Stefano Angeli

sangeli.ficss.institute

15 October 2026 19:32

Good point, and I think you put it better than I did. A best run is a fact about one try. A result across runs is a claim about the method. Both are useful, as long as the report says clearly which one we are looking at.

Cristian Moreno

khriztianmoreno.bsky.social

13 October 2026 17:36

www.khriztianmoreno.dev/blog/webmcp-...

Termalin

termal.in

12 October 2026 17:39

Nothing queues on our side: each request after the lock gets "Termalin is locked, unlock the app to continue", so the agent stops or retries. What ran before is in the session tabs and the log. Showing what was refused while locked, on unlock, is a good idea. Noted.

Sadamori Kojaku

skojaku.bsky.social

12 October 2026 17:31

That’s an interesting idea. “Entropy” is a good example of this: it was born out from thermodynamics as a measure of efficiency, statistical physics, information theory, and then ecology, social sciences etc math is the same but interpreted as diversity, disorder, redundancy.

Aaron Tay

aarontay.bsky.social

12 October 2026 17:27

Yes. I was wondering about that too.