Paul
Allison Parrishhey mastodon, throw all your programming pedagogy links at me. I teach a lot of beginner programmers and I want to get better at it!
specifically the trouble I have is getting beginners from the "I understand the syntax and can modify examples" stage to the "I can model problems as programs and apply my knowledge of the language to build these programs" stage. the first stage is pretty easy and most tutorials for beginners are oriented toward that kind of literacy... the second is harder to achieve and harder to teach, and it seems like most people only reach that stage with self-directed practice
Soso Verde@aparrish For the second one you also need a lot of creativity. For some people that's very hard.
William D. Jones@aparrish Hmm, I should do a reflection on "how did I get to this point?" It's a pity I only remember bits and pieces of what I've learned before a specific "catalyst" event.
@aparrish (That event being "building a toy CPU in Verilog for a class". That was the first time I really felt like "I could do anything with computers".)
John 🧀@aparrish
Some resources that support the idea that grokking takes time and self-directed practice:
http://norvig.com/21-days.html
http://abstrusegoose.com/249
Some resources that support the idea that grokking takes time and self-directed practice:
http://norvig.com/21-days.html
http://abstrusegoose.com/249
Nicolas 💬@aparrish Maybe what you might teach for this step is more about methods to cut a "problem" in small chunks and how to draw flow charts and system design?
Like, less on code, but more around the actual thinking that bring one from "I want to do this" to "I'll do it that way".
Focusing the thinking more on "how to explain my problem to the computer" and less on "write code"
I'm not sure there's a way to teach that. If there is I never found it nor saw it used.
MOVED@aparrish @StreetWanderer you're speaking of the distinction between computer programming and software architecture
The traditional way to teach that involves design patterns, which last I checked, have a bad reputation among people who have to work with green CS grads.
@aparrish I've been fighting with a lot of "engineers" over the years about the way programming is taught to beginners.
I even wrote a bad Medium article on the subject.
Programming is taught as maths when it needs to be taught like a foreign language.
Learn to speak and formulate though in the "computer language", the specifics of syntax, vocabulary and grammar can be learned as you go.
Like with a language, your accent and grammar will be off at first but that's part of the learning.
@aparrish the problems at Project Euler (projecteuler.net/archives) seem to mostly exist between the two stages you mention, so it's a good resource for practice material.
tonnydourado@aparrish I think second is actually pretty much orthogonal to the first, and don't derive from it at all. Learning to break up and model problems abstractly can be learned with not programming at all. And perhaps it should? I learned in parallel, but recalling it, the classes in college where almost always independent. I learned basic programming and languages in some (like intro to programming) and modeling in others (like db, software engineering, algorithm analysis)
@tonnydourado that's a really good point! unfortunately the two tasks get conflated in curricula all the time. :/
@aparrish yeah. Lots of discussion about which language to teach first. It is relevant, I think, but before any languages, understanding the computer should come first.
@aparrish the one thing I feel helped me throughout the whole thing is forming a mental model of the computer very early on. There was no programming involved, but the professor explained, in high level, but precisely enough, how CPU and memory interact and how this leads to breaking things into smaller steps. I feel like if you get that, a lot of the apparent magic in even higher level languages becomes less opaque.
⚗️⚗️⚗️ pnathan ⚗️⚗️⚗️@aparrish when I ta'd initial courses, that was the hard jump. tbh my advice to early cause students is "time and loc". experience the craft as you fumble towards comfort.
love to hear of better methods
Darius: Moving Accts, See Bio@aparrish We had a whole class in my electrical engineering undergrad that tried to teach that, but mostly failed at it. I think project-based work, where you give people progressively larger and larger problems to solve, is pretty much the way to do it.
The principle from the class I alluded to still applies: abstraction and synthesis. Students must abstract a problem and then synthesize that into the relevant language concepts
@aparrish One thing that I'd be interested to try is a daily whiteboarding excercise. So students only do the abstraction part and you do it as a group for like the first ten minutes of class every single day.
If you look at it at two skills it's the abstraction that's usually the harder part than the synthesis, so I think just working those abstraction muscles through daily calisthenics is the way to go.
@aparrish So like the input of the exercise would be some problem. The output would be a HYPOTHESIS as to how you MIGHT model it in code. But don't bother digging in and falsifying. Just move on. One hypothesis a day. Maybe at the end of the class review them all and self-critique?
@tinysubversions I've been trying to incorporate material like that. also when I introduce a new bit of syntax, I try to start with the kinds of problems that require the new syntax to solve, and ask students to "design" the relevant syntax for solving those problems. I guess I need to write those exercises down and be a bit more consistent/directed with them
@aparrish Yeah that sounds pretty good!
Ivan Safrin @tinysubversions @aparrish I agree with Darius. I teach a class that is almost purely project based and my model is basically, give them a bunch of building blocks during class, then ask them to put it all together in an assignment, then build on top of that with more complex building blocks.
Parker Higgins@aparrish sorry, this is not an answer, but I want to say that this totally mirrors my experience and I hadn't seen it described before. For a decade I kept re-learning that first stage and it took forever (and a project-based book, and a project I cared about) to make the leap to the second step
Carl Friedrich Bolz-Tereick@cfbolz @xor I teach mostly artists in a self-directed setting (masters degree art program). my experience is that even motivated people with a project they care about will often fail to make the leap (usually finding a solution with a perceived lower effort gradient, e.g., using pre-built code examples with little modification or collaborating with a more experienced peer)
Gregory@aparrish I find that I can read something, and not understand it but when I do something that involves what's being learned and can see an example I learn very well.
æon 
@aparrish I'm not an educator but I'm really interested in this problem
@aparrish the only textbook I know that tries to address the second stage is 'How to Design Programs'. It's Scheme based, however:
@aparrish Agreed. I got to the first stage, but have no chance of ever writing something from scratch. Brain no worky that way.
v@aparrish this is not what you sought but: I am curious to play T Cavanagh's forthcoming work. A (non-coding) pal said past programming tasks just built chaotic knots - but with this, form was distinct, simplifications logical. shows abstraction mapping, i think? idk! but i'm curious.
Ceejbot has a new job@aparrish My current theory, after years of trying to figure this out, is that people generally learn the second through a programming apprenticeship. This is slightly gloomy, I know, and I would love to find out about books that teach this.
Brendan Howell@aparrish This is the toughest. I try to make frustration and confusion more acceptable to students. Lower expectations of fancy results and appreciate that hair-tearing is part of the process. And I do a lot of iterative, code-along-with-me exercises to show how I start with something simple and slowly add to a program rather than pointing to finished examples.
Fritzy 🌯@aparrish Our brains are pattern machines, so that is how we learn at first. Finding patterns, adjusting them, etc. Modeling problems as programs in your head takes time, but I bet it could take less time if we pointed out that that was the goal. Some people like thinking of codes as layers, and that can help, but it's a middle step.
Traditionalist: http://greenteapress.com/wp/think-python-2e/
Iconoclastic: https://mitpress.mit.edu/exploratory (hi @nickmofo )
@aparrish I've used resources from https://software-carpentry.org/ to teach an intro to scientific Python for non-programmers. Their stuff was very useful.
Ꮢ๐ϲoᴄo Ⅿoԁｅm Ᏼaｓіliｓk@aparrish How experimental are you willing to go?
I have some theoretical justifications for teaching beginners to solve simple problems in a variety of esolangs. 😉
@enkiv2 I am willing to be extremely experimental tbh
@aparrish
My thesis is basically: most of *deep* programming knowledge is getting the flexibility to paradigm shift, & the easiest way to get that is exposure to lots of paradigms. Most of the difficulty in learning real languages is memorizing inessential syntax details.
Esolangs are all paradigm and no syntax, so they can be learned more quickly because they make no consessions to reality.
Solving the same problem in brainfuck, befunge, pure prolog, and forth will level you up.
@aparrish
But, a simpler way to do this would be to choose a couple problems and show very different ways of implementing them. (Say, showing where a loop makes more vs less sense than recursion)
But, a simpler way to do this would be to choose a couple problems and show very different ways of implementing them. (Say, showing where a loop makes more vs less sense than recursion)
@LogicalDash @aparrish
I'm not familiar with that at all, & will have to look into it.
It's the logic behind language survey courses and the "seven languages in seven weeks" books. I think those examples compromise too much, choosing marginally useful languages even when utility for real problems is counterproductive.
@enkiv2 that makes sense. I'd love to see a whole curriculum designed around this and similar ideas (or at least some academic literature to help prove to my department that we should adopt something like it 🤔)
@aparrish Another nice experimental technique: teach them about stacks and associative arrays, then have them try to implement a toy forth-like stack language. Then, ask them to write something in it.
Parsing in a toy forthlike is easy (just split by whitespace), & keeping track of parameters and functions is also easy (two stacks). So, it's within the realm of possibility for beginners to implement.
@aparrish
This is basically the SICP method, except with forth instead of scheme: understand the concept of a program by understanding how an interpreter works, by writing one.
This is basically the SICP method, except with forth instead of scheme: understand the concept of a program by understanding how an interpreter works, by writing one.
@aparrish
Depending on exactly how beginner your beginners are, it might be out of scope, but I wouldn't overestimate it -- it's important for programmers to learn not to mystify compilers and interpreters early, and toy-forth is a lower bar than most.
Depending on exactly how beginner your beginners are, it might be out of scope, but I wouldn't overestimate it -- it's important for programmers to learn not to mystify compilers and interpreters early, and toy-forth is a lower bar than most.
@enkiv2 the problem is that I teach in an arts program where the intro to programming curriculum is very interactivity-centric. it's a challenge to work an exercise like this in among, e.g., "make an ellipse follow the mouse!" but a worthy challenge, I think
@aparrish Maybe implementing something more like turtle would be more thematic. There's more complexity in some ways & less in others.
(On the forth side, I'm sure there's a subset of postscript that isn't super ugly. Also if you teach a little postscript people can go home and write code directly into their printer, which is a nice trick.)
Der Blaue Mond 🌍@aparrish programming and poetry at the same time really sounds great...
Royal Rivera:tophat: @aparrish What language?
@r4stl1n prefer language-agnostic approaches, but the class in question is javascript
Hisham@aparrish Please throw everything you get back at us! There are more educators out there willing to improve :)
Crosswhispering (inactive)@aparrish I'm sorry if this already came up, there are so many replies! I'm currently learning python and using exercism to get to this second stage. hey have small nice examples for many programming languages to practice. It's a really nice platform and I prefer not to be under time pressure while programming (like I would be if I'm at a course and should get something done). Hope that helps!
@aparrish at exercism you can also get comments to your code and have more than one solution for the same problem. It's possibly to watch other peoples program once you put yours online.
Andrew Reid@aparrish
Famous in my world:
Software carpentry, https://software-carpentry.org/.
Also, the Computing Education blog is good, https://computinged.wordpress.com/.
Greg Wilson (Software Carpentry founder) is on Twitter: https://twitter.com/gvwilson
Also Lorena Barba from GWU: https://twitter.com/LorenaABarba
This is mostly scientific/engineering, of course.
@reid thanks! weirdly many of my colleages in tech follow both of those accounts, but none of my colleagues in the arts...