Old, and imho not very good, models of programming ability were pretty strongly informed by the cognitive science backgrounds of the few people who worked on them. E.g., the argument that a strong working memory and trained better 'chunking' might lead to large indiv diffs in programming
@grimalkina I remember taking a test for CS ability and totally flunking it.
The main reason I failed was that for many of the multiple-choice questions I could make valid arguments for more than one choice - so had to try and guess what the question-setter was trying to get me to pick.
It was deeply frustrating as there was nobody to ask clarifying questions.
Turns out that the ability to find logical flaws is rather useful for programming! Also, asking clarifying questions is critical!
This is laughably simplistic when it comes to modern programming, and there are approximately a million other things we can also care about when it comes to people learning and working.
So many myths about the skills of programming derive from this obsession with (human) memory
It's pretty interesting to think about how these attributes may have mattered because of the constraints of the day, but the ways we've carried them forward in stereotypes is just, remarkable
I keep trying to wrap my mind around it because it's such a chain of claims that are slightly warped referencing other claims that weren't actually the thing that person said....loads of noise. Very frustrating.
Another thing I notice is that things that are probably generally true if you selected ANY cognitive task, for any group of people, are taken to represent programming ability in a unique way just because a study was done solely with programmers.
So few questions about what might be a general trend across pretty much all forms of complex work (e.g., working memory could also matter for managing a restaurant), and assumption that finding in a population of programmers = "unique to programming"
@grimalkina I agree and I can also relate on two counts. Firstly the general cognitive skills. I've seen a claim (don't remember where) that good programmers are often also good writers. Maybe also vice versa. I strongly suspect that is not a coincidence and that there is a large overlap in the skills needed for programming and the skills needed for writing. I also teach a programming course for beginners and noticed that most course participants are able to learn to program. While all participants are academics, they are usually in the humanities and not all of them are intrinsically motivated. This has given me the impression that programming is not nearly as exclusive as some people like to portray it.
Secondly the working memory. Mine is not exceptionally good. In fact, I have trouble remembering a shopping list with three items. I am gifted but I also have ADHD. According to an IQ test, my working memory was above average but not nearly as much as my verbal and spatial intelligence. Yet my programming skills tend to be regarded as quite good by people who have seen me do it (I am a tech lead in a professional team of nine, I maintain a few well known open source libraries and I am also somewhat active in online programming forums, so my work is relatively often seen by others). I don't feel like I need to rely much on my working memory while programming, because I can constantly see the code I'm writing laid out in front of me. The claim that working memory is specifically important for programming, which I see for the first time in your thread, does not strike me as very convincing for all these reasons.
To add to the point on working memory, I have a teammate who is even more forgetful than I am, while he is also famous within our team for always being able to get code to work. I hope you will forgive me for mentioning you here @sheeanspoel, maybe you can share your thoughts on this.
@juliangonggrijp thank you for sharing, I love this reflection and all your points. My general pov on this is that diverse people can reach the same levels of success by learning to capitalize on different strengths and also that because we need complex outcomes we need to learn to value many types of minds and their problem-solving. Where we can support, supplement, and understand so that people reach their full potential, why wouldn't we?!
Early work on this deeply underestimates the need...
@grimalkina I think some skills are or at least used to be like 'gates' where if you do not master it (syntax) it prevents you from going further, yet if you get past it, being awesome at it does not do you any good. Yet other skills like weaving abstractions and basically coming up with new languages to describe solutions to a problem only come up later and make all the difference once you are past all the boring stuff... Which means models for students and masters are wildly different.
@ignas I think that's a really interesting point. And an important one to not think of these as nothing but linear relationships. As you say maybe things are just about passing a threshold, and even there, we might be able to compensate with other strategies and competencies
@grimalkina
Yeah, hard to believe that 'traits X, Y, and Z are why people are better at programming' when X,Y, and Z all mean "better". Like maybe: "Smelly are good at programming because it's less social than other tasks."
Personally, I think a lot of it has to do with adventurous experimentation, which is often seen as a positive in boys and a negative in girls..
@ThreeSigma @grimalkina I'd say there's also something about having, for lack of a better term, "reflexive abstraction focus". The innate drive to drill-down on the essentials of an orchestration and see/feel it from a variety of perspectives and abstraction levels on an unconscious level.
@grimalkina I think working memory affects what type of software development or practices come more natural to you. Also, mental rigidity, hyperfocus tendencies, visual pattern matching, and even learned behaviours like prioritizing effectively affect how you solve the problems.
Are these related to what you were saying or are you only focusing on ambiguities surrounding correlating memory to programming?
@rojun I think the last one you list, effective strategies, is probably what I would call metacognition which is part of the class of things that I believe has been far too overlooked in favor of fixed trait explanations. We have strong evidence that you can help people learn better metacognition and this is one of the more effective ways to improve their learning and problem solving. On the other hand very small effects from working memory over a certain threshold
@rojun so if you care about improving things, care about the things that have more predictive value for your outcome and that are more possible to change. With a large global population you're going to have a distribution of working memory plus over their lifetimes it'll change. It's absurd to act like small diffs which may be detectable in simulated laboratory challenges will predict any large amount of the variance in real world outcomes for this population, and nothing else does.
@grimalkina With small diffs you refer to the working memory and other personal differences?
Even though metacognition sounds like something to be learned, I've noticed some people just seem to fall into it more intuitively. If a person can filter priority more efficiently, they are bound to be more proficient in creating software.
I guess this relates to learning to sing does not guarantee one to land the highest level musical productions, or learning to run faster to break world records.
@rojun yes I mean the effect size of the individual differences on these measures used for this purpose tend to be small (particularly when looking at a population of professionals which has already been selected to be at a certain level of functioning)! And working memory changes across a lifetime and is coopted by many many things (eg anxiety will steal it away from your problem solving, so will needing to solve social problems) so it's a complex factor to talk about
@rojun metacognitive strategies are more teachable though, and frequently win out as the largest intervention effect size in large scale tests across educational interventions (eg, this is old but gets at some main points https://journals.sagepub.com/doi/10.3102/00346543066002099 )
@rojun often things that seem "innate" in some people are in fact the product of learning, and can be learned. Just because they come easily to some doesn't mean they can't be taught to others! Metacognitive skills is one of those things where there might be high individual variance on them but that doesn't make them a bad intervention target, what we tend to assess is whether those people who start low can end up high (rather than simply who is high and low without intervention)
@grimalkina I agree, I started programming about 45 years ago, and then knowledge of and being able to hold all the basics in your head was vital to the task. It's been a long while since I professionally programmed, but new tools obliterate memorization needs.
