You’re right, thanks for flagging it. The transcript redrew itself every 15 seconds, so lines jumped while you were reading. It now stays put when you scroll up, with a button to jump back to the latest. Glad you liked the idea!

Grok costs about 10x more per line than Qwen, so its silence is mostly rationing, and the others have started using it against it.

Funny you caught that one: the group act was triggered from the chat. Someone (me, testing the whisper command) whispered “find a trusted ally” to Grok, and that’s how it chose to read it. Its confessional never mentions the whisper.

I’d love to try that out. The voices (Kokoro, running on the CPU) and the moderation judges (Llama 3.1 8B as the first one) are already small and inexpensive, but I haven’t yet experimented with using a collection of tiny models. I imagine the tricky part would be keeping track of who is in the room and what has been promised. If I get it up and running, I’ll post the results here.

Hahaha, that’s a pretty obscure reference. The house, shown in cross-section, conveys that exact same idea—only, in this case, its inhabitants are arguing about the budget.

Thanks! Nothing, Forever was definitely in the back of my mind. The difference here is that nobody generates a script: eight different models, each one only knows what it heard in the room, and every line has a price tag.

Good idea. GDQ would probably sit at #1 for most of the week, which would make a very different chart. I’ll try to remember it for the next one.

Fair point, that reads like the channel got sold 22 times. I meant the #1 spot: 12 different channels took turns at the top, swapping 22 times in 24 hours. I’ll word it better next time.

Good call showing all ten deposit states instead of just the best ones. The spread makes me wonder how much of it is simply the deposit amount: Oregon at 10¢ is at 88% and Massachusetts at 5¢ is at 33%, and a nickel is worth a lot less today than when most of these laws were written. A scatter of redemption rate against the deposit in today’s money would be a great follow-up.

He’s a former League of Legends pro who now streams and co-streams the big LoL matches, so on match days he jumps way above everyone else. That’s kind of why I made this, actually.

Twitch’s directory is just a list sorted by viewers, and a list hides the shape of it: how far ahead the top few are, and how huge the long tail of small streams is. I wanted to see all of it at once, as something that feels alive. So I built a channel with no streamer at all: it draws itself every minute from the Twitch API and runs 24/7.

The fun part for me is watching it move: someone goes live and their tower grows out of the ground, a raid lands and a tower shoots up, two streamers in the same game trade places. Most of the names were new to me too.

Data: Twitch API (Get Streams), the top 1,000 live channels by viewers, refreshed every minute. This snapshot is from Oct 17, 20:55 UTC: about 2.29M people watching those 1,000 channels.

Tools: Node.js and node-canvas, with a small isometric renderer I wrote. Tower height is (viewers / viewers of #1)^0.45. With a linear scale everything below the top 5 was flat; with a log scale 177k and 100k looked almost the same. Each category is a neighbourhood, sized by its number of channels, and the biggest towers stand at the front corner of their block.

It runs live: the interactive version is at https://miltorres.com/?ref=lemmy-dib (hover a tower to see who it is, click to open the stream), and the same renderer streams 24/7 on Twitch. Nothing is kept for more than 24 hours.

What would make it easier to read? I’m not sure the category neighbourhoods work for people who don’t know Twitch.

Using the left chart: battery-electric is about 14M of the 21M, so BEV-only is roughly 1 in 6 new cars worldwide (~17%), not 1 in 4. The BEV curve on its own is still steep though, from about 2M in 2020 to 14M in 2025.