@gjm I suspect your conjecture is correct, which means that some ambiguity doesn't really matter that much. (Though I won't pretend it's intentional.)

@gjm Agreed. Although there are a few questions where people are overwhelmingly giving a percentage near 50% and then almost universally saying agree/disagree. Which seems to suggest they aren't confused?

Lesson learned: Always test surveys on real people before releasing!

@gjm Unfortunately, I've gotten a LOT of similar comments, so I don't think you're alone in being confused. My intention was to ask if you agree with the proposed policy, not if you agree with the majority of Western adults.

(Fortunately, the most important info for me is the predicted percentages, which I hope is more unambiguous.)

@gjm Shame though, because I use Cryptpad in part because I think of it as being friendly to people with eccentric browser setups, etc.

@gjm That's concerning. (Concerning enough that I just exported all the current responses to be safe.) Not sure what's going on—my account still seems to be active. Have you used cryptpad in the past?

Please take my weird moral puzzles quiz https://dynomight.net/puzzles

Seems about right.

@surprisetalk incidentally, is this a feature of a bug?

(Your page is weird enough that I can't tell, which I greatly admire.)

@khinsen What do you think of the argument that for array languages, it's been very common for new languages to come about and then actually get adoption? Does that suggest we need a "big tent" of interfaces? I feel like this is pretty unique in the world of programming languages and might reflect something unique about the space.

@khinsen You know, I bet you could use the same strategy I did to easily make that API work now—have the J-style API compile everything into vmap statements.

Personally, I *think* loops and indices will still be more clear, but I'm a dilettante with APL-type languages.

@khinsen Very interesting! Is there anywhere that gives a picture of the initial API how it's evolved over the years?

@isaacg I believe this concern is valid and more or less acknowledged. I think the generally suggested solution is basically what you suggest, you'd need a highly liquid market, so it's not worth "buying" the outcome".

(Though I think that solution kind of presupposes that the problem I'm pointing out doesn't exist...)

Futarky's fundamental flaw https://dynomight.net/futarky

lots of books featuring sugar alcohols

Optimizing tea: An N=4 experiment https://dynomight.net/tea/

My advice on (internet) writing, for what it's worth https://dynomight.net/writing-advice/

I finally got one of these, the most uncool of all home automation products. It's... really good.

I think I'd prefer one that worked with whistles.

@trurl I hadn't seen this particular paper. It seems maybe sorta related to jaxtyping? (Which I must admit doesn't make a lot of sense to me... Static typing in Python always seems to descend into madness when I get beyond simple cases)

@BenediktEhinger My impression is that Julia is much better for the "run loops on CPU" type scenarios. But is it really easier when using a GPU? I know there are a million experimental julia packages that try to do incredible things, but it's very hard to know how smoothly things really work in practice.

@adredish Really! Can you provide any refs/links? I knew matlab had some auto-JIT magic that was pretty decent, but I had no idea that it could vectorize and GPU-ify, and I'm having trouble finding much info by searching.