didericis

I’m a “biological optimist” and think a lot of what we consider “bad” can be and often is at least somewhat adaptive. Even things like aging and death; there are ways to age and die adaptively to give room to the next generation, to allow societal renewal/avoid sclerosis, etc. My understanding is that aging is thought to be in part an evolutionary adaptation against cancer, and a kind of pattern of creative destruction we see all over the place in both nature and society. There are also maladaptive tragic ways to age without remembering/learning how to live in your context, and maladaptive things like premature aging, disease, etc.

All adaptation is contextual. This picture is probably familiar to a lot of people, and is a perfect illustration of what I’m talking about: https://www.boredpanda.com/athlete-body-types-comparison-how...

There is no universal “better”. At any age, in any environment, in any physical condition, there are contextually appropriate optimums, and always two categories of direction: 1) using whatever tools are at your disposal to adapt as much as you can while accepting (and testing) limitations 2) ignoring what tools you do and don’t have and either giving up or trying to get the world to adapt to you.

The first direction is always better.

I highly recommend lap swimming as a general physical/mental sharpening exercise you can do with virtually zero joint or other injury risk your whole life.

I had the good fortune of being coached for a brief period by this guy: https://magazine.wsu.edu/2009/08/11/channel-swimmer/ Having that kind of an example was awesome.

The downside of swimming is trekking out to a pool and finding a time when lap swim lanes are open, and climbing the learning barrier if you’re completely new. In my experience, a lot swimmers also tend to be instinctually quiet, standoffish and very competitive, which can be very intimidating if you’re going for a lap swim as a newbie. Especially if it’s crowded and someone wants to lane share.

But there’s almost always a time slot somewhere when it’s virtually empty, or a friendly lane sharing buddy who’s relatively new/not doing flip turns and stuff, or a friendly expert who’s casual and happy to give learners a wide berth. A local YMCA or one in a neighboring area is usually the most friendly place to start swimming, but I think a decent number of big commercial gyms have pools too, and universities/schools sometimes have public lap swim times as well. And if you’re completely new/never learned to swim, lesson instructors often LOVE private adult lessons because they actually listen/you don’t have to additionally do baby sitting/kid wrangling. I lifeguarded for a bit and taught swimming, and liked teaching kids too, but adults were so much more relaxing.

It’s absolutely worth climbing those barriers. Especially if you get one of these: https://www.underwateraudio.com/products/waterproof-ipod-shu...

I haven’t really checked the landscape post covid, though. I’m guessing a lot of places with pools got hit really hard. All the more reason to get out and find/support one, they’re a really awesome community resource.

Speaking of which, I’ve turned into a stressed out couch potato/haven’t been to a pool since covid, I should take my own advice and get back out there.

If you’re validating client url state against server session state that kind of defeats the purpose of trying to put all the state in the url.

You’re technically correct/sounds like you agree it’s still a bad idea, even if there’s a way to duplicate certain state for validation and make it work. Staying shackled to the earth via stupid rules of thumb like minimizing client state in banking apps does keep you from flying straight into the sun, though, which is nice.

Compression is one of the closest things to real life magic I’m aware of. Especially now that people have basically compressed the entire internet into a couple gig AI models.

And then when I think about biology/genetics and what kind of unfathomably weird and convoluted compression is going on there, and then think about whatever it is our brain is doing… and then when I think about how like all of our perception is probably some kind of weird inescapable compressed representation of true reality…

It’s all magic. Real life magic.

I dream of a world where there’s some kind of ipfs like thing for storing state in some sort of distributed commons, all while maintaining user privacy/allowing for state to expire if untouched, and monetized in a balanced way to incentivize hosting without discouraging consumer adoption too much/trending towards gouging.

We’re so close/all the pieces needed seem to exist, but getting something like that off the ground is super difficult.

The opposite is also true. People who love seeking out alien perspectives have a virtually unlimited amount of ability to find and commingle with whatever community they can think of online.

I think the optimism of the early internet was due to a dominance of users with that kind of personality. It wasn’t apparent to that type of person that others are not excited by that perspective hopping opportunity and not only would not take advantage of it once being online was more accessible, but would actively use the same technology to further silo and entrench themselves.

There’s a more practical and tragic reason to desire a “one true method”.

Imagine spending your whole life working under method X. You study its precepts, gain practical knowledge, go out into the field and test your assumptions. Your desire is to end human suffering using method X, and build a better world where good triumphs over evil and everyone can pursue their bliss.

But you must sacrifice in order for method X to work. The precepts demand you honor the gods. If you are an Aztec, you end up with pyramids covered in blood and tens of thousands dead, but are still woefully unprepared for disease and Europeans, your people suffer immensely and are virtually extinguished, and you die realizing your method was wrong and your sacrifices were in vain.

You can avoid that with a meta method that helps you select proper methods.

The amount of human suffering that can be avoided if people are able to distinguish the effectiveness of method X from method Y is extreme. That is where the hard earned victories of modernity come from. Science is the preeminent comparative meta method that identifies which methods are most effective at alleviating human suffering and people are right to uphold it (and to distinguish it from scientism/confusion with scientific bureaucracy)

There is no universal, rule based, propositional method of betterment, but to give up on the idea of any method being objectively better than another is to give up on the idea of meta negotiation and the pursuit of universal peace and prosperity. The pursuit is worthwhile, even if it may not be fully achieved in billions or trillions of years and may be more of an art than a system of computation.

The fact that we’re able to talk openly about the unaccountability of the NSA, CIA, FISA courts and Gitmo is a form of accountability that does not exist in China.

I do not condone the actions of all of the US institutions and believe many are largely unaccountable, but there is still far more accountability in our least accountable institutions than any in China. Culture within institutions, the populace at large from which institutions hire from, and the history of institutions is a very important but subtle factor that westerners take for granted and is not identical between countries.

The fact that we even call US actions war crimes is another form of accountability China lacks. China forcibly annexed Tibet. China has concentration camps in inner Mongolia. China massacred it’s own people during the Mao era and is run by the same party today. China releases dams upstream of cities and drowns its own people. China committed war crimes in its support of North Korea during the Korean War, and Vietnam in the Vietnam War. China turned hundreds of its college students into mincemeat in Tiananmen square.

These are not on the radar of most people in China the same way Gitmo and Iraq and Vietnam are in the US because the CCP has a level of unaccountability westerners do not understand. Their stranglehold on media and public perception is extreme, and they have been exerting influence on western social media like reddit, youtube, and facebook, as well as google search. They do so through virtually all corporations that want access to Chinese labor and the Chinese market.

The ramifications of this are severe. The lockdown policies that have destroyed a lot of the global economy were pushed on the world by pressure from China (and our own fear and incompetence), despite existing pandemic plans in western nations that were far more sane, more humane, and less destructive.

They were able to get us to discard our own pandemic plans from our own institutions because of the influence they have been gaining. China exerted influence over the WHO recommendations, and was very aggressive/in PR panic mode when the virus first escaped. Our own internal division is largely our own doing, but they were extremely aggressive in the early phases of the pandemic online and greatly exacerbated western problems precisely because they knew what talking points to copy and paste for maximum division and lack of consensus.

Giving them your data and more ability to influence you and people that fit your demographics online is incredibly naive and foolish, no matter how much you may dislike your own government or how valid that dislike may be.

I wonder how difficult it would be to extract this kind of evolutionary history from LLMs.

> If I was forced to pick one government to share my secrets with, it would be the Chinese, because there’s nothing they can do about it.

What makes you so sure about that?

I worry about China because there’s no internal checks to prevent them from doing anything.

Western governments and allies have a long culture of court systems and thinking about balancing constituent needs. That is eroding and becoming more dangerous to the extent western leaders are envious of dictatorial powers and trying to emulate Chinese totalitarianism, but there is a lot of institutional and cultural bulwark against it.

Any powerful totalitarian country should worry people. People underestimate the level of covert aggression in all facets of foreign involvement in regimes with no internal accountability.

The more I understand about how many risks and problems are involved in running an organization as large and with as high a social profile as Meta, with all the accrued obligations and entrenched loyalties and overpromising managers that glom on over time, all while building new unproven technological infrastructure, the more sympathy I have for how blurry the line between corruption and implosion is.

I don't think anybody trusts Facebook at this point, and the idea of being forced into a pay for play virtual world controlled by a single company is understandably scary and deserving of criticism, but at the same time I've done social VR and I truly and honestly believe it's a vast improvement over any other existing kind of remote interaction. I actually get and believe in the vision, it just seems like the execution is really bloated/bad and they tried to way over abstract things/make things way too big way too early. I could see how spending that amount of money would be easy to do if you're trying to create universal abstractions for making, selling, creating, stitching services for an interactive set of 3D environments together that is easy for people to make "apps" for, building data centers, accounting for latency, building out supply chains, subsiding hardware, doing R&D on tracking, doing social research, and dealing with entrenched employees and middle managers accustomed to "the cushy tech worker lifestyle".

You really have to try social VR to understand how much better it is than video chatting or text channels. There's absolutely something to it. There's so much more information in the way people interact about what their intentions are, how they feel about what you're saying, what they're interested in, etc when you're interacting in VR with a 3D avatar.

All that being said, the virtual real estate and clear corporate "branding" angle is really gross, regardless of whether the money gets shady/you could really call that "corrupt". I prefer the wide range in quality but genuine effort and creativeness you get with a free and open ecosystem like VRChat.

There's a company named Thorcon building a modern nuclear salt reactor in Indonesia that's actually doing it in what seems like a legitimately safe, scalable, cheap, easily reproducible way using shipyard/ship manufacturing techniques.

https://www.youtube.com/watch?v=_qtJE7VmyoU

What people describe as "tech addiction" is I think better described as "consumer product addiction".

Technology has created an unprecedented opportunity for us to turn the world into a human friendly imaginative playground.

That opportunity is squandered by people who think in the short term about control and money and status and grievance.

Call me what you will, but I have a hard time watching old videos like this and not getting emotional -> https://youtu.be/Uz__bJTlOjk?t=37 Great engineers using the most advanced technology of the day to make childhood fantasies a reality is one of the most beautiful uses of our time on Earth I can think of. Tech could make the world the ultimate fantasy adventure. We have so much capacity to alleviate material want and create amazing dreamworlds and make people happy, but laziness, greed, ego, jealousy, and resentment all get in the way.

Technological complexity obfuscates and entrenches a lot of existing problems and creates lots of room for abuse, but at root most modern problems are not fundamentally tech related. They're rooted in our difficulties cooperating with each other.

I strongly disagree.

Humans have perceptual systems we can never fully understand for the same reasons no mathematical system can ever be provably consistent and complete. We cannot prove the reliability and accuracy of our perception with our perception.

The only thing which suggests the reliability of our perception is our existence. The better ways of perceiving make a better map of reality that makes persistence more likely. Our ability to manipulate reality and achieve desired outcomes is what distinguishes good perception from bad perception.

If data directed by human perception is fed into these systems, they have an amazing ability to condense and organize accurate/good faith but relatively unstructured knowledge that is entered into them. They are and will remain extremely useful because of that ability.

But they do not have access to reality because they have not been grown from it through evolution. That means that fundamentally they have no error correcting beyond human input. As systems become increasingly unintelligible due to increasing the scale of the data, these systems are going to become more and more disconnected from reality, and less accurate.

Think of how nearly every financial disaster occurs despite increasingly sophisticated economic models that build off of more and more data. As you get more and more abstraction needed to handle more and more data, you get more and more error.

There is a reason biological systems tap out at a certain size, large organizations decay over time, most animals reproduce instead of live forever. Errors in large complex systems are what nature has been fighting for billions of years, and tend to compound in subtle and pernicious ways.

Imagine a world in which AI systems are not fed carefully categorized human data, but are operating in an internet in which 5% is AI data. Then 15%. Then 50%. Then 75%. Then what human data there is gets influenced by AI content and humans doubting reality based categorizations because of social pressure/because AI is perceived to be better. Very soon you get self referential systems of AI data feeding AI and further and further distance from original source perception and categorization. Self referential group think is disastrous enough when only humans are involved. If you add machines which you cannot appeal to and are entirely deferential to statistical majorities, which then become even more entrenched self referential statistical majorities, you very quickly become entirely disconnected from any notion of reality.

The medieval church used to be much more about esoteric doctrine and strange theological reasoning than I think most people realize. It was both complicated and nonsensical; different traditions had different levels of complexity, and many were notoriously convoluted. Byzantine life centered around theology and the overcomplicated bureaucratic state dependent on it to such an extent the word “byzantine” became synonymous with impossible to reason about complexity. While the way in which AI works involves much more rigor and empiricism and has wildly different principles at its foundations and observable results, I’d argue it’s similarly impossible to reason about. What brought about the scientific revolution was a rigorous deference to simplification wherever possible. The convoluted orbits with Earth at the center “worked”, but were much more complex. Our deference to complex convoluted statistical models where the underlying natural rules and phenomenon at play aren’t explicitly identified seems like a step backwards in terms of creating more actual knowledge to me (although the practical utility and the opportunities to refine outputs into new knowledge discoveries are massive)

And I realize what I was replying to was a hypothetical/not a serious proposition right now, but people seem very eager to push towards a future in which we defer to more advanced versions of these things without much additional thought. I think that’d be a major mistake.

I'd go further and say most major modern day problems, of which ludicrous education expenses are one of the best examples, are problems that come from perverse monetization incentives and increasing disconnect between needs, effort and reward stemming from societal complexity and abstraction, and that technology is likely to make a number of those problems worse.

We've figured out how to increase physical goods and enable efficient use of scarce resources that can't be infinitely replicated incredibly well, despite a huge range of different emotions, motivations, and conflicts about how those physical resources should be allocated. I'd argue that's mostly because signals related to physical goods are relatively clear, direct, and measurable.

Things like "knowledge" and "skill training" aren't as clear, direct, and measurable, and when you walk up the hierarchy of abstraction, it's a lot more fuzzy and difficult to measure what you're actually getting as a consumer, what you're delivering as a producer, and what the source of problems are, especially if there's a major time delay and variability for payoff. That's the case with modern education, and also applies to "investment payoff in quality and effectiveness of health care", "investment payoff in quality and effectiveness of living space", "investment payoff in quality and effectiveness of public outreach", etc. It makes proper negotiation very difficult and leaves a lot of room for confusion and abuse.

One possible solution to this, which seems impossible/idk how this would work in the modern era, is to try to return to a system where a whole town would work more like a single unit so as to try to reduce some of the sprawl and complexity. Earlier in America there'd be a town known for making shoes, a town known for making lightbulbs, a town known for making steel, etc. There'd be a rallying, measurable "physical thing" the town could help use to calibrate whether it's healthy/allocating resources correctly. If the town isn't working as a cohesive unit, the central "thing", like lightbulbs, would get measurably worse. High school education is bad? Fewer competent lightbulb workers, fewer lightbulbs produced, clear signal. Healthcare isn't working? More sick days, lower moral, fewer lightbulbs produced, clear signal. Everyone in the town would have a better map of the intricacies of what's working and not working in that smaller system than they could possibly have in a global/gargantuan system, and they'd have a relatively objective measuring stick to determine whether they're working effectively as a unit.

I’m quite sympathetic to that line of thinking. Your language is suggestive of an interview on Lex Fridman’s podcast/am guessing you’ve listened to it. Nietzche also basically recognized this over a century ago, though he put it in different terms/couldn’t articulate in as much mathematical detail as evolutionary game theorists allegedly can.

That’s actually precisely why I think evolutionarily adapted embodied perceptual systems need to be a part of any true knowledge: reality is so incredibly complex that we don’t know what our perception is or is not doing. We just know it’s doing something reality based because we exist. Perceptual models don’t need to correspond with reality like actual embodied perception (and corresponding cognition)

I don’t buy the argument that the correspondence with objective reality in evolved perceptual systems is zero as is claimed in that interview, simply because we exist. I think that conclusion is an artifact of precisely the same kind of over abstraction and non-correspondence that invalidates language models as knowledge systems. Our existence means our perception corresponds with at least some kind of compatibility with and ability to navigate through reality.

A system which does not have that embodied evolutionary history with millennia of required correspondence with reality, simplified or not, does not have all of the essential embedded context and understanding of complexity that goes into our sense making. Science is about discovering more of that hidden context, and perhaps these language models can help identify patterns we haven’t seen yet that do in fact correspond with reality. But the model can’t test it’s own correspondence.

This is a deep topic that I don’t feel fully qualified to speak precisely about, but the key point is that knowledge isn’t about mapping symbols to other symbols like what language models do, it’s about mapping symbols/mental representations to reality.

I think knowledge is in fact a good word for ‘sensical’ language, but the point most people like me are trying to make when talking like this is that language models can’t determine what is ‘sensical’ language.

The amazing thing about these newer models is that they seem to get remarkably close to properly understanding real world concepts just through symbol mapping on massive amounts of data, but that’s highly dependent on having being fed sensical data where that mapping of symbols to reality was already done.

I’m trying to argue that knowledge only really exists in the data feeding step, when we map symbols to reality.

What the language models do is create a map of symbols. But they mix things which are knowledge with things that are not, and cannot determine what about their output and their mapping is actually valid knowledge.

I think most of why this is has to do with the nearly endless amount of embodied physical and historical context that goes into our own sense making. A surprisingly large degree seems to be embedded in language, but there’s a lot more that is not.

EDIT: I rambled, sorry; direct answer to your question about what I’d call the property that makes these language models useful is “relatively high degree of correspondence with reality”. What I consider “knowledge” is directly obtained by something which can interact with reality/directly maps reality to some representation.

> isn’t ‘holding a representation of a fact or relation’ basically what we call ‘knowing’ something?

No. Knowledge is fundamentally and inextricably tied to our experience of reality. The billions of years of evolution that created our perceptual framework has created a bridge to understanding reality machines don’t have.

I’d argue a chicken brain has a far better greater amount of true knowledge than any of these language models.

That does not diminish the utility and sophistication of these language models. But they are not knowledge. They are a disconnected meta map of what we have written down.

We calibrate these language models to knowledge because our language usually corresponds to things we know.

If the internet was mostly populated with sentences like “a druid flew into houston, cats pool into mud vampires wealth umpire onion buy now saxophone”, a language model mapping it all would be useless.

This is an extremely important point.

These are very large, very complex fuzzy language maps, and are best used like a crystal ball for autocompletion suggestions, not definitive answers.

I’m unsurprised but also very disappointed by people’s desire to give these fuzzy answers final word. The desire to cede final authority on truth to these machines just because they’re giant and impressive is incredibly misguided and a major step backwards.

These machines are literal embodiments of group think.

They can only add value if people understand their limitations. If people get carried away with them and start assuming they’re authoritative we’re in for serious trouble.

> just takes a sufficiently large models word for it

The renaissance and enlightenment happened because we stopped doing this. The church was that era’s big model, and doctrine was similarly impenetrably complex and indecipherable. You’d just take the church’s word for something.

We’re going backwards.

Is that true of the upper atmosphere? Wouldn’t it clump there due to gravity?

And if it doesn’t clump there, does it clump anywhere, or is there some reason it’s naturally diffuse?

Is there a possible future in which lowered cost to orbit and helium scarcity makes it economically viable to harvest helium in giant sacs in space, and then use “gas stations” floating in the upper atmosphere at different buoyancies to shuttle it down to airships?

I have no idea how many problems there are with that, or how diffuse helium is/how hard it would be to harvest from space, but floating airship stations like Bespin would be awesome.

The article focuses on human overrides, but I think the more obvious and gaping security issue is lack of any significant ability to verify output correctness whether it’s intentionally gamed or not.

I predict nearly all of the upcoming LLM products will end up being fancy autocomplete suggestions a user will then have to feed into a more constrained system with some sort of manual confirmation/tweaking.

Whether or not those quoted are actually representative or cherry picked, and whether they’re as opposed to internal practices as is suggested or simply think there’s room for improvement requires a trusted investigator to determine.

Quite frankly, the vast majority of the journalistic enterprise has been so heavily discredited and so abused for hit pieces and marketing that I don’t trust that the reporting reflects the internal reality they’re depicting.

The timing, the citation of multiple federal boards, citation of a competitor, and the group think Eye of Sauron that used to be aligned against Trump seemingly gearing up to go against Musk makes me highly skeptical of the legitimacy of the complaints.

This is a tragic situation, because it is now very difficult to distinguish between the types of people opposed to high pressure pursuits of excellence in general, regardless of how effective and voluntarily pursued and in good faith they are, and people who are legitimately waving a red flag about a serious problem.

This does not strike me as a serious problem, this strikes me as a hit piece.

The beneficial purposes are improving prosthetics.

And the only thing that makes those goals attainable and products better is experimentation.

If they are needlessly experimenting and making no progress, that’d be tragic, and comparable to existing tragedies that happen in labs all over the world.

The timing, intended audience, and flavor of the criticism suggests this is much less about any ethical breaches and much more about reigning in someone who is perceived to be breaking out of bureaucratic control.

I think the neuroscientists Musk has hired are scientists, and yes, I think he’s competent at hiring competent people.

The specific example I gave was bad, yes.

Pork might have been a better food to make this comparison with, and in this particular example I get that monkeys being closer to humans is a lot of what makes this controversial, but the phenomenon I’m describing is still pertinent. There are undoubtably other experiments using monkeys in far less scrupulous ways for far less beneficial purposes, but of course Neuralink gets the spotlight due to Musk.

I used to think a lot of ostensibly moral decision making which had a net detrimental effect was something that would change if people better understood the full picture and where moral outrage would be most impactful, but I’ve seen envy dominate decision making over and over enough to learn that a lot of people are in fact quite content to harm everyone on net if it means the outliers are harmed more.

The saddest reality of human nature that I don’t think I’ll ever come to terms with is the aggressive, unrelenting desire to hold the most competent and boundary pushing people to intentionally impossible standards.

A million people could torture a thousand chickens each over the course of a lifetime and throw away the meat, overeat and make themselves a diabetic, and do all kinds of things that actively make the sacrifice of that chicken meaningless, and one scientist who tries to maximize the gain of that sacrifice and better humanity by studying it’s brain is hounded and harassed not because he’s worse than the average person, but because he’s better.