LessWrong 2.0 Reader

Based on about a dozen of Said's comments I read I don't expect them to update on what I'm gonna write. But I wanted to formulate my observations, interpretations, and beliefs based on their comments anyway. Mostly for myself and if it's of value to other people, even better (which Said actually supports in another comment 🙂).

• Said refuses to try and see the world via the glasses presented in the OP
  • In other words, Said refuses to inhabit Aella's frame
• Said denies the existence of the natural concept frame and denies any usefulness of it even if it were a mere fake concept
• It seems to me that Said is really confident about their frame and is signaling against inhabiting other people's frames

Most people usually aren’t onto anything good, so this, again, ought to be the default assumption.

• It seems to me that Said actually believes there is no value in inhabiting other people's frames

This seems bad, actually. It seems to me like a sign of insecurity and unjustified submission. I, for one, have no interest in having my conversation partners signal that they’re vulnerable to me (nor have I any interest in signaling to that I’m vulnerable to them).

Everyone has vulnerabilities. Showing them and thus becoming vulnerable doesn't signal insecurity or submission, actually the opposite. It requires high self-confidence (self-acceptance?) and signals openness and honesty to the other person. The benefit is that it leads to significantly deeper interactions.

And the benefit of inhabiting another one's frame? If I use the "camera position and orientation" definition of a frame mentioned by Vaniver, inhabiting other person's frame allows you to see things that may be occluded from your point of view and thus give you new evidence. The least it can give you is a new interpretation of data that you gathered yourself. But it can possibly introduce genuinely new evidence to you, because frames serve as lenses and by making you focus on one thing they also make you subconsciously ignore other things.

If you have access to the episode rewards, you should be able to train an ensemble of NNs using bayes + MCMC, with the final reward as output and the entire episode as input. Maybe using something like this: http://people.ee.duke.edu/~lcarin/sgnht-4.pdf

This get's a lot more difficult if you're trying to directly learn behaviour from rewards or vise-versa because now you need to make assumptions to derive "P(behaviour | reward)" or "P(reward | behaviour)".

What we'd want is some neural-net style design that generates the coin reward and the move-right reward just from the game data, without any previous knowledge of the setting.

So you're looking for curriculum design/exploration in meta-reinforcement-learning? Something like Enhanced POET/PLR/REPAIRED but where it's not just moving-right but a complicated environment with arbitrary reward functions (eg. using randomly initialized CNNs to map state to 'reward')? Or would hindsight or successor methods count as they relabel rewards for executed trajectories? Would relatively complex generative games like Alchemy or LIGHT count? Self-play, like robotics self-play?

But new algorithms also don't work well on old hardware. That's evidence in favor of Paul's view that much software work is adapting to exploit new hardware scales.

Real propose of Peresvet was published only 2 days ago https://lenta.ru/news/2021/12/02/peresvet/

Events in Kriorus are beyond crazy, but it looks like that more people learned about cryonics then before. Wired is preparing a piece and even Netflix showed interest (according to rumours) in making TV series about the story. So people are still coming for cryopreservations, and there was a few last fall in Valeria facility. 

Zirkon hypersonic missile was tested a couple of times recently, but no updates on Burevestnik. 

Getting from here to there is always the tricky part with coordination problems, isn't it? I do have some (quite speculative) ideas on that, but I don't see human society organizing itself in this way on its own for at least a few centuries given current political and economic trends, which is why I postulated a cooperative ASI.

So assuming that either an aligned ASI has taken over (I have some ideas on robust alignment, too, but that's out of scope here) or political and economic forces (and infrastructure) have finally pushed humanity past a certain social phase transition, I see humanity undergoing an organizational shift much like what happened with the evolution of multicellularity and eusociality. This would look at first mostly the same as today, except that national borders have become mostly irrelevant due to advances in transportation and communication infrastructure. Basically, imagine the world's cities and highways becoming something like the vascular system of dicots or the closed circulatory system of vertebrates, with the regions enclosed by network circuits acting as de facto states (or organs/tissues, to continue the biological analogy). Major cities and the regions along the highways that connect them become the de facto arbiters of international policy, while the major cities and highways within each region become the arbiters of regional policy, and so on in a hierarchically embedded manner.

Within this structure, enclosed regions would act as hierarchically embedded communities that end up performing a division of labor for the global network, just as organs divide labor for the body (or like tissues divide labor within an organ, or cells within a tissue, or organelles within a cell, if you're looking within regions). Basically, the transportation/communication/etc. network edges would come to act as Markov blankets for the regions they encapsulate, and this organization would extend hierarchically, just like in biological systems, down to the level of local communities. (Ideally, each community would become locally self-sufficient, with broader networks taking on a more modulatory role, but that's another discussion.)

Anyway, once this point is reached, or even as the transition is underway, I see the ASI and/or social pressures facilitating the movement of people toward communities of shared values and beliefs (i.e., shared narratives, or at least minimally conflicting narratives), much like in Scott Alexander's Archipelago. Each person or family unit should move so as to minimize their displacement while maximizing the marginal improvement they could make to their new community (and the marginal benefit they could receive from the new community).

In the system that emerges, stories would become something of a commodity, arising within communities as shared narratives that assign social roles and teach values and lessons (just like the campfire legends of ancient hunter-gatherer societies). Stories with more universal resonance would propagate up hierarchical layers of the global network and then get disseminated top-down toward other local communities within the broader regions. This would provide a narrative-synchronization effect at high levels and across adjacent regions while also allowing for local variations. The status games / moralities of the international level would eventually attain a more "liberal" flavor, while those at more local levels could be more "conservative" in nature.

Sorry, that was long. And it probably involved more idealizational fantasy than rational prediction of future trends. But I have a hunch that something like this could work

FWIW I agree that bit also rang hollow to me--my sense was also that neurons are basically as energy-efficient as you can get--but by "computational efficiency" one means something like "amount of energy expended to achieve a computational result."

For example, imagine multiplying two four-digit numbers in your head vs. in a calculator. Each transistor operation in the calculator will be much more expensive than each neuron spike, however the calculator needs many fewer transistor operations than the brain needs neuron spikes, because the calculator is optimized to efficiently compute those sorts of multiplications whereas the brain needs to expensively emulate the calculator. Overall the calculator will spend fewer joules than the brain will. 

Makes sense:

• Tech brings in a lot of per-capita income
• Tech brings in utopian, disruptive voters
• One of the most progressive states

Neuron computation is electro-chemical through voltage gated ion channels. If the voltage is at or below the Landauer voltage, then ion motion through the gate is pure noise. As the voltage climbs above the Landauer limit, you start to get meaningful probabilistic state transitions (error rate below 50%) in reasonable time; you can then implement analog computation using many such unreliable carriers reducing error/noise through central limit binomial.

'Pure' chemical computation is protein machinery.  Biology evolved voltage based signaling for high speed longer distance communication/computation.

There’s an optimistic way to describe the result of these trends: today, you can’t start a cult. Forty years ago, people were more open to the idea that not all knowledge was widely known.

This doesn't seem to have aged well in light of the rampant spread of misinformation and conspiracy theories on social media (especially Facebook!)