This is the question asked by John Cook on Twitter. He lists responses from different people:

  • R
  • Version control
  • Linear algebra
  • Advanced math
  • Bayesian statistics
  • Category theory
  • Foreign languages
  • How to not waste time
  • Women

Mine are: quantum mechanics, Python, cooking, the language of philosophy.

What learning curve do you wish you'd climbed sooner? Give reasons and stories if you feel like it. Do you think other people should climb the same curves?

New Comment
41 comments, sorted by Click to highlight new comments since: Today at 10:39 PM

Linux/Unix & their associated command line stuff. The number of times it's come in handy to be able to SSH into a machine that's way over there and do stuff is immense. Sadly I waited till Uni to learn, and I wonder where I'd be now if I'd internalised these concepts by the age of 15.

Need to log something you've just done? Redirect the output into a file. Boom. No longer do you have to find the bit of software that does everything, you just need programs that do simple stuff you can repurpose*.

Reading FOLDOC a bit to get the history was handy too. Thinking of computers as CPUs attached to teletypewriters, with all that fancy graphics stuff as optional explains a lot of how current software ended up the way it is.

*Totally not advocating mainstream computing for every user end up like this, btw. (Linux still makes me cry on a regular basis) Just that being able to drop to a command line and chain together commands or write scripts is so powerful that it's a game changer.

Majorly with you on this one. Piping in shell changed my life, and then piping in R (via Magrittr via dplyr) changed my life again.

Vicarious regrets: Negotiating salary. Basic investing. Budgeting well below ones means. Sampling many subjects and careers to see how they fit.

Personal regrets: Ability to make good impressions on non-nerds (e.g. dressing well). Meeting lots of people.

Somatics. Interacting with your own body in a decent manner brings all sorts of advantages. If I would go back 20 years and meet my younger self that's the most important thing.

Formulating knowledge for SRS like Anki.

Social skills.

Somatics. Interacting with your own body in a decent manner brings all sorts of advantages. If I would go back 20 years and meet my younger self that's the most important thing.

I'm still young and can learn. Some ressources to point me at?

If you want a book I personally liked Thomas Hanna's Somatics Reawakening The Mind's Control Of Movement, Flexibility, And Health. It can be found as a PDF on the internet.

The books starts by explaining why you should care about somatics.

He might exaggerate some things but I think that no matter what you do in terms of the factors of aging that Aubrey de Grey describes, if you don't address the somatic factors you won't make it to 300 years of age even if you do gene therapy.

At the end the book describes exercises that you can do at home before going to sleep and after waking up. I'm not sure whether those exercises as described by Hanna are optimal but they do something. There a higher time investment to get into it but after that the regular upkeep is at 5 minutes per day after waking up.

Finding local classes is also good. At the moment I still lack the knowledge to judge the quality of all methods. Doing Feldenkrais with a teacher that has years of experience would be a straightforward route.

Would you mind explaining why you think Somatics works? That >=3 people on LW take it seriously certainly lends some credibility, but as far as I can see there's not much else to recommend it. I'm genuinely curious what convinced you, since this seems potentially useful.

Looking at the Wikipedia pages on Somatics and on sensory-motor amnesia, I see no significant evidence that this actually works.

Nor does their official website provide any convincing evidence.

Edit: This is a review of 6 Feldenkrais RCTs, and concludes:

[The six studies] were all burdened with significant methodological weaknesses. ... All but one trial reported positive results. Conclusion: The evidence for the FM is encouraging but, due to the paucity and low quality of studies, by no means compelling.

Would you mind explaining why you think Somatics works?

I hold my beliefs on the subject mainly because of things I have seen with my own eyes and experiences I made. I spent years in standard physiotherapy. As part of my QS regiment I measured my lung function daily. After a 20 minute intervention at a bar by a practitioner of somatic psycho-education I woke up the next day with an 50% increase in my lung function. At that time that wasn't permanent but after 4 months of weekly session I stabilized on that new value. Specific improvements in that timeframe often happen right the day after a session.

Hard QS numbers where what hooked me initially. Then as time went on things got more complicated. I started perceiving a lot more. I made a lot of experiences but they don't fit into a structure. Today I could spent 5 minutes to describe you how my body feels and I don't really have a way to compress that into a decent number that I could track. This means that studying the subject via the scientific method is very hard.

There are a bunch of things I perceive which I can't measure with any method I could think of and where I'm not calibrated and therefore I can't tell whether the variable actually changes or my perception of the variable changes.

For example at the start of this week my physical perception increased. Practically that means that I can dance a Salsa turn pattern comfortably at my Salsa lessons with eyes closed. At the same time my increased perception comes with slight nausea I can perceive and I get diarrhea. But overall I feel more comfortable.

If I would use some random checklist than I would mark nausea as something negative. But I do thing the problem was there before and it's just my increased awareness that brought it to the forefront.

If you study something like back pain, we don't have good objective methods to measure back pain. The standard is to ask people and when your primary modus operandi is about using changing in perception as a means to get results that's problematic.

We have X-rays to diagnose the shape of bones but we don't have similar high powered equipment to measure the amount of contraction of all the muscles in the body and have standardized that equipment to be useful for diagnosis of illnesses that have to do with contraction of muscles that don't relax themselves when they should relax themselves.

A Feldenkrais practitioner might say: "Hey you have this muscle in your back and is tense. If that muscle would be relaxed your back would feel better. I can teach you to relax it." That seems like the kind of question that should be easy to investigate scientifically. You would just need an objective measurement for whether the muscle is tense. Unfortunately we don't have this for clinical practice. Scientific discovery has a lot do to with having good tools.

I have a friend who writes a software that uses the kinect for analysing movement patterns to study another physical therapy. It's hard to get academic funding for it. In biology there a lot of money for buying fancy fMRI's and gene sequencing equipment. There not much money for buying good camera's and software to analyse movement patterns of humans so she has to repurpose closed source gaming hardware. There's nobody spending the millions to get an open source solution that's optimised for the needs of researchers of human movement. Such tools might be a better target than telomere research for someone who wants to found the fight against aging.

I don't capitalize the term somatics because it's a field that contains many methods. I do think that the book of Thomas Hanna is good but I don't specifically endorse Hanna Somatic Education over another brand like Feldenkrais. Thomas Hanna unfortunately died in a car crash so his trademarked method isn't as strong as it would have been if he would have spent another two decades in research.

Somatics

30 seconds of research leads me to believe it's quackery. Should I investigate further?

It's completely an issue about how you form your beliefs. I didn't form mine on the subject matter through scholarship but through empiricism.

There limited peer reviewed research on Feldenkrais but those studies rather indicates that it works than indicating that it doesn't work.

If you need peer reviewed studies to believe that investing more energy is probably not having a good return. On the other hand from the initial list of this article, I don't use Source Control Software either because the software was shown to be effective in peer reviewed experiments.

I second the recommendation of Somatics-- it's a good simple explanation of Feldenkrais Method, and the exercises are good for my lower back.

I think there's a rejuvenating effect because Feldenkrais reverses a lot of accumulated movement habits, but it's plausible to me that gene therapy might have the same effect. On the other hand, we have Feldenkrais and we don't have anything like that level of gene therapy.

I like Somatics, but think that it's a very thin skim on what Feldenkrais had to offer.

Feldenkrais have very interesting stuff on volition and action in The Potent Self. Body and Mature Behavior is good too.

As far as aging goes, Aubrey de Grey lists aging as being about a bunch of factors. He proposes if we fix those factors through techniques like gene therapy we can get 1000 years old.

Aubrey de Grey's list misses "bad movement habits" or as Thomas Hanna calls it Sensory-Motor Amnesia (SMA). I think it's important for the discourse about fighting aging to understand that Grey's list isn't complete.

One thing I'd add to the list of Aging factors is a generalization on the unnecessary tension and lack of sensation noted by somatic practitioners - feedback loops and signal transduction pathways pegged into insensitive operating points.

Some signal gets too large, which tamps some sensitivity down, when then leads to positive feedback making the original signal even larger. Hormones, neurotransmitters, muscle actuation/sensing. System compensation helps in the short run, but they lead to getting trapped at suboptimal operating points that are local minima, that require some "kick" to get you out.

With the latest post on Neoreactionaries, I felt the urge for some Moldbuggery.

He had an example of the human organizational version of the principle described above, which I realize now is largely his whole Cathedral analysis as well. When the feedback loops get screwed, so do you.

Nearly every scientist in a field can be working together to promote a falsehood because they all get their money from Joe Romm and company. And if the falsehood is exposed rather than promoted, there is no field left. It is no more surprising that all USG-funded scientists are unanimous in promoting AGW as a global emergency, than that all Philip Morris-funded scientists are unanimous in promoting tobacco as a vitamin.

This question is kind of weird, in the sense that what I really wish is that I had climbed some shallow (big wins with not too much effort) learning curves earlier. In lots of domains, an 80/20 rule applies, where you get 80% of the benefit with the first 20% of the effort expended.

I'm curious about the domains where you get shallow wins.

exercise, food, meditation/planning/metacognition training, some positive psychology tricks, social interaction, clothing and personal presentation, playing an instrument. There are probably lots of others.

Playing an instrument

Really?

strongly disagree with "learning an instrument". I wasted lots of time on that. The costs are high, and all the benefits are either dubious or can be more cheaply obtained through other means.

At least for classical instruments, getting good enough to not embarrass yourself is quite expensive, both monetarily (equipment, teachers) and in terms of time. At >1k hours of practice and ~$10k invested, I'm just barely good enough to get paying gigs (which I'd lose in a second if the people paying me knew anyone who'd attended conservatory).

How is playing an instrument a big win, and how is it not much effort? (I'm assuming it involves not learning a classical instrument.)

Playing the piano or guitar, even at a semi-beginner level, can be a lot of fun. You won't make money off of it, but you can show off to friends and family members. It's like playing tennis or basketball: you need a high level of skill to compete seriously, but you don't need to be even as good as the local high school team to have fun with your friends.

Incidentally, I think it might be easier to achieve minimal competence on a guitar than on many other instruments - you can perform a lot of songs just by learning a few chords and singing the melody.

Learning to play an instrument is probably not something most people can get an 80/20 sort of benefit from, but it belongs in a class of activities in which some/many people can put minor effort in to reap the significant benefit of becoming a more interesting person, depending on one's innate proclivities. Other examples may include dancing, singing, drawing/painting, certain sports/physical activities, craft-work, etc.

Off the top of my head: fashion, public speaking, and knowledge of the tax code.

Programming. Writing essays. Learning an instrument.

I have to +1 'Writing essays'.

Too easily overlooked in a technical environment, but it really pays off. Both to assist in consolidating your own ideas, and also in communicating them to others.

Agreed. Also: request for canonical sources on the art of essay-writing. It's something I could use improvement on, and I don't see anything related in the best-textbooks thread.

Strongly agree with "programming", but strongly disagree with "learning an instrument". I wasted lots of time on that. The costs are high, and all the benefits are either dubious or can be more cheaply obtained through other means.

I, on the other hand love my cello. I also happen to enjoy practice itself. This helps a lot.

Spotting and establishing connections with Gervais's Sociopaths. If I would have done this in university I would have a more useful network.

Doing CBT - not that may not be aplicable to people with better mental health

Steep?

String Theory and related math. Math is not my strongest suit, so taking the same courses a decade earlier would have been much easier and more productive.

Some musical education. I have no relative pitch to speak of (or at least a gross mismatch between identification and reproduction), but early training would have at least enabled me to carry a tune without being horribly off key 100% of the time. It sucks that I can't even sing to myself without cringing.

Hm. I thought a bit. Looked through the list. Thougt a bit more and concluded: I'm quite happy with my learning curve,

Only after looking thru the comments did I consider Investing. Advanced Investing actually. That would have been handy not only because of the money but also for more interesting job opportunities. Problem is only that I'm not clear how I could have had a path to that knowledge.

I'm quite happy that women and style knowledge are placed so late in my life. The mastering of these can eat insubordinate amounts of time and effort and cannibalize other areas.

I', very happy that math and the sciences came first (and deep) and have made learning all the other areas much easier.

See also http://c2.com/cgi/wiki?BreadthFirstLearning an old wiki page about how people acquire knowledge.

The hows and whys of refactoring and DRY. How and why to achieve proper separation of concerns (I still have lots to learn there).

Social protocols, especially around initiating and maintaining friendships and other levels of relationships. Being empathetic.

The hows and whys of refactoring and DRY. How and why to achieve proper separation of concerns (I still have lots to learn there).

Do you know what you have to learn in that area?

Statistics. I took it in High school, but it was so poorly taught that I learned almost nothing from it. Now I use statistics every day.

Scripting Languages. I learned Java in High School and rode that knowledge all the way through my CS degree. But when I got a job as a software engineer, Java was among the worst languages available for solving the type of problems with which I was dealing. Looking back at my old code, I could have saved hundreds of hours if I had learned Python instead of Java.

Related, curves I haven't climbed but wish I could/would/intended to:

A European foreign language. I dabbled in a variety of languages in school and settled on Chinese, wasting many years and having little to show for it. If it had been an Alphabetic language (and better yet, a Latin-alphabet language), I'd have a much higher level of proficiency.

A Martial Art. I love the martial arts, but I've never been able to devote myself to any one of them.

Seconding statistics.

I'm tempted to say "social interaction", but I'm not entirely sure it would be honest. I wish I knew it now, but at the time I wish I had learned it, the measure of my interest in people was an imaginary number.

Playing an Instrument for sure (I'm a classically trained Flautist) Photoshop/adobe programs Business Myself and my own core distortions Writing (I started that journey when I was 12 and am just coming into my own) Design

I should probably get better at Math but I don't really need it and I don't code or program and I'm sure there are people that think I should do that.

Most things that I have taken on I have put in effort and gotten a result as well as manifesting the right people to help me.

I do not like "steep learning curve" the way people use it. It raises my probability estimate that person using it has done no study of learning. This reduces my probability estimate that they have sound insight about learning.

Sometimes I work with learning curves for machine learning or human learning. These curves are plots of a measure of learning (say, correct score, or number of mistakes) versus time or number of trials or number of training examples or number of iterations. When topic is hard, curve is shallow! It takes more learning to improve score or reduce mistakes. Steep learning curve means topic is easy. Mastery comes with very few repeats or little time or little data.

Does anyone know why so many use it wrong, to mean topic is hard and progress is slow? I believe Wikipedia article is right and "learning curves" must come from scientific study. How did meaning get reversed? Wikipedia article says "Arguably, the common English use is due to metaphorical interpretation of the curve as a hill to climb." but has no citation.

The first post didn't use it only to mean progress is slow. Version control software is no subject that takes much time to learn. The problem is that if you want to use it for the first time, it takes time to wrap your head around it before you can use it productively.

I don't think the progress on learning R is much slower than the progress on learning Excel but learning R is harder. You can use Excel while have relatively little skill at using Excel.

Climbing a steep mountain takes more effort than climbing a mountain that isn't step. It doesn't necessarily take more time.

That's what I understand by a "learning curve" too. I would tend to agree with the wikipedia proposed explanation: "steep" is, in some contexts, synonymous with "difficult". Saying "you have to climb a shallow learning curve" would certainly be interpreted wrongly by most people.

Concerning the OP question, I wish I had learnt something interesting at school, instead of the thoroughly irrelevant, utterly boring, mindboggingly wrong pseudoknowledge.

To zedzed and D_Malik: I think what you have in mind is learning to play an instrument as a profession. I that case I agree that the rewards are probably not commensurate to the effort, unless you are exceedingly talented and lucky. But for everybody else it's probably a good idea. If done for its entertainment value it's surely worth it, and it helps you understand and enjoy music at a deeper level. On the minus side, it also makes you realize that most popular music sucks. Really really sucks, to the point of irritating me.

I understand the "steep learning curve" to be a curve on the plot where the X axis is the measure of progress, how far into the subject did you make it, and the Y axis is the measure of difficulty.