Notices by Evelyn fra denne andre øya (evelyn@misskey.bubbletea.dev), page 6

fedi lawyer is eepy, sue ya

influencers are mortal, they bleed, they can be destroyed!

The idea that "AI safety" is about AI not becoming fucking skynet and taking over the world is very much something being pushed by AI companies, because on the one hand it helps feed a misconception of AI as being all-powerful and all-useful, even on the verge of sentience, on the other, it distracts from ways AI is already being used to harm people, from hiring algorithms to erasing the commons, to safety-critical machinery

a lot of the the things described as "enshittification" could imo be described as rugpulls, a company wilfully destroying good will and fucking over contributors to try to make a few extra quid now they reckon they're past needing it

"enshittification" really does mean nothing now

@newt@stereophonic.space dsee nuts

i think a reasonable tradeoff when it comes to fireworks is that you're allowed to set them off at random times, but people are allowed to "have a polite discussion" with you about how annoying it is

I've been past "mosquito" type devices seemingly intended to deter animals, it's great walking past someone's front garden then finding yourself in pain. my only regret is not going into the garden and smashing it

entering my dragon era

gagarin

nuclear planes are a silly and dangerous idea, a nuclear reactor? in the sky? preposterous!

this is why I think my proposal, the trolleyplane, combines the best of electrification and nuclear power,

I got a call yesterday from a recruiter asking me yet again about an erlang developer role with bet365. Since gambling is a scourge, I still say "no"!

you are a capitalist, i am a lowercaseitalist, we are not the same,

A company did a "total rewrite" of mastodon, then deployed it with soapbox as the frontend, to demonstrate that their product supposedly allows you to build in "twitter scale" with "only" nine months of one person's time.

i.e. they wrote new fedi software, which inextricably depends on something that they supposedly spent ten years writing, and are clueless enough that they're still calling it "mastodon". This should make you wonder what other corners they cut.

https://blog.redplanetlabs.com/2023/08/15/how-we-reduced-the-cost-of-building-twitter-at-twitter-scale-by-100x/

The instance itself is interesting, because it's inhabited by a spambots (100 million of them, according to them) constantly posting nonsense (3500 posts per second, again according to them), ostensibly to demonstrate that they've achieved "twitter scale"

Instance is: mastodon.redplanetlabs.com

vriskification

if someone seriously suggests relying on machine learning models for safety-critical systems, never take them seriously again. If someone cannot understand why life and death should not be decided by unnecessarily complex and unauditable software, I'm not sure you can trust them with safety in general.

someone on hackers.town accused me of writing a post using gpt a few months ago, and no, i regret to inform all of my bad takes are worded by myself

there's something quite horrid about being so casually treated as inhuman simply for not having the same range of language as a "normal" person

@shibao idk, i live in the UK and I own one, soooooo

@shibao i see, it's not tacticool enough for you