Mark Gubrud

It was 76.4 M to 73.7 M. Neither number comes close to the historic 81.3 M who voted to fire Trump in 2020.

Fascism only wins if we believe the Big Spin and if we don't fight hard from right now until MAGA is defeated.
bsky.app/profile/nvre...

👇Aftermath of today's Russian attack on Ukraine's Odesa region. At least 2 people killed, per local governor. (source: t.me/dsns_telegram)

Threads's fatal problem is that it is owned by a psychopathic zillionaire only slightly less virulent than Elmo.

Story of how 2 accidental nursing errors led anti-vax RFK Jr to end up killing 80 people in Samoa.

The country did not shift. Blue just didn't get its voters out like in 2020. And that's basically because Biden sucks, and Harris didn't distance herself from him. Probably couldn't. But that's why only 73M showed up to vote for her while 81 M voted to fire Trump in 2020.
bsky.app/profile/nvre...

I am blocking you now and I do hope you do feel censored.

This is basically Lysenko, Great Leap Forward stuff. This is dictator stuff. Dictators do big, insane things, ranging from sparrow holocausts to human ones. bsky.app/profile/nvre...

This is basically Great Leap Forward stuff.

This was the ambition of decoherence theory I believe but it remains an incomplete project, or so is my understanding. Einselection seems to explain classicality but collapse itself remains unexplained.

I don't know how you recover any coherent notion of branching from this.

Rather, I think that a web of elementary interactions that spreads at the speed of light in some DoF (slower in others) collapses all wave functions into one local, collective reality.

Every molecular collision must entail some uncertainty, which combines with chaos to ensure total alteration of futures at human scale some time further, not only from each interaction, but from all of them together at once, so that the background of classical & quantum noise is truly impenetrable.

'What will GPT look like in 2030'
An incomplete but not inaccurate preview of AGI.
bounded-regret.ghost.io/what-will-gp...

blocked

Blame Biden. Blame Biden and everything he stands for: the mediocrity, the compromise, the delusion, the absence. The support for genocide. The half-halfway leadership on climate. The continuing misery of covid and no leadership toward mitigation of airborne infectious disease. The nothing. Biden.

More have died of covid under Biden than under Trump. Biden's public health policy has been thoroughly miserable, an utter dereliction of duty for dim-witted political convenience.

Harris lost because Biden sucks. In many ways. In all the ways that he sucks. That's why she lost. Because he sucks.

Way to open the door for MAGA. I mean, don't. Just don't. It's hard enough to pass any Constitutional amendment in normal times. We need to resist MAGA's attempt to establish a dictatorship and protect our Constitution. Not open it up for rewriting by Trump & Elmo.

Right, and that would be a continuum. I know that some people claim to know how to count worlds in MWI. I don't know if there is any consensus about this, or whether it requires some additional assumptions.

Does it make sense to speak of being "constantly thrown to random places within the timeline"? This seems to require the assumption of a kind of super-time, which isn't the normal "timeline" but rather the sequence of parts of it that we would be visiting.

I can't make sense of the notion that time isn't real. But I also can't make sense of the notion that it is flowing. Rather, things exist, and events happen, causing things to change, and that is what we call the flow of time, but time isn't doing anything.
+

But QM is not the same as saying things just happen without cause. If a nucleus decays, that is because there was an unstable nucleus. The probability of decay in any time interval can be measured or calculated and known with high precision.

But if you look at what happens before that, you see semantic computation using the vector representation of meaning. This is learned from the prediction task, but it is not in the form of statistics, and its structure will be the structure of meaning, knowledge, features, not of words on paper.

Again, LLMs do not work by compiling or learning statistics of words. That is a denialist canard.

If you look at the last stage of the neural computation, out pops something that can be interpreted as a probability estimate over words, or tokens, actually.
+

So, the absoute void is not a possible alternative to the existence of... something. And perhaps one day we will have enough understanding in physics to be able to say that the kind of universe we live in is the only kind that is possible, that actually makes sense.

Indeed. But I think I have an answer to that one. Try to consider the alternative: nothing existing. At all. Ever. Anywhere. We may think we can imagine this, but then we would be there, beholding this void. How could it, on its own, *be* so?
+

Seems to me that time as I experience it is the happening, is the decision of what happens out of the range of possibilities. This seems the only way to construct the notion of a now. (Which is necessarily only a local construct.)

Why is it so hard to accept (constrained) randomness?

Interesting article there by Ball but I have never thought a modified Schrodinger eqn or any kind of "physical collapse" mechanism was likely. In contrast with Everett, who wants the Schrod eq to be Reality, I think physical reality is fundamental, math only an approximate & partial description.

Why would randomness be more in need of explanation than determinism? It is in fact what we experience. What is better about a clockwork universe or multiverse? If everything is predictable, isn't that boring and lifeless? If everything always happens, does anything ever happen?

It is not accurate, as many people think, that what LLMs have done is compile statistics of text and that it is generating the next word with no understanding, with no semantic representation. That would actually be impossible (intractable).

Belkin says "the model estimates the conditional probability... for each next word." OK, that's formally true, but that "estimate" is just the output of a softmax function, whose input is the output of a neural computation which is just tuned to produce the most accurate predictions.
+

Agree and have said the same many times. But GPT is NOT "simple statistics." It is a neural model that is trained to predict language as successfully as it can, and this success entails building an internal semantic model, a model of meaning, not of words.