( aka Rates not States)If anything, I consider this to be my primary contribution ot the blogosphere these last 5 years. In education? Feedback systems dominate, and therefore lecture is moronic. In government, all government systems that have ever existed have feedback systems that push towards the impoverishment of the citizenry in favor of those in the government. In software, the feedback systems determine how good the software is. In Manufacturing, feedback is all of quality. Evolution. Science. Bayesian Epistemology. It's all feedback all the way down. 1 idea...that's all.
Well, this afternoon, thanks to a couple influences, I had my second important insight. I've danced around it before, and I've referenced topics near it in Aretae's 4th law...which is about to be promoted. The 4th law is:
Brain Bunnies (previously known as Monkey Brains)Your brain doesn't work how you want it to. Ever. There's lots of little bunnies running around in your head, each with it's own goals/systems. The conscious mind is only the press secretary for the rest of your brain...making up answers about why in order to support the herd of brain-bunnies who have independently decided what to do, and now need verbal cover.
Moving on to the real insight...thank you Tom and Andrew for setting up my insight:
Verbal learning is THE education problem.Human beings' brains are a very small percent verbal.
Only a very few excessively verbal people are even poor (as opposed to abysmal) at verbal reasoning...and the folks who can make any sense at all of verbal nonsense are hailed as heroes in our age. They end up as the denizens of the cathedral. And of course, as the most verbal folks, the ones who spin the words, their inclination is not to point out how bad the verbal capabilities of the best of us are.
Therefore, any learning process that goes through the verbal cortex will be massively inefficient.
Don't use verbal systems to teach or persuade.
They stink.
Instead, use the visual cortex, the musical capabilities of the brain, the social capacities. And leave the crappy verbal part out of the learning business.
Of course, that is a tremendously subversive notion. That would perhaps eliminate from many learning systems the advantage accruing to folks with high verbal IQs. Which violates everything that the Cathedral holds dear.
Posts
17 comments:
Do you think it is possible to teach verbal reasoning through non-verbal methods? Or are you just giving up on verbal reasoning entirely (or maybe just for an elite few)?
Skill acquisition is obviously done through these non-verbal channels, and learning skills seems incredibly valuable. Are you proposing also transmitting knowledge? What does non-verbal knowledge look like?
Will,
For instance, neither Math nor programming is verbal. And I want to go one further and say "at all". With math, software, and software process education being my primary education focuses...I've been focused on these topics primarily.
I'm actually willing to concede for the moment that many verbal topics don't have immediately obvious non-verbal paths.
Further...I'm arguing specifically that verbal-ness is mostly the problem, not the solution.
What is knowledge? Knowledge is often explained as being in two categories. procedural knowledge and list-style knowledge. The assertions there are that procedural knowledge is best learned via practice. We can thus eliminate that category. And then list-based knowledge is best learnt (via every direction I've seen) via Music, via Spatial awareness, via smell associations. Trying to do list-based learning verbally fails almost every time.
The other direction that I play into is the constructivist paradigm. You don't trasmit knowledge. You create a situation where their brain builds knowledge. 99.998% of teachers attempt to do this via verbal activity. I prefer Montessori...and the advanced versions.
Can you expand your question?
Feedback has important limits, so it doesn't always dominate. In the case of learning, feedback learning is http://en.wikipedia.org/wiki/Local_search_%28optimization%29 so no matter how ferociously effective it is at finding suitably-nearby solutions, it becomes glacially slow at finding some kinds of suitably-distant solutions. (This is fundamental: notice famous NP-complete problems like "traveling salesman" and "satisfiability" on that Wikipedia page.)
I've learned several subjects largely through self-study. (E.g., enough computer programming to redesign parts of an optimizing compiler, enough electronics to build one-off gadgets of a few dozen components, and the game of Go to a 4 dan rating.) It's routine for a "verbal" shortcut to skip me over what would, without the hint, be an impractically long period of practice and discovery. (I scare-quoted "verbal" because you seem to mean general symbolic abstract textbook-style information, esp. including illustrations.)
My self-study examples are pretty abstract, but the same thing happens in less abstract fields, too. E.g., verbal information about the transition from rolling to skidding is helpful when learning to control a car on a slick road. At a higher price point, it's reportedly helpful to get verbal instruction about conserving one's aircraft's energy in air combat. (Roughly: energy gives one options, and once bled off can only be restored slowly.) Otherwise it's easy for people to make the same kind of mistakes in practice dozens or hundreds of times without noticing the problem.
You wrote "leave the crappy verbal part out of the learning business." I grant that one can teach more effectively in many cases by avoiding relying on symbolic words and illustrations. E.g., students of driving will probably "get it" better if you can demonstrate transitions from rolling to sliding, perhaps with a rubber block on a board which is tilted until the block suddenly starts to skid. But many useful concepts can be hard to convey without relying on the "verbal part." It might take quite some time to communicate nonverbally how it is the sum of kinetic plus potential energy that is conserved, or how energy constrains your options.
Purely symbolic knowledge can be clearly second-best when alternatives are available: if you are a military commander or a real estate developer, no matter how good you are at reading maps, it is probably a really good idea to actually visit and look at the terrain. But that doesn't make symbolic knowledge crappy. Being good at reading maps is a powerful skill, enough to make an army that's good at it significantly more effective than illiterate opponents. Even being kinda bad at reading maps can be powerful: having a map and mediocre map-reading skills can be much better than going in blind. Language is a powerful skill, even mediocre symbolic hints about hard-to-notice insights can be much better than having to find those insights unaided, and various useful insights resist expression in nonsymbolic form. (Mendelian genetics, the practicalities of sterile technique for food storage and for medicine, evolution by natural selection, compound interest, chemistry, the thermodynamic constraints on efficient heat engines, celestial navigation...)
(You also wrote "that would perhaps eliminate from many learning systems the advantage accruing to folks with high verbal IQs. Which violates everything that the Cathedral holds dear." But Blogger claims that my comment is long enough already.:-)
WIthout actually responding, Mr. Newman, I am actively calling out the distinction between verbal and symbolic. Math is symbolic. But you can do math non-verbally. Maps are symbolic, but GPS Navigation instructions usually suck in comparison to the map. Words are NOT the right solution for the problem of location. They're a crappy 3rd or 4th best. Neither do you use much words in your Go capacities (I was never better than ~8 kyu -- 40 stone losses to 4kyu players). Sure you want to be able to reference atari, but doing your Go processing in words is almost pure loss.
My assertion: Use the other brain capacities. For location, use maps, not words.
And...I don't dispute that words have been a nudge in cases where you didn't have anything else. But the other stuff would have been a lot better in almost all cases.
If anything, I consider this to be my primary contribution ot the blogosphere these last 5 years.
True, and it changed how I view many problems.
I was reading another pedantic anti-Cathedral libertarian programmer and his take on verbal learning is surprisingly different:
http://takimag.com/article/anti_amnesty_talking_points_john_derbyshire/print#axzz2KlXEDJGQ
Words are NOT the right solution for the problem of location.
If I'm going to be hanging around a city, one of the first things I do is memorize a list or two of the street names in areas I'll be spending time. To go somewhere, I imagine a basic map (because my ability to memorize detailed maps is ~nonexistent) and what big roads get me to the right area, then orient myself by the memorized list of streets in a zone. Memorizing a list of streets in one direction and reciting it the same way every time makes for brain-friendly work.
Meh,
Thanks for the bit about feedback. I hadn't been sure why you read my blog. That's good to know.
This is the most text-oriented blog I can think of ... which makes this post pretty funny.
Meh,
I hate it when I find that my skills (talking conceptually in front of an audience), and what I care about (education) don't work well together.
It's true that words are a poor fit to many things which want pictures. Explaining how to knit or how two DNA strands fit together without using demonstrations or illustrations would be an interesting exercise in masochism. But words are natural for many important things that are hard to communicate otherwise, e.g. causality, possibilities and risk and reliability and honesty, numerousness (?) and commonness and rarity, cost and benefit, physiological and emotional state, complexity and simplicity, and growth and success and failure and injury and death. These things are broadly important.
(Humans are also quite good at thinking about those things. Remember your earlier argument that human brains are evolved primarily to deal with (social) people problems? I disagree about the evolutionary pressure: I think toolmaking and predation and war and agriculture were even more important than in-group competition. But regardless of why we are good at those things, we are good at them, and it's useful to be able to communicate our insights about them, and we mostly do that with words.)
You write "words have been a nudge in cases where you didn't have anything else. But the other stuff would have been a lot better in almost all cases." In my experience the proportions are reversed: words are appropriate in most cases. Not when teaching how to drive nails with a hammer: if you use lots of words then, you're likely doing it wrong, maybe so wrong that you'd be better off taping your damned mouth shut and just showing how it's done. But it is difficult to avoid words when teaching safe operation of a chainsaw or safe handling of firearms, or when teaching rockclimbing belaying technique or effective cafeteria sanitation or army encampment sanitation. And I expect for many students it would be inefficient to teach effective marksmanship without ever verbally expressing the concept of "[avoid] flinch[ing]". Of course a student shouldn't try to learn belaying technique or marksmanship from a lecture alone, but a student also shouldn't try to learn from a deafmute mime with a principled objection to sign language.
If you were 8 kyu in Go, you probably had a pretty solid understanding of how a group of stones needs two eyes to live. Essentially all players learn that principle from verbal explanation. Once you use the principle enough, it becomes second nature and you don't need to slow yourself down by reasoning about it verbally or even explicitly. But that the knowledge eventually becomes fluently nonverbal doesn't mean that its verbal transmission isn't important.
Teaching seems to me a horses for courses thing. Concrete examples bog down for "sterile" or "two [true] eyes [in Go]": they should be accompanied by a verbal distillation of the intended abstraction. Diagrams and demonstrations are inefficient for some things like states of knowledge. Words are inefficient for some things like shapes. When I read something like "don't use verbal systems to teach" it seems mistaken in the same way as "don't use Clydesdales" or "don't use steam turbines" or "don't use direct current".
As an alternative to a rule of avoiding words, I nominate "avoid unnecessary abstraction when teaching" and "when teaching things which are necessarily abstract, strive to give relatively concrete examples which illustrate the principle" --- ideally with feedback, noticing how each particular student seems to be misunderstanding and choosing new examples which illustrate the particular misunderstanding. When it is particularly practical to avoid words, as with hammering nails, my rule calls for behavior similar to "avoid words." But my rule degrades more gracefully in the all-too-common cases of teaching someone marksmanship or GPS navigation or Go.
Aretae,
I hate it when I find that my skills (talking conceptually in front of an audience), and what I care about (education) don't work well together.
You've stated that your skills are ideally suited to motivating students and that motivation is the biggest key to learning. Are you questioning either of these premises? Have you been reading Piotr Wozniak recently or something?
Stepping back a moment, I think that "99% of all first attempts fail, more if it is a difficult (read interesting/useful) problem"
has a greater claim on being your (Aretae's) 2nd great idea.
I've personally found this very useful.
Meh,
Neither of those premises. But that's not what I WANT to do in education. I think, for instance, that in person I have superb explanatory skills. And explaining hard things is what I want to do. Unfortunately it's my enthusiasm that makes most of the difference. As a friend calls it: Christmas tree teacher theory: "Look at me, I'm shiny, so this tree must be pretty"
Dr. Pat,
Thank you for the kind words. I think of that as a connected idea/prequel to my feedback loops theory.
I'm also inclined to believe that I acquired my Militant Humility from the old greats and the GMU Libertarian Mafia. I don't think of it as my idea, but rather one I'm simply preaching. Hume and Hayek (and Aumann). Then Cowen and Hanson and Kling and McArdle. And I see the truth in their position. I just carry the message forward.
Feedback systems as universally dominant, contrarily was not said anywhere that I'd seen. There were hints of it, but not collected together sufficiently. Calling out Agile software and Evolution as being the same mechanism... the Yudkowskian Cthuloid God.
At the same time, I don't hear anyone else as aggressive on this point as I am...so you might have a fair point.
And, I do frame the feedback systems this way:
Feedback systems dominate.
Why?
Because we're wrong a lot.
Regardless all that, thank you for the call out. Folks finding my ideas useful, even those I've just promoted unoriginally, is why I blog.
Mr. Newman,
(you're the second William who posted in this thread, and I don't know you in person)
If you've been 4th Dan level go and you have a European name, you're well into the 3rd or better Sigma. When reading your writing and coupling that with your systematizing skills, that puts you in the 1/1000+ range on both verbal and math/spatial IQ. The point that I'd make is that your internal observations of your state are not a reliable gauge of how anyone else learns. You're so different that you don't count.
The interesting point though is that folks who have trouble seeing why most poodles can't pass the Mensa entrance test (it's so easy) don't gauge well how normal-ish folks learn.
I'm also playing in the high sigma space both verbally and spatially too, but with 40K+ contact hours in education...and education as my primary avocation.
My primary assertion is that introspection might tell you a little bit about a highly non-representative sample learns...but from that you will learn nearly nothing about how someone as stupid as an average Harvard student will learn.
This insight was basically...every time I have ever seen a deity-level hack in the education space, it has been entirely non-verbal. I've seen them in a large number of topics...and every time I rewrite a course to use less words and more carefully defined sensory/ experiential learning it gets better. Not, mind you, that that's easy...it's much harder to do as a teacher. But...it's universal. If something is teachable non-verbally, it's better taught non-verbally. Indeed, in most of those cases in normal people, verbalness is the problem, not the solution.
I'm reminded of a conversation I had with a guy who was COMPLETELY verbal.
He even claimed that he did not dream in experiences or pictures, he dreamed in verbal narratives.
As one friend of mine put it "He must have really boring sexual fantasies."
I second most people here. Yours was the first blog to emphasize 'feedback' in my reading list and now when I try to learn new (programming) languages, I immediately just try as many things out as possible to get the maximum feedback in the shortest amount of time. BTW can I ask you for your favorite resource on how to go about becoming a better programmer/learning new languages etc from your perspective? I would even pay for a short but dense ebook. I'm sure you could promote it on Hacker News.
Post a Comment