I have updated SoP with a better definition of the fork choice rule: monero-project/research-lab #146

@gingeropolous you can release your LLM on it.

tevador: fyi the irc/matrix bridge is down

ai says that "33% attacker gets a 3-block reorg about once per [0.83] days" not 10

The once per 10 days number comes from here: monero-project/research-lab #102#issuecomment-2402750881

Can you explain how the AI got to 1/0.83 days?

tevador sorry for the delay , ignore this is if it's noise, appears to be pointing at something in your sim privatebin.net/?0ee1a2b7dc85041b#9C…3Lap57gh9jtfU1QfPFyQWvB4pt1upDQ4KLj

I've reproduced the AI numbers. Chapter 6.2 is updated, I've removed the 3-block reorg claim and left only the main result with the 10+ block reorg frequency.

I think I know what happened. The rewrite of the specs lost one property I had in the original version. The reused honest workshares didn't count toward the 48 work object limit.

plowsof: Can you let your AI check the current version of the specs? The weakness should be fixed.

Note: my published simulation code no longer reflects the specs.