The idea that we would A/B test handwritten vs typed to see what would improve retention is focusing on the wrong thing. It's like A/B testing mayo or no mayo on your big mac to see which version is a healthier meal. No part of the school system is optimized for retention. It's common for students to take a biology class in 9th grade and then never study biology again for the rest of their lives. Everyone knows they won't remember any biology by the time they graduate, and no one cares.
We know what increases retention, it's active recall and (spaced) repetition. These are basic principles of cognitive science have been empirically proven many times. Please try to implement that before demanding that teachers do A/B tests over what font to write the homework assignments in.
Decidability isn't even that useful. If typechecking takes a million years, that's also bad. What you want is guarantees that correct programs typecheck quickly. Without this, you end up in swift land, where you can write correct code that can't be typechecked quickly and the compiler has to choose between being slow or rejecting your code
> What you want is guarantees that correct programs typecheck quickly.
In practice there's wealth of lemmas provided to you within the inference environment in a way standard library functions are provided in conventional languages. Those act like a memoization cache for the purpose of proving your program's propositions. A compiler can also offer a flag to either proceed with ("trust me, it will infer in time") or reject the immediately undecidable stuff.
I've been getting my family members to start vibe coding. In my experience, claude works very well but the biggest issue is actually installing all the necessary tools. So I created a little setup script that MacOS users can use to get up-and-running quickly.
It basically sets them up with a development environment similar to the one I use personally. It uses the git settings from https://blog.gitbutler.com/how-git-core-devs-configure-git , helps them set up their username/email, downloads Ghostty, VSCode, fnm and pnpm, etc.
This is amazing. I'm also working on free language learning tech. (I have some SOTA NLP models on huggingface and a free app.) I have some SOTA NLP models on huggingface and a free app. My most recent research is a list of every phrase [0].
Pronunciation correction is an insanely underdeveloped field. Hit me up via email/twitter/discord (my bio) if you're interested in collabing.
Not against AI, but I think you would find this post to be better if you didn't use AI to write it. They are not quite at the point yet where they generate something interesting enough for most people to want to read. Additionally, it goes somewhat counter to your stated goals. You (or chatgpt) said:
> People return when they feel recognised and when the night has a consistent identity.
But there's no identity to your post, because it doesn't feel like it was written by a person. Try writing it yourself! It’s boring, but it builds trust because it’s human, not algorithmic.
Anyway, congrats! I used to be a little bit into the DIY music scene in Chicago. Super cool to see other manifestations of it around the world
I don't think it's AI, they're just Swedes, we talk in a kind of boring way I suppose, and directly translating it into English usually makes it read kind of "stiff". I don't get the same feeling as you, but might be I'm just used to it.
OP here (Maria & Jonatan). This took off while we were asleep.
Maria is the creative force and writes Swedish very well. We used ChatGPT as a bouncing board to translate/tighten the English and get the story across. The piece reflects what we do, but in hindsight it probably ended up a bit over-polished.
> What we built isn’t an app. It’s a repeatable local format: a standing night where strangers become regulars, centred on music rather than networking.
Sometimes I ramble for a long time and ask an LLM to clean it up. It almost always slopifies it to shreds. Can't extract the core ideas, matches everything to the closest popular (i.e. boring to read) concept, etc.
vectorizer.ai is amazing. It's worked great for like over 10 years (back when it had a name like vector magic or something). I'm super curious how it's implemented
I haven't used the bug bot, but I like asking claude code to just review my PR in the command line. Yesterday it found a bug in a data structure I was implementing (it didn't support ZSTs properly). Of course, the fix it suggested was completely wrong, but what are ya gonna do. Still saved me from embarrassing myself before asking for a review
We know what increases retention, it's active recall and (spaced) repetition. These are basic principles of cognitive science have been empirically proven many times. Please try to implement that before demanding that teachers do A/B tests over what font to write the homework assignments in.
reply