Sasha

Performance Engineer. Tokyo based 🇯🇵

[Sasha @devdotsasha]

for some reason the wifi in tokyo in public is really bad

[Sasha @devdotsasha]

@CryptoMikli Me after 2 highballs

[Sasha @devdotsasha]

https://t.co/MpvnULJMYV git stuff

[Sasha @devdotsasha]

the annoyance -> pipe to codex -> write as nvim plugin pipeline is beautiful

[Sasha @devdotsasha]

fast rust analyzer stills seems to exist in the realm impossibility however

[Sasha @devdotsasha]

not sure I've been to a worse major airport than bucharesti

[Sasha @devdotsasha]

I couldn't encapsulate everything frustrating about LLMs better, even if I tried

[Sasha @devdotsasha]

@thekitze he's not leaving git

[Sasha @devdotsasha]

time to replace copilot, in theory its useful but the execution is pretty terrible, I'll probably just wire in a codex instance into a nvim plugin, unless there exists something better already?

[Sasha @devdotsasha]

@IterIntellectus depends how you setup the game theoretic model, doesn't strictly have to be rational actors

[Sasha @devdotsasha]

@yacineMTB yeh I just pointed it at an old QMK file and asked to setup a new corne LP with ZMK, one shotted it, no toil, beautiful

[Sasha @devdotsasha]

@pomato_v @BigFatHater @marywitha4 Quite easy to conceptualise if you map it out:
- MM
- MF
- FM
- FF (ruled out)

MM / MM + MF + FM = 33%

[Sasha @devdotsasha]

I think its this: it depends what you define as a rational actor.

In a vacuum, one should think selfishly - if thats the case the intelligent choice is red - zero risk to oneself under any case. A rational actor.

However, in reality, peoples incentives aren't simply selfish (their utility function has an altruistic component) - there is a group consideration. Also a rational actor in this case.

If we then assume, that there is at least one irrational actor that votes blue (child or whatever), then it becomes rational to vote blue but only if you also assume a majority of other actors have group consideration and those actors have knowledge that the other actors have the same consideration.

The reason I suppose people think the red button is the smart choice, is because it is: but only in a vacuum or toy problem.

[Sasha @devdotsasha]

although, rationality is utility function dependent. If that utility function has a altruistic component for >50% of rational actors and there exists common knowledge of social altruism and you assume at least one irrational actor that would vote blue, then blue is the likely to be the rational choice.

[Sasha @devdotsasha]

@recmo @waitbutwhy what about if you assume a fully rationale set of voters and they all press red?

[Sasha @devdotsasha]

@JoelWBerry Complexity can be built with simple stated axioms - besides programming each atomic unit is not the only way to engineer, see LLMs / neural networks

[Auguste Prompt @augusteprompt]

I don't even know where to begin.

£500k over 18 months as tech investment into a town of 70k population to create a blue print for UK innovation in AI, education, and healthcare.

Anthropic is hiring for a single role at £630k and that's *excluding* stock compensation.

[Sasha @devdotsasha]

oh yeah

[Sasha @devdotsasha]

gfs visa has been the biggest pain in the ass and its driving me insane; that being said - anywhere to cowork from in BKK with monitors?

[Pierre Terdiman @PierreTerdiman]

man these new SIMD instructions are out of control