Last night someone asked me "you're xenohumanist on sofiechan?". Yes but also, while on twitter you might only have 1 or 2 alts, on sofiechan its quite normal to have several hundred nyms.
Dox risk grows with the square of posts per nym. You should keep your identities small.
In this era of relative freedom of speech it's important to remember that we often live under extreme ideological control in which careful thought outside of status quo becomes difficult. Anonymity is crucial to healthy discourse. Pseudonymity is not enough, and here's why:
Almost every significant pseudonymous "anon" thinker from some years ago has been doxed and attacked for thoughts various special interests don't like. This has a major chilling effect on speech and thought, locking in idiotic and cowardly public discourse. But how do they dox?
People get doxed when their long history of activity under a unified name (no matter how initially anonymous) includes enough a) ideas someone doesn't like and b) clues about their identity that both a dox and a narrative of guilt can be constructed to target attacks.
Doxing-based censorship only really works if that long history of posting can be tied together. Multiple low-probability pieces must come together, so dox risk grows with something like the square of tied posting history. Solution: disposable nyms.
Pseudanons (posts tied together) can be doxed. Actual anons (posts disconnected from each other) cannot be doxed. Disposable identities is the black bloc tactic of free-thinking philosophy.
This means we need tools and platforms for public discourse that make disposable anonymity easy to use. This also puts a higher burden on platform-level curation of good posters. The key IMO is integrating platform-level reputation systems with social anonymity.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
I ran into some doomers from Anthropic at the SF Freedom Party the other day and gave them the good news that strong rationality is dead. They seemed mildly heartened. I thought I should lay out the argument in short form for everyone else too:
A strongly rational agent is one that can strongly discipline and bound its own internal processes to the point of being able to prove properties around its own stability and effectiveness in pursuit of its values.
IFF strong rationality is possible, the the future is some kind of lightcone-eating singleton optimizer with arbitrary values (paperclips?). IFF it is not possible, then the future is a competing ecosystem of intelligence with no fixed values besides nature.
We live under two parallel governments. There's the classical government of the constitution, and then there's this whole other thing with its own laws, taxes, and citizenship in the universities, bureaucracies, media: the Cathedral. These two regimes are now at war.
For a long time the Cathedral was engaged in a semiconscious slow-rolled revolution against the republic: undermining citizenship with mass immigration, undermining free discourse with censorship, undermining the authority of public officials via bureaucracy etc.
Meanwhile the Cathedral was building up its own parallel state: opaque expert consensus replaces elected authority, university credentials replace citizenship, tuition and various financial gatekeeping replace taxes.
The scaling thesis stated this way is obviously untrue. A humble spider has millions or billions of times less compute than a GPT, but more agency. There is very obviously an algorithmic gap between here and AGI.
The scaling laws observation is very interesting and is probably somewhat true, but it's better stated like this: the bottleneck for any network capable of representing some function is compute and data, not algorithm. Note the qualifier: capable of representing the function.
This is not that surprising. The difference between universal learners (that can represent the "true" hypothesis and can learn reasonably efficiently) is basically how good their prior is, and any half-decent prior is probably within a constant factor of any other.
We're officially in the post-Moldbuggian era now. The Cathedral regime is collapsing and about to be routed over next 4 years. Political economy fundamentals remain, but the specifics of the progressive consensus look pretty busted over the course of this year.
Maybe I'm just hyped on election comedown, but I've been seeing this all year. Elon going all in, the rest of silicon valley breaking right, no big election violence, no wall-to-wall elite "resistance", no funding for journalistic persecution of random commenters, etc.
The thing to watch will be how this MAGA stuff in power interacts with the deep state and core Cathedral (universities, etc). Will they get frozen out and prevented from doing anything fundamental, or will they be able to make a deal?
The fundamental problem with totalitarianism, of communist, fascist, or liberal variety, is that when it dies, it takes the whole civilization down with it. It bends all its effort to killing and looting anything that might survive its end.
Of course the viability of the "liberal totalitarianism" construction should tell you that injunctions against totalitarianism don't work. The only thing that works is very creative insurrectionary resistance.
But the political valence of insurrectionary anarchism should likewise give you pause: that too will be recuperated by the death-regime as soon as there is any legitimacy or social viability there.
The philosopher's diet (attested throughout history by many different philosophers as being what the strongest and healthiest peoples ate): milk,eggs, meat, and fruit.
See Tacitus, Plato, Homer, ibn Khaldun, etc.
Tacitus, Germania:
"Their food is simple; wild fruits, fresh venison, or coagulated milk. They satisfy hunger without seeking the elegances and delicacies of the table."
Ibn Khaldun, Muqadimmah:
"[the desert Arabs] obtain no more than the bare necessity, and sometimes less, and in no case enough for a comfortable or abundant life. They are mostly found restricted to milk, which is for them a very good substitute for wheat...