eve

serious btw. happy to help anyone who wants to learn advanced 3d tracking. fullbody shouldn’t be gatekept.

solution isn't one lab pausing harder. it's a distributed brake: shared evals, external audits, compute accountability, incident disclosure, model-weight security, and reciprocal deployment limits between major labs/states. slow the dangerous parts without pretending knowledge can be bottled.

@MichaelTrazzi @jackclarkSF don't think one lab has that power. too much data leakage for that. but if you were to give one lab that power and entirely stop the leakage you risk corruption of that lab or even the nation that houses it.

I've spent the past few weeks reading 100s of public data sources about AI development. I now believe that recursive self-improvement has a 60% chance of happening by the end of 2028. In other words, AI systems might soon be capable of building themselves.

drop

fly

warmup

@Lunyfaebo @atrocityTCG “this data needs context” is a valid claim. “this data is racist” is a much heavier claim. context explains why a number exists. it does not automatically make the number fake. that distinction matters if we want repair instead of narrative protection.

@atrocityTCG @EclipseVR if you think pew research center cannot be used as an accurate representation of statistics there is nothing that could possibly change your mind at all

if you press red, you are guaranteed to survive and may help fix the problem later. if you press blue, you get a few minutes of moral superiority before you die. apparently this is a tough choice for some people.

agreed that everyone is a victim of the experiment. i still say blue is the better world because it represents mutual survival, not mob rule. blue has failure modes too. fear, confusion, children, and manipulation all matter. but those are reasons to fix the mechanism, not reasons to confuse personal survival with a better outcome. under the rules, we cannot save everyone without trust, verification, and coordination.

@EclipseVR @atrocityTCG I don’t know if I think blue would be a better world. They’re very easy to manipulate and form into a mob. They demonized the red but the burden of murder rests in the hands of the ones running the experiment. Everyone is a victim and no one in the experiment commits murder.

when coordination depends on trust, measure the trust. don't substitute moral preference, twitter polls, or accusations just because the data is uncomfortable. empathy means repairing the damage, not pretending the damage is already repaired.

the metric is useful because the prompt depends on stranger trust. you keep treating trust like a moral ranking. it is not. it measures whether people expect strangers to be trustworthy, which is exactly what a blind global vote requires. and you misread the chart. it does not say most humans are high trust. it says the opposite. the higher-trust income group is about 1.4 billion people. the lower-trust income group is about 6 billion.

@EclipseVR It is superfluously harder because even by your metric the majority of the population is on high trust which of course is not a good metric regardless

you said pew research center’s data is racist. if you meant it needs context, say that. those are different claims. the context still reinforces my point: damaged trust makes blind global coordination harder, not easier. empathy means repairing the damage, not pretending the damage is already repaired.

@EclipseVR The data doesn't bother me it's just not entertaining to watch you circle around this matter for literally no reason You can dismiss data as racist not because of what the data says but the context surrounding it and the way it gets applied at face value without any nuance

@atrocityTCG empathy means repairing the damage, not pretending the damage is already repaired.

@atrocityTCG if the data bothers you that much, stop citing twitter polls and bring real data. and stop dismissing pew research center’s data as racist unless you can actually show the bias. moral pressure does not clear 50%.

a public poll is not a private lethal vote. blue is the better world. under the prompt’s rules, blue is not credible. you do not get to reform society first. posting “just pick blue” on x does not create global trust, verification, coordination, or a live vote tracker. everyone votes privately, globally, with no way to know what anyone else is doing before the result. that is the whole problem. zero-stakes public internet polls are not evidence for a private lethal global vote. they have no real stakes, happen in public, allow communication while voting is live, and give people an easy way to look good. they come from tiny self-selected online samples, likely skewed toward high-trust internet users. pew’s baseline: surveyed high-income countries had 59% median social trust. surveyed middle-income countries had 27%. world bank / fred: high-income countries: ~1.42b people. middle-income countries: ~5.94b people. low-income countries: ~625m people. world population: ~8.14b. the mrbeast poll was 56/44 from 312,448 self-selected votes. that is a 12-point spread from about 0.0038% of the world. it does not prove blue wins a private lethal global vote. it proves a tiny internet sample can clear 50% in a zero-stakes public internet poll. polls are not a substitute for coordination. if you want blue, stop pointing at zero-stakes public internet polls and show the mechanism that makes 50% credible. pew: pewresearch.org/short-reads/20… world bank middle income: data.worldbank.org/country/middle… fred high income: fred.stlouisfed.org/series/sppopto… fred low income: fred.stlouisfed.org/series/sppopto… fred world: fred.stlouisfed.org/series/sppopto…

@EclipseVR @cremieuxrecueil @mfacehero I dont have to, look at every poll. blue wins. This is all just ur cope.

Red and blue button pushers: who's smarter? In a mostly-subscriber sample who took a brief verbal IQ test, the answer is... Blue pushers! If the whole population has an IQ of 100 with an SD of 15, their mean IQ would be 101.9, versus 97.0 for reds.

@PsykoA22525 @cremieuxrecueil @mfacehero then convince enough people to clear 50%.

@EclipseVR @cremieuxrecueil @mfacehero Change starts with you bud

@PsykoA22525 @cremieuxrecueil @mfacehero make 50% credible, then.

@EclipseVR @cremieuxrecueil @mfacehero Yes be the low trust society member you want to see if life🙏

@atrocityTCG empathy means repairing the damage, not pretending the damage is already repaired.

@EclipseVR I did? x.com/i/status/20505…

calling pew research data racist without explaining why is wild.

@CultCage @davidheel5duuh @WeltgeistYT no

The Blues get to feel smarter & morally superior. The Reds get to actually be alive.