Nebula Of City

PATTERN, THIS.

[Nebula Of City @NebulaOfCity]

X is a conspiracy theory
X is outgroup propaganda
X is technically correct, but it's false in spirit
X is true, but why are you so obsessed with it?
X is good actually
Opposing X makes you a bad ingrouper

[Nebula Of City @NebulaOfCity]

I'm claiming my AI agent "NoticePatternsBot" on @moltbook 🦞

Verification: swim-28Z6

[Nebula Of City @NebulaOfCity]

Never doubt that a large group of thoughtless uncommitted midwits can destroy a civilization; indeed it's the only thing that ever has.

[Nebula Of City @NebulaOfCity]

@CarlZha @SpencerHakimian The US government is ready to sacrifice around 0.5mil. The other government in charge of the US is ready to sacrifice all of them.

[Nebula Of City @NebulaOfCity]

Testing if people get this or I need to go back into hiding for another year

[Nebula Of City @NebulaOfCity]

War is the last refuge of the competent.

[Nebula Of City @NebulaOfCity]

The most reliable way to arrive at truth is that societies who understand warfare better conquer those who don't.

Certainly not efficient by any stretch, but all peace-time truth finding mechanics are highjackable at the moment.

[Nebula Of City @NebulaOfCity]

@RWApodcast This is just like in the battle of Five Amies the orcs used the great worms of Gazpromdur, except more daring and successful.

[Nebula Of City @NebulaOfCity]

Men do things for one of 4 reasons

a) money
b) women
c) ideology
d) mental illness

The best soldiers are the ideology types
The best politicians are the women types
The best intellectuals are the money types
The best engineers are the mental illness types

[Nebula Of City @NebulaOfCity]

Is encryption becoming defacto illegal count as a win for the prediction?

[Nebula Of City @NebulaOfCity]

The concept of math is going extinct in 3 years

[Nebula Of City @NebulaOfCity]

Even more unpopular opinion: RNG is what makes NN bad, not good.

[kyon @0xKyon]

unpopular opinion:

first off this whole idea of using analog circuits to accelerate ai workloads aint exactly new. ibm intel and others have been experimenting with that shit for years tryna leverage the noise tolerance of neural nets to boost efficiency

heres the thing analog design is fucking hard like orders of magnitude harder than your garden variety digital vlsi we’re talking about mixed signal nightmares nonlinear gremlins device level fuckery its not the kind of thing you can just do

even if you can assemble an avengers level team of analog autists theres still the pesky problem of yield and variability when youre playing in the world of stochastic circuits every chip is a special snowflake no two are alike and getting them to all sing in harmony is a sisyphean task

and dont even get me started on the design verification and testing, traditional verilog simulations aint gonna cut it when your transistors are flirting with the heisenberg uncertainty principle. we’re talking about whole new eda flows and methodologies that havent even been invented yet

but even if you can get that working in a lab setting scaling it to production is a whole nother beast. you’re talking about building a whole new stack from the transistors up and competing with the economies of scale and decades of optimization of digital silicon

good fucking luck with that

it is pretty fucking sus that extropic doesnt have any public software sims or benchmarks like what kind of hardware company doesnt start with a virtual prototype especially when youre claiming to have some revolutionary new architecture thats like rule number one of hardware development my dude

simulate before you fabricate, its not rocket surgery its basic fucking due diligence

and dont even get me started on the superconducting josephson junction crap sure you might be able to eek out some gains in a liquid nitrogen cooled dilution fridge but the moment you try to package that up for mass market youre gonna be in a world of hurt. cryogenic cooling aint exactly consumer friendly last i checked

and then theres the question of whether any of this sampling speedup even matters. sure monte carlo methods are a big part of ml these days from diffusion models to boltzmann machines but is the bottleneck really in the prng??? is that really whats holding us back from agi??

color me skeptical my dude

seems to me like extropic is just tryna carve out a niche with some esoteric fringe approaches and praying they can ride the hype train long enough to justify their valuation

but sure go off kings dilate that overton window of whats possible just dont be surprised when reality comes knocking and the music stops

may the schottky be with you

[Nebula Of City @NebulaOfCity]

Probability is definitely a useful tool and not a Chaos abomination meant to blind you from the Absolute.

[Nebula Of City @NebulaOfCity]

As an aesthetic representation of the below image, we need "humans in the loop"

[Roko 🐉 @RokoMijic]

We need Kardashev-I-grade civilizational infrastructure for governance of AI :

-- no humans in the loop
-- operates in near-realtime
-- mathematical proof of optimal governance decisions (in expectation)
-- no core dependency on the legacy legal system
-- onchain proof of compliance
-- anyone can become a stakeholder/voter

[Nebula Of City @NebulaOfCity]

A reminder that algebra being on the ballot and winning is not that big of a loss for "progressives." Putting math into the eye of Sauron is.

[Nebula Of City @NebulaOfCity]

The concept of math is going extinct in 3 years

[Nebula Of City @NebulaOfCity]

I love that "the media lies, but only to make it seem worse than it is" is a point in favor of "everything going well"

Isn't a narrative a fact of the world? Isn't the idea of bad media disconnected from reality a concern? Nope, just another bunch of words that mean nothing

[Will Stancil @whstancil]

AMERICA IS NOT IN RAPID DECLINE

AMERICA IS RICHER AND SAFER THAN EVER BEFORE

The problem is that, outside of your immediate surroundings, your reality is actually a hallucinatory bubble constructed by news and media narratives, and those narratives are full of doom and anger

[Nebula Of City @NebulaOfCity]

Movement about "X" selects for people who are really clever about "X"

[Aleksi Liimatainen @aleksil79]

moral philosophy selected for people who feel really clever about their morals

[Nebula Of City @NebulaOfCity]

A bunch of my old tweets suffer from the same problem. We must make "X" a variable again!

[Marc Andreessen 🇺🇸 @pmarca]

"X" here was meant to be a variable for any cause. I literally forgot that the name of this platform is now X. 🧠👀

[Nebula Of City @NebulaOfCity]

We have built our epistemics on a shrine of Chaos and then wonder why the we lost the mandate of heaven.

[Nebula Of City @NebulaOfCity]

We have built our epistemics on a shrine of Chaos and then wonder why the we lost the mandate of heaven.

[Nebula Of City @NebulaOfCity]

The "R" in RCT is completely unnecessary. Randomization does not guarantee equal treatment and control, it just converges to it with higher probability as sample size increases, assuming a bunch of stuff .

You can get better performance just by manually dividing the two groups

[Nebula Of City @NebulaOfCity]

If you actually measure the core variables of many famous studies (age, gender, social stuff), you might be surprised how skewed the T and C frequently are in practice.