darkmighty

Interesting. Apparently he needs to assume a particular multiverse theory to prove it. While I don't object to those on principle, I don't believe they're needed to prove heuristic, good enough versions of Ockham’s Razor that work in the real world (albeit without guarantees), based on the arguments outlined on the previous comments.

> The latter is important because some theories trade off between these two: e.g. a multiverse theory might have simple equations ("every possibility happens somewhere") but require very precise "coordinates" to pin-point the actual possibility that we observe.

I think this is an important observation that's quite obvious for ML researchers et al but again seems to escape current physics discussions. An example is the endless drama about "Fine tuning": if your new theory requires many less bits for equation description, that it requires fine tuning is irrelevant as long as the additional model parameter precision uses less bits -- then it should be the preferred candidate.

W.r.t. [computational] multiverse theories (and variants such as Tegmark's MUH, Schimidhuber's, and others), I do believe they're an inevitable progression of physics/philosophy. I just think it's a bit pretentious to have any certain about a particular flavor. I feel there's still much philosophical and mathematical ground to be covered; it tests the limits of our imagination. It seriously feels like a very important step for humanity at large though -- finally approaching metaphysical theories that actually make sense, and explain the basis of much about humanity, existence, ethics, etc. I think it's an important void to be filled after the decline of religion, hopefully in coonjunction with the spread of humanism.

I guess that's a good overview. It's related to the phenomenon of overfitting in machine learning: you can always easily find a sufficiently complex (or complicated, large if you prefer) theory fitting all data points. Because this theory simply encodes each observed case (including progressive sophistications of encoding), you naturally expect it to fail on unobserved cases -- it makes no effort at generalization. The simpler theories have a greater chance of generalization, they're more likely to be the "true" mechanism of the process that doesn't simply encode "edge cases" as you cited, and thus are more likely to also work on unobserved data.

Honestly I haven't seen attempts at making this process more rigorous, when applied to physics. There's the large corpus of machine learning study which provides concrete results and even concrete comparison tools, but the times I've asked a physicist I've been dismissed; while it seems incredibly valuable in the face of lack or large cost of experimental data, which is quite relevant today.

Meh. Competitive race to dehumanization, ruthless proggressivism, are not physical laws. If there are sufficiently many actors unwilling to make accords, sure, it's inevitable. But there is enough structure and organizations, and interested humans, such that we can alter society to give a little more weight to basic human values than surrender to the endless march of domination.

We needn't halt it -- after all, if we can create a good desirable society, a billion more societies would be welcome.

But we have, and should continue to, what's truly valuable and worthwhile to pursue. An universe filled with automatons (even if scientific or mathematics god-like savants) is boring -- and I love mathematics -- but in general problem-solving ability is certainly not the whole story of cognitive experience, even general problem-solving ability. The life experience of Lee Sedol is more interesting than the synthetic "experience" of AlphaGo, even if it is superior at his life's work. If we had an AGI strictly geared towards solving technological problems the same might be said, even if ultimately other aspects of human experience are "wasteful", unnecessary for conquest and dominion.

Life can be material-constrained instead of energy-constrained (I know there's E=mc^2, but this may not be technologically feasible hence irrelevant). Environmental domination is more fundamental than energy source tapping imo (you can imagine life in systems that don't even have precise energy analogues, like conway's game of life) -- that is, dominating whatever the basic resources are (space, materials, energy, etc).

Tbf, those are trivially not mathematically equivalent (even taking profit instead of revenue). Suppose each contracted worker cost $1/yr in salaries, and there were 1000 contractors/1000 in-house employees. Then without including contractors, the profit per employee is (X-1000)/1000 ~ X/1000, while including it is (X-1000)/2000 ~ X/2000, much lower. Including or not contractors does have a potentially large difference.

> If GPS satellites were to cryptographically sign their signals, would that be sufficient to prevent spoofing of GPS signals (assuming the receiver of the signals treated the signing appropriately)?

Nope, classic replay attack case (just record the GPS signal and replay to the device at the desired location). You'd need a true time signal within the device, e.g. and atomic clock to make it work (so you'd authenticate the signed time against true time).

---

There is another way, however. If we assume the hardware is tamper-proof (unless drastically different methods are needed), then with strict timings we can device a challenge-response system that's immune to replay attacks due to relativity: simply transmit a signal A, have a known 3rd party (e.g. US government servers in cellphone towers) sign your signal Sig(A) and retransmit, and check that the delay matches the propagation delay you'd expect from the cellphone tower distance, plus the fixed (and immutable since it would be gov-controlled) processing delay. Your tamper-proof crypto-camera would record its location and whether it trusts the location. Using cellular signals is also better because GPS doesn't work indoors and is sensitive to interference.

Since we're adding a cellular connection to our device, it would also be a good idea to log its position on the state-controlled servers (again can be done with cryptographic safety assuming non-tampered device), along with some kind of intrusion detection system. As soon as it'd detect an attempt at tampering, it would relay this attempt to the servers, storing the intrusion and invalidating the authenticity subsequent recordings; a self-destruction attempt of the key would probably be also wise.

---

And now that I think about it, you'd probably want to put several keys/auths in the device, from different organizations -- not only governments. That way if the government authentication is positive but NGO's mismatch, you can suspect a government-backed forgery attempt (analogously vice-versa).

Systems of logic should not be hard coded into AIs (or AGIs) imo. The basic problems with logic is that it doesn't take uncertainty into account, and doesn't support the fact that almost every statement about the real world is only approximately true, or a useful model; it also usually can't handle inconsistencies very well.

It would be unable to prove or in fact have any useful facts about the real world. For example, consider the statement "The sky is blue.". It is not categorically true -- the sky becomes red near sunset, (depending on chrominance vs luminance interpretation) it gets black at night, may be other colors in other planets, etc. So the system would be stuck trying to convey all necessary conditions, might not be able to formulate simple thoughts or communicate well. It would almost certainly lack abstraction capabilities.

There are some logical systems that make slight modifications to logic to accommodate probabilities and uncertainty, but those still might be insufficient or incomplete.

The fundamental issue is that, while logic works and is useful for us, it is just a tool. The end goal of any real world agent is not logical correctness, it is achieving tasks, understanding/predicting its environment, communicating, etc. Because of the specificity of those environments and tasks, there isn't going to be a general method that is efficient in them. You need a evolutionary-like, or leaning system that can adapt to the problem with minimal underlying assumptions. The assumptions of logic that everything should be binned according to falsehoods, probabilities, etc, is ultimately not necessarily a good way to (for example) control a robot in a complex environment, make it assemble a product, play videogames, or even do mathematics. Perhaps they can be augmented with side models (again tools) allowing them to formulate subtasks as logical problems (although its not clear those approaches would be better than just training neural networks). Nothing can be set in stone, everything must be learned and modifiable.

So again the main problem with AGI is formulating good frameworks that 1) Have good, general goals aligned with what we expect from this AI; 2) Is able to adapt in almost every way to achieve this goal, without degrading its ability to recognize its rewards/goals.

This is such a hard problem that it occupies a significant chunk of even humans' lifetimes. We spend lots of time thinking about long term goals, which in part are dictated by society, in part dictated by biology, and which we reflect upon to conclude what we should do -- from small daily goals like eating or getting to work, to long term career advancing and relationships. Then we set to achieve those goals by learning and acting on the world around us. It does happen often that our goals degenerate, ultimately because of our ability to modify ourselves and set our own goals: the reward achieved from a drug is a degeneracy that overrides other long term goals; various addictions exploit loopholes in our reward system; we suffer from despair, depression or lack of motivation that can even override the basic biological rewards.

If we could not override basic biological rewards, setting our own goals, then we would fall to addictions and inability to live in complex environments which require complex unnatural goals related to careers, learning, and other abstract concepts.

And too much of this power can leave us in desperation, depression, meaninglessness.

So there is a constant balance and search for genuine rewards, and constant refinement of them. That is central to human existence, and to conscious experience in general.

> Turtles all the way down, I guess.

Indeed it is important in this case of course that this does not happen :) To see the increased reliability and P(glue failure)<P(single failure) you need to assure the glue systems are very simple and well built -- and preferably they need to be much smaller than the system you're protecting.

Another adequate expression to apply here is

"Who watches the watchmen?"

The answer again is the watchmen must watch themselves and be very reliable.

On this topic I recommend von Neumann's (the brilliant mathematician) "Computer and the brain" book, where he explores how computing systems can be reliably interconnected and how those failure probabilities interact. He was interested on how the brain could be so robust to failure -- don't worry there's no time spent speculating on how the brain works, instead he derives from first principles properties of reliable computing components, and possible reliable designs (the brain's unknown internal workings at the time, and now to a lesser extent, would follow as a special case). He used this same approach in analyzing the principles of life, where he came up with a self-replicating machine with a tape encoding of itself, predating the discovery of DNA -- it's a very inspiring and powerful approach. Unfortunately he could not complete 'Computer and the Brain', he was in declining health due to cancer and died while writing it. What was left is still very interesting imo. He is one of those giants whose shoulders we can sit on to peek over the horizon :)

The key part of redundancy is that your "redundancy glue"[1] must be significantly more reliable than each component, including its software and implementation -- because often the glue failing in isolation itself can cause outages. So the probability of failure was simply P(single failure); now for 2x parallel redundant systems it is P(single failure)^2 + P(glue failure). If P(single failure)^2 ~ 0, we need P(glue failure) < P(single failure), at the very least.

[1] i.e. the systems that interconnect the multiple redundant system, detect failures, redirect traffic, etc.

I think that's a fair questioning. Transistor scaling seems to have run its course, sitting right at the atomic scale (but not beyond -- hence it would be probably undisturbed by subatomic discoveries).

At least there are plenty of biological examples that we haven't gotten close to replicating technologically -- while we have substitutes for almost every biological functionality, there are areas in the parameter space where the bio versions comfortably lead. The brain of course is an example, leading in efficiency and capabilities; but there are many others like compact self-replication, self-repair of structures, micro and nanomachinery (exhibited in cells), and so much more. None of that of course requires advancing physics.

Even for long term technologies, I don't think a deeper understanding of physics would have much to contribute, unless we found something wild like a compound made from new particles. That isn't to say it's not worth pursuing; it just takes a different relation to society, similar to math and philosophy.

I can't comment much about Marx as I don't know the details of his claims or how they relate to implemented politics.

What I enjoyed very much was the claimed position of Raymond Aron's w.r.t. intellectualism:

"Aron’s generosity of spirit was a coefficient of his recognition that reality was complex, knowledge limited, and action essential. The leitmotif of Aron’s career was responsibility. He understood that political wisdom rests in the ability to choose the better course of action even when the best course is unavailable--which is always.

The subject of politics, Aristotle noted, is “the good life for man.” What constitutes the good life? Aron cannily reminds us that the more extravagant answers to this question are often the most malevolent. They promise everything. They tend to deliver misery and impoverishment."

I think this is a great thing to be reminded of, constantly -- in particular for intellectuals and scientists. It is almost too easy to be fooled in thinking the scientific principles or intellectual ideals will apply nicely to humans, the complexities of the real world; that we can be certain about what is truly good or the true nature of reality.

It's not that we shouldn't investigate political systems, moral systems, economic theory, metaphysics, etc; it's that we should be specially skeptic about grandiose claims that go against common sense or personal freedom.

And quoting the recent Aaronson AMA:

"(...) partly it's because I take it as almost an axiom of rationality that, if a metaphysical belief leads you to do "obviously insane" things with your life, then it's probably time to look for a better metaphysical belief. :-) (I wouldn't say the same about scientific or mathematical beliefs.)"

Library Genesis/SciHub[1], and its various access portals[2].

Honestly LibGen+SciHub is doing more for humanity than almost any organization I know of. It's the next generation complement (not replacement) of Wikipedia -- truly opening the entirety of Man's knowledge to anyone.

I don't oppose paying content creators, but I suspect the vast majority of people accessing this knowledge wouldn't do so under traditional publishing. This shows copyright is largely outdated, not serving society optimally as it should.

As for Bitcoin and blockchains... my opinion is that they are extremely useful for efforts like those (and other niches like totalitarian states), if not much else. Perhaps paradoxically I also (controversially) think they should be outlawed and/or heavily regulated (outlawed but as a lightweight non-criminal offense). The potential for theft and money laundering is too large otherwise.

In the limit where everyone used perfect secrecy coins, paying taxes would need to be largely voluntary. And I suppose we all know how well that would turn out.

So I think cryptocurrencies are meant to be underground. It will always be possible to acquire some under the censorship conditions they are really useful for. So restricting their usage mainly curbs trivialized laundering.

[1] https://sci-hub.nu/

[2] e.g. http://booksc.org/howtodonate.php, it does support amazon gift cards too but I find that inconvenient

Seems fairly "conceptually straightforward" (if not so easy): compress data (internal dependencies), use predictions (external dependencies), save states server-side, simply reduce garbage.

You're talking about launching a Mars colony, to build a massive industry, to provide materials for Earth-orbit systems... that looks like a massive complication to say the least. Yes, Earth launch costs are high -- but those exact launch costs would be necessary for almost every equipment put on Mars, at least until you have a large self-sustaining colony. Imagine the cost of power that needs solar panels brought from Earth, necessary tools, the costs of maintaining the crew, ... . $50/kg of steel produced under those conditions seems absurdly cheap.

And then to sustain this massive operation you need a massive demand. You'd need millions of tons of steel of demand to make the colossal investments (of perhaps billions/year) viable. And then the demand for the primary applications needs to be even larger. I don't know much about the applications you've cited, but I'd find it hard to believe you could get demand in tens or hundreds of billions per year. They look essentially like niche applications that can be fulfilled by a low-cost launch system. Remember that the FH costs about $1700/kg (and cost probably be cut in half or better with several billions in annual demand).

I.e. I'm not convinced. That said, I think under a very long term (perhaps by the end of the century), the objective of a self-sustaining colony is worthwhile.

Before economic viability of any enterprise, a massive amount of systems will need to be developed enabling manufacturing basic tools and materials in situ. It's basically reinvention of industry from scratch but with much more strict requirements and much thinner margins for sustenance. A first step would be probably local production of solar panels, food, structural materials, habitats. The basic elements of self-sustenance. Then gradually moving up the technology ladder until import dependence from Earth would be at a minimum. Only then it starts to become viable and thinkable to have small martian exports for orbital systems, tourism, culture and media (I imagine showcasing life on Mars would get quite an audience on Earth), perhaps low-bandwidth scientific computing. This will probably take centuries but is an important step for humanity. Sooner or later we have to get off this rock, there's an entire universe out there waiting for life...

Just use little inductive proofs. How many days lie between day 3 and day 37? Is it 37-3=34 or maybe 37-3-1=33?

Well, between day 1 and day 3 there is 1 day. By finite induction between day n>3 and day 1, there are n-2 days, and analogously between days n>k>1 there are n-k-1 days.

Actually I'm not so sure if much would change (w.r.t. performance) if they switched to a more democratic regime.

In the end I were always a bit scared/reluctant about the US often touted ultimate justification for democratic capitalism that it is simply more efficient -- "Cultivate freedom -- after all, in the end it's just more efficient!". What if... it isn't?

China is turning the tables on this argument quite a bit. Heavy handed government control, aggressive state-sponsored corporate espionage, citizens surveillance and strict punishment of detractors, exploitation of workers, all combined with hand-picked aspects of market economy to lubricate the system. China is designing itself as an efficient economic machine with minimal regard for personal freedom or expression.

It's still unclear whether China will be able to achieve technological parity, but it's quite probable it will overtake the US as the foremost economy this century.

If you're interested have a look at FromSoftware's actual Dark Souls (spiritual) predecessors. In particular King's Field. The setting is very similar; although the gameplay is very primitive, it has the same feeling of exploration and similar difficulty. Also it's 1st person (quite innovative for the time).

I don't have first hand experience but I believe a big problem with sound channels is multipath propagation/challenging frequency response. FSK naturally deals quite well with those; it's similar to OFDM but more robust due to single carried symbols. You can extend the symbol period as long as you like while using an adequate number. Finally in challenging environments it'd a good idea to use error correction, not just error detection as done here -- you can (in theory) achieve reliable transmission over arbitrarily noisy channels by adding redundancy.

It was just a comment on how it is possible to accelerate testing. I wasn't suggesting anything, just discussing and exchanging information. Whether they used this, it's probable, but to what extent only an employee could tell.

Well there is little, but there is also some. There's a whole field of accelerated testing of components, by degrading various variables, e.g. increasing temperature, vibration, sounds, corrosive environment, materials, etc., and then using curves of exaggerated failure times to predict failure times under nominal conditions. (after making tons of models and analysis)

You can also improve your simulation (here the sky is the limit).

It really depends on the cost of mishaps and design failures, if it's high enough you can almost always improve design with additional expenditure.

The extent to which we usually overestimate danger (i.e. are fearful) in non-physical situations can hardly be overstated.

We often behave as if some social triviality or temporary financial issue were actually life threatening, when indeed near-worst case scenarios are almost inconsequential. That person will judge you for a week or so then forget about it; if you fail financially you can rely on family, friends, savings, social security to not go homeless. The probabilities of bodily harm or true social outcasting are often very remote (and even outcast is often a but choice, since there are always other accepting people elsewhere).

Compare that with how much we fear climbing a simple step ladder -- we wouldn't give it much thought at all. But the probability of falling and having major bodily harm or even head trauma is perhaps much greater than the probability of poor outcomes from frightening daily situations. No that you shouldn't fear climbing ladders, just to give a perspective of how incongruent our intuitive fear can be to reality unless we police and adjust ourselves.

I wonder what portion of time on those supercomputers is used in optimizing the code itself; and how much could be done.

With those massive clusters you could afford testing in parallel trillions of mutations to your simulation kernels and prove the correctness of the fastest ones -- pruning should be extremely fast by finding counterexamples. Or even higher level architectural optimizations. Surely they could afford at least a few % of the total time on this pre-optimization (although the tools to achieve this automatically would need to be quite sophisticated!).

Perhaps the assistants could have more initiative then.

Ex: "Hi (...), since you always turn off the heater when you leave, would you like me to schedule to always do it for you?"

The problem is as soon as you're speaking of "frequency" you're already on a time-scale where the power doesn't interfere with other frequencies (as long as the medium is linear, etc -- which is almost certainly true here). In the mathematical/technical sense (in linear system theory)it's quite clear that different frequencies would interfere, it's a basic entire principle of linear circuit design that's been used for centuries -- systems have a Frequency Response where each frequency is independent as a result of the eigenfunction property. Personally I just think it's a big oversight, maybe they were trying to give a laymen explanation and confused themselves. What they're doing must be some form of phase modulation, maybe they're interpreting as short bursts of frequency modulation (with a stable steady-state frequency), but as I said this is not an illuminating view imo.

Edit: Actually I think they might have frequency control indeed. That would be a good idea to find a peak in frequency response (i.e. "resonant power transfer"), not for any directivity gains through constructive interference. That might be the source of confusion.

Isn't this description flawed though? Waves of different frequencies are independent in power. Thus they don't interfere (when time-averaged) -- at all. Zero. (this is because signals of different frequencies are orthogonal in the time-average integral of product sense)

What creates interference is difference is phase among waves of the exact same frequency.

It's odd that the paper author is making such basic EM theory mistake, and I wonder what the actual device construction is like. (since it cannot work the way he describes at steady state)

Maybe he's interpreting phase modulation as frequency modulation? Of course at steady state all frequencies would be the same, making this a weird/incorrect interpretation I think.

You can always make a mass driver system (which is not limited by the rocket equation), but given how insanely hard it is at our own gravity, it would be more insanely difficult at a few times great g.

So there's definitely a quite low maximum gravity allowing practical space access.

Indeed, cooling would be quite bulky and require lots of extra energy. But for an environment like Venus cited (instead of small sensors), I think it should be a good option. The Carnot efficiency to cool from a Venutian environment (462C) down to 50C (reasonable chip temp) is a surprisingly good ~60%. In practice you might get 15% (? plus insulation leaking), that should still allow cooling a chip of less than ~1 watt of tdp without too much hassle inside a robot.

You didn't address my other points, where I question if OP should rule that out.

> Otherwise, they may spend their whole lives fruitlessly chasing after immortality.

What about spending their whole lives fruitlessly chasing egoistic hedonism?

Again you didn't present a convincing argument that it is totally fruitless. As I've said, it certainly shouldn't be one's singular goal. But I think there's a role it should play.

> And you'll be too dead to enjoy it if you are. Are you sure you have come to terms with that?

Honestly, I don't think it's all that easy to rule out valuing the future beyond death, or to categorically reject notions of continuation of identity beyond death. That we are a coherent, super defined identity that has a definite beginning and a definite ending is in large part a convenient, and comfortable, myth we tell ourselves. In reality ourself from moments ago is a (slightly) different being, and you from many years ago, say from your childhood, is an almost completely different being, not only physically but a completely different mind. I don't think most people have come to terms with this either (which is understandable imo because those concepts are very tricky to interpret in a non-nihilistic way); while this truth give perspective as to why other forms of propagating your impact in the world and on other people that go beyond just your future self are perhaps justified. That the continuation of part of your identity in your work, teachings, children, friends, beyond your death can be "rational".

Ultimately I think (and so seems your position) one of the main motivations people should have is to simply have a good, happy, meaningful life themselves (where meaning isn't fame or grandeur). But, again "rationally" (in a classical economic-ish sense), there's definitely a big place for helping other people and building a legacy imo, as previously justified and based on ethical principles like the Categorical Imperative[1]. Of course if everyone were singularly concerned with helping others, no one would be reaping the fruits of labor, then what would be the point of help? That's why the one's main motivation should always be your own happiness, whatever that even means.

[1] https://en.wikipedia.org/wiki/Categorical_imperative

I wonder if simply cooling isn't an option in those applications? I guess that'd depend on the chips power output. Otherwise you could just surround it in a really good insulator and pump residual heat out somehow (maybe can peltiers withstand such high temps using multiple stages?)