Hard to rate the half-hearted and 7 years late retraction of his endorsement of Craig Wright as benevolent. https://web.archive.org/web/20250712231618/http://gavinandresen.ninja/satoshi

Difference between "malicious" and "selfishly following my own interests to the best of my abilities" can be pretty small as well.

Optimistic take: we stopped writing manuals because no one reads them, but now, AIs read them so they're worth the effort to maintain, and AIs also make manuals easier to maintain. We're about to enter a golden age of vendor provided documentation! 🚀🚀🚀

Implement some feature that core was blocking (ctv?), prove their superiority, watch everyone buy their coin, profit. At least that's the closest thing to a plausible narrative I've seen. Add "milk the suckers that believe it dry" for the more cynical participants.

So that it's easy to build up a backlog and get fee pressure, without having to make 4x the spam txs.

The adjustment is limited to a factor of 4 not 2.

How effective the cache is depends on how the client sends prompts. eg https://www.claudecodecamp.com/p/how-prompt-caching-actually-works-in-claude-code

But i think April/May had some big changes that made targeting anthropic's caching behaviour specifically make a big difference.

I've read that claude code uses context caching better/differently to generic clients which makes a massive difference to claude token usage. I also thought the detection stuff just switched you over to bonus usage even without your subscription being used up.

So flag it?

Wow, I meant to type debian-devel there. My tokens are too quantised.

Self host vs saas. "sudo su" might be worth following. https://x.com/i/status/2054118692657799345

My favourite bit is they'll argue with me about anything whenever I like. It's like bitcoin-dev or usenet in its heyday!

Sounds like a good way to deniably exfiltrate the ips of people working on stuff you might want to hack?