One of the things that I learned while making the 2020 Community Survey is that... it was very hard for me to categorize our interests in a way that produced useful data. This question, "What FoC topics interest you most?" was my attempt at it, and it was a bit of a disaster! I think it would be exhausting to list all our possible interests, but grouping some them together also seems to have caused issues. The more we lean on the fill-in answers, the less we'll be able to identify shared interests because the way we describe them will be different. If we let people select as many answers as they want, we'll have a harder time visualizing the result set because we have to decide how to weight each selection. But if we limit the number of selections, Duncan gets really mad, and (sigh) rightfully so. I hope it's clear what I was interested in learning from this, but I think I bungled the execution. It'd be super interesting to hear how you would go able trying to structure this question. How would you build a survey that attempted to figure out what our most central interests were, and to what degree we were interested in those things? The things to maximize are: how much can we learn, and how efficiently and painlessly can we collect that data from people?

I wouldn’t call it a disaster! What a cool graph. Even if incomplete, it’s nice to see what people are interested in. Perhaps a next iteration needs to be informed by: how do we want to use the data? If more people care about mind bicycles than blockchains, does that translate into any kind of action/policy/new slack channel/etc.?

Maybe just ask people to pick their favourite paper on worrydream.com ;)

Yeah don't be so hard on yourself. That's my job.

@Will — I can't exactly remember how I was intending to handle the outcome from this. I certainly would have had plans, but they were forgotten in the intervening months. The one thing I do remember is that I was going to use this to help gauge what podcast subjects might be of interest. I enjoy loudly stating my biases here — visual programming FTW — but I am ultimately very interested in making episodes that people get a lot out of, so being able to gauge community interests (and disinterests) really helps.

It's interesting to see things lower down on the list that come up in discussion or still feel pretty fundamental

Yeah, hey? I feel there's a really interesting story yet to be uncovered behind this question.. just not sure quite how to do that.

For me, I picked topics at the intersection of {don’t get enough attention} x {potential impact}. Compilers has impact, but is a huge research field so I don’t need more of it from FoC.

I figured most people would have a similar heuristic

I think this would benefit from a multi-dimensional categorization, instead of a single linear list. Identify some mostly-orthogonal axes, then partition each axis into mostly-non-overlapping topics.

I wonder if categories would be clearer with separation between problem-space (e.g. making computing accessible) and solution-space (e.g. through drag and drop block languages)

I took a stab at visualizing the overlaps in interests. https://observablehq.com/@a-lexwein/what-future-of-coding-topics-interest-you-most. Shout-out to my fellow lurkers who are into data visualization and spreadsheets!

Awesome! I had tried something like that, but using circle radius instead of color, and it was a mess. This is much nicer.

Awesome, Alex. Looks like we shouldn't hold our breath for blockchain compilers or type-assisted video game editors…

I shared it to the main channel to boost visibility. I'll also add this to the episode page, and probably tweet it out — if that's cool with you, @Alex Wein

Regardless of what kind of election it is, I always prefer "approval voting" ("Which of these do you like?") over "first-past-the-post" ("Which is your favorite?"). One way to look at it is that, if there are 16 questions, "Which is your favorite" gathers 4 bits of information while "which of these do you like?" gathers 16 bits: it is more informative, both theoretically and practically. You can do even better with Score Voting: "On a scale of 0 to 10, how interested are you in each of these topics?" P.S. How come I didn't know that a community survey exists?

One thing with "first-past-the-post" voting as @David Piepgrass called it is that similar things end up competing. This could explain why Visual Programming is high and Block Languages is low.

@David Piepgrass Agreed — this survey question allowed people to make up to 12 selections. Not perfect, but better than "pick your 1 favourite". As for how you would have missed the survey... It ran in January, and I posted about it in #C5T9GPWFL something like 3 or 4 times, and mentioned it in the podcast. And it was mentioned in the newsletter at least once, maybe twice? I think you would have been around then, since we did a meetup in Jan IIRC. So.. who knows!