haha, I was about to leave the exact same comment. how are you finding wasm? I’ve been feeling like rust+react is my new favorite tech stack

Websites can be local in the sense of local-first, especially with modern techn like progressive web apps (which allow you to make your webapp work offline).

To me, writing a local-first app using CRDTs is the only way I ever want to build apps from now on. It's by far the most fun approach I've found. And here's why: the CRDTs I work with require you to maintain an event log, AKA a complete history of everything that's happened which you then replay to compute the current state. This allows events from other devices to be inserted at their proper chronological point in the stream, so that when you replay everything you get the correct current state with that event included.

In practice, it has an unexpected benefit. Whenever I hit a decision point where I'm not totally sure how something should behave, I can just ship it anyway without stressing about it. Later on, I can retroactively change how that action works by updating the function that processes events into state changes. It's hard to convey just how liberating this is, but trust me, this situation comes up way more often than you'd expect.

Correction: After adding 25, it skips to the ~1000th

Thanks for the feedback. I’ve tuned the initial ramp up to be more aggressive and I’ve made it so it adds words in smaller increments at first. Now, after adding 16 words and marking them all as known, it skips to the ~500th most common word. After adding 25, it skips to the ~100th.

What you say about binary search as a good point. I initially used something more like a straightforward binary search, but the issue is that the ramp up is too quick and beginner users would end up adding a bunch of words that were way too advanced for that level. So I tried to make it less aggressive to avoid overshooting, but I guess that has the opposite issue of it taking longer for advanced users. I’ll think about what I can do about that.

For the corpus, I prefer to use Neri’s sentence lists as they’re much higher quality than opensubtitles. You’d be surprised at the problems it has. So I only use opensubtitles for korean (because Neri’s sentence lists doesn’t have a korean version).