Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

The Feynman Technique: The Best Way to Learn Anything (2012)

fs.blog

Post reply on HN
Print view View original

41–49 of 49 posts

Re: The Feynman Technique: The Best Way to Learn Anything (2012)

#41

Post by teacpde »

judofyr wrote: post #12
barefoot wrote: post #7

"If you’re not learning you’re standing still" This is a large understatement. In our industry, if you're not learning you are drifting backwards rapidly. I aim for twenty hours a week of (unpaid) reinvestment and learning. Some weeks I still feel like that's not enough to stay current. Does anyone else feel the same way?

I’ve had that feeling, and I started making an effort to improve myself. I started time-tracking my side projects (“you can’t improve what you don’t measure”), scheduling time for learning, and set up an automatic summary sent on email every week. It didn’t work for me at all, and ended up basically killing every joy in my side projects. If I missed a scheduled hour (because I was doing something else fun) I felt lik…

I share similar experience, I find I can only achieve learning at a high effective rate with strong short-term motivation, and it can become tiring very quickly. On the other hand, I can always enjoy having fun and learn without worrying about effectiveness.

Re: The Feynman Technique: The Best Way to Learn Anything (2012)

#42

Post by naasking »

barefoot wrote: post #7

"If you’re not learning you’re standing still" This is a large understatement. In our industry, if you're not learning you are drifting backwards rapidly. I aim for twenty hours a week of (unpaid) reinvestment and learning. Some weeks I still feel like that's not enough to stay current. Does anyone else feel the same way?

> Some weeks I still feel like that's not enough to stay current.

Staying current is overrated. You'd be better off spending most of those hours on CS fundamentals, written and oral communication skills, and probably exercise.

Re: The Feynman Technique: The Best Way to Learn Anything (2012)

#43

Post by james_s_tayler »

wenc wrote: post #18
dajonker wrote: post #6

That's usually how you can identify good people: when they can explain complicated things using simple words, in a way that their audience can understand it. People that need lots of expensive words and leave you feeling dizzy at the end of their presentation often don't know very well what they are talking about either.

An alternative explanation is that some people are just good explainers. I've had profs in college who were brilliant but couldn't teach, and profs who could teach but weren't brilliant (didn't produce much original research). I wouldn't say the former group "didn't know their material well", but they didn't build up the material systematically. Pedagogy was neglected. Explaining things is an art that not every has t…

Professors can massively improve their pedagogy if motivated to do so.

I once had a professor who was new to teaching. It was a very painful experience and the entire CS class spent every lecture murmuring among themselves how terrible the lecture was. I gave him the benefit of the doubt and waited to 2 weeks to see if the lectures would improve. They didn't. At the end of one lecture I went and spoke to him and told him directly he was by far the worst teacher I had encountered at University and that as students we expected higher quality instruction.

From the next week on he was a good lecturer. It was a total transformation. I was seriously impressed. I guess he just never really thought about the pedagogical side of things before but to get such a rude wake up call really jolted him into action and he actually had a pretty good capacity to teach after that.

Re: The Feynman Technique: The Best Way to Learn Anything (2012)

#44

Post by james_s_tayler »

peter303 wrote: post #26

A variant is to 'explain it to your spouse'. (Ignoring these days your partner might be better educated than you.)

I've started just explaining it to myself. You don't need to actually teach it to someone.

In fact this is kind of a subtle point. There are two functions that teaching it to a third party is performing. The first is that of burning the material into your memory and the second is getting to an exposition that the third party understands.

Well... the second part is not necessarily necessary. It depends on your goals, but let's say you are only trying to further your own understanding then perhaps all that is really required is getting the material into your memory in a robust way. Just explaining it out yourself enough times to actually wire all the nuerons up in a complete graph is massively beneficial.

Re: The Feynman Technique: The Best Way to Learn Anything (2012)

#45

Post by 8bitsrule »

"Step 1: Teach it to a child..."

Key. Teaching is a learning process for both teacher (who must condense and reduce to essentials, and so integrate) and student.

If the child is old enough to ask good questions (and is encouraged to), so much the better. The feedback is valuable, and the teacher can continue to reduce/integrate at a somewhat higher level.

Re: The Feynman Technique: The Best Way to Learn Anything (2012)

#46

Post by tugberkk »

kmarc wrote: post #25
tugberkk wrote: post #8

There is a youtuber junior md called Ali Abdaal. He has very good scientific videos about learning. Talks about subjects such as “active recall” and “spaced repetition”. Worth a look imho. Here is the youtube channel link: https://m.youtube.com/user/Sepharoth64 Video on a.recall and s.repetition: https://m.youtube.com/watch?v=ukLnPbIffxE

Should you be interested in the science and research behind it, I warmly recommend reading the book "Make It Stick" where the concepts are discussed in length. [1] [1]: https://www.goodreads.com/book/show/18770267-make-it-stick

Thank you!

Re: The Feynman Technique: The Best Way to Learn Anything (2012)

#47

Post by MaxBarraclough »

tejohnso wrote: post #5
bitwize wrote: post #3

I thought the Feynman technique was: 1) Write down the problem 2) Think real hard 3) Write down the solution In other words, the "draw the rest of the fucking owl" of general learning/problem solving.

No, it's 1) Write down the problem 2) Frame it in such a way as to pique Feynman's curiosity 3) Wait for Feynman to tell you how he solved it

These days we call this approach 'lazyweb'

Re: The Feynman Technique: The Best Way to Learn Anything (2012)

#48

Post by MaxBarraclough »

naasking wrote: post #42
barefoot wrote: post #7

"If you’re not learning you’re standing still" This is a large understatement. In our industry, if you're not learning you are drifting backwards rapidly. I aim for twenty hours a week of (unpaid) reinvestment and learning. Some weeks I still feel like that's not enough to stay current. Does anyone else feel the same way?

> Some weeks I still feel like that's not enough to stay current. Staying current is overrated. You'd be better off spending most of those hours on CS fundamentals, written and oral communication skills, and probably exercise.

Doesn't that depend on your work?

Many web developers have no choice but to chase fads. The whole industry is based on fad-chasing, after all.

Re: The Feynman Technique: The Best Way to Learn Anything (2012)

#49

Post by MaxBarraclough »

8bitsrule wrote: post #45

"Step 1: Teach it to a child..." Key. Teaching is a learning process for both teacher (who must condense and reduce to essentials, and so integrate) and student. If the child is old enough to ask good questions (and is encouraged to), so much the better. The feedback is valuable, and the teacher can continue to reduce/integrate at a somewhat higher level.

I'm always surprised how rarely anyone makes the obvious point here, and sure enough in this thread it looks like no-one has: no, you can't explain complex things to a two year old. Not without dumbing it down to the point that you've discarded everything of value.

Is this some odd kind of hyperbole about teaching techniques? I don't see that it's helpful.

If I'm struggling with it, the odds of a toddler being able to comprehend it are slim to none.

The kernel trick? Quantum computing? Even good old eigenvectors? (Or, heaven forbid, genuinely advanced mathematics?)

The best professors in the world couldn't convey these things to a 2 year old, and not for lack of understanding.

Post reply on HN