@sameQCU

back to the regularly scheduled cryptic posts DMs open. some published models: https://nitter.cf/t.co/YAbJvGkgKO some published code: https://nitter.cf/t.co/ZE5Y59WayI

 1 regional flight from you
sqcu.dev
Joined September 2020
  • Tweets51K
  • Following1K
  • Followers3.4K
Oh yeah. I guess I'm in San Francisco from March 3rd.
3
30
7,086
nitter.cf/sameQCU/status/2110252… for reference, a lot of the design notes i'm pulling together on stellaris (nu) are because the game is so radically unforgiving in the game systems that it might only be *fair* if most factions were played by llm-style rl-optimized policies!
サメQCU
@sameQCU5h
Replying to @sameQCU
(at 15 years into my game, you can already see my first neighbor trying to evict *me* from the game!)
1
7
224
no, not rl-optimized to *win*, rl-optimized to have motivations or goals with semantics besides 'maximize aggregate resources given resource multipliers and featureset enabled per faction'. you can use rl for that too, not just winning!
1
3
37
OOOOOOH. mining stations in nu-stellaris are an obvious trap and i just didn't estimate how bad they were as a rate limiter on growth.
1
2
192
more replies
the only reason to not take all of these actions as fast as possible in the history of your game is because you were making combat ships to kick the neighboring player in the dick so hard it turns inside out and they quit the game; if you didn't force them to quit you're *losing*
1
2
38
(at 15 years into my game, you can already see my first neighbor trying to evict *me* from the game!)
1
1
3
245
the game was not always like this; i actually played the launch version of the game, for example, where there was optimal play related to splitting corvettes off in all directions to scout all systems in parallel for what to survey in what order. that was *ten years ago*.
1
3
11
anyways the pareto dominance of 'mining stations' by this more important choice escaped me until this point; i didn't realize how much the incentive structure of the resource production relations had shifted...
1
1
9
Load more