How are they so close??

Anyway my next goal is doing All Languages runs, where I do them all back-to-back

I did these by studying a bit of the languages for speaking, but I want to study them *for Scrabble* now. Learn the two and three letter words and such. This is what I consider fun these days.

To be clear, these times all suck real bad

My English time on this version of the game is like 15 minutes. A great time would be like 5 minutes.

But somehow in three languages I barely know, I got atrocious times, for very different reasons, years apart, all within 20 seconds of each other !!!

I just want to know now who that stock photographer was and why the New York Times is using pictures of German Scrabble tiles for English articles

Again I don't really have a problem with this, I just saw it and got confused about why it exists. It activated the part of my brain that thinks about how things are made.

At least this is miles better than the garbage AI Scrabble images you can find.

Or like, if an English set isn't readily available, all you have to do is get the Ö out of the frame and not spell anything with a Z in it, and then I would guess almost nobody would notice

I'm looking at other stock photos now of Scrabble tiles spelling something, and a bunch of these are spelling English words with German tiles

Like it's totally fine that a German stock photographer got some work spelling English words with Scrabble tiles, but why not use an English set?

I'm in a bad timezone for being able to watch the livestream, but the few matches I have caught have been great. And it's even been fun just to see how tight things are at the top of the standings! Some truly great players clearly brought their A-game to this!

I have a computer science background, and although I haven't given much thought to how to implement something like this, I have spent quite a bit of time thinking about what challenges would need to be overcome, and how you could try approaching those challenges, and I feel like you nailed it there.

Section 3 in particular was satisfying to read. I think you covered basically every "why speedrunning is a difficult and interesting problem for AI" thing that I've thought of in the past.

I've read enough of your posts the last few years to trust that you'd be that way, but seeing it called out in the paper that everyone in this research space should do that as well was nice.

That was a neat read! As someone who was actively speedrunning for most of the last 8 years, I really appreciate your approach of wanting to work with and communicate with the communities in question.

I'm enjoying looking through the random "Assorted Artwork" CDs in that collection, and seeing development screenshots and art from a bunch of games I know and love that have major differences from their final releases

Jersey Devil is one of my favorite games, and it had art hidden in one of those

I think the variety is what gets me the most. It's not just "another" Spider-Man chair every day. It's another chair that is unlike any of the ones we've already seen. Why are there so many kinds of chairs? How many ways have we invented to sit? let alone sit on Spider-Man