LessWrong 2.0 Reader

The most ambitious I could be would be spending twenty-five hours a week getting better at and doing activism for my internship.

That sprang to my mind as the perfect solution to this problem.

Why do you think the methodology of calculating piracy damages by taking the number of downloads and multiplying by the retail price utter bunk?

Great reference! I found myself explaining this repeatedly but without the right terminology. The "but comparative advantage!" argument is quite common among economists trying to wrap their head around AI advances.

I think it applies for worlds with tool/narrow AI, but not with AGI that can do whole jobs for much lower wages than any human can do anything.

I am not talking about any 'good enough' answer. Whatever you deem 'good enough' to some arbitrary precision.

I am talking about the correct answer every time. This is not a matter of sufficient compute. Because if the answer comes even a fraction of a second AFTER, it is already too late. The answer has to be immediate. To get an answer that is immediate, that means it took zero amount of time, no compute is involved.

That sounds like either a nonsensical or a fundamentally impossible constraint to me. 

Max human reaction times (for something like responding to a visual cue by pressing a button) are about 150-200 milliseconds. Just the input side, for a signal to travel from the retina, down the optic nerve, through the brain to the visual cortex takes 50-80 milliseconds.

By the time your sensory cortices receive (not even process, just receive) the raw sense data, you're already a fraction of a second out of sync with reality.

Possibly you're not concerned with the delay between an external event occurring, and when the brain parses it into a subjective experience, only the delay between the moment of subjective experience^[1] and one's response to it? But the same basic issue applies at every step. It takes time for the visual cortex to do image recognition. It takes time to pass info to the motor cortex. It takes time for the motor cortex to send signals to the muscles. It takes times for the muscles to contract.

No response can be immediate in a physical universe with (in the most extreme case), a lightspeed limit. Insofar as a response is triggered by some event, some time will pass between the event and the response.

Or to put it another way, the only way for a reaction to a situation to not involve any computation, is for the response to be completely static and unvarying, which is to say not even slightly responsive to the details of the situation. A rock that has "don't worry, be happy" painted on it can give a perfect truly instantaneous response, if "a rock that says 'don't worry, be happy'" is the perfect response to every possible situation.

Am I being blockheaded here, and missing the point? 

Do you mean something different than I do by "instantaneous"? Or are you putting forward that something about worldviews sometimes relies on faster than light signaling? Or is your point that immediate responses are impossible.

1. ^{^}

   Notably my understanding is that there isn't a moment of subjective experience, for all it seems like sense data coheres into a single unified experience. Rather, there are a bunch of not-quite consistent pseudo-impressions that don't all take place at a exactly the same moment, but the differences between those are papered over. (You can do various experiments that mess with that papering over process, and get confused, mixed, subjective experience of events happening in sequence.)

My real view on P vs NP is that at this point, I think P almost certainly not equal to NP, and that any solving of NP-complete problems efficiently to the standard of complexity theorists requires drastically changing the model of computation, which corresponds to drastic changes in our physics assumptions like time travel actually working according to Deutsch's view (and there being no spurious fixed-points).

I did some analysis of audio watermarks a long time ago, and concluded that the answer to the title is "yes, this is a lost cause analytically, but viable socially and legally for quite some time".  The reason is mostly economics and standards of proof, not technical.  You start out with 

1) The watermark should be decisive: It should be almost impossible for normal text to appear marked by accident.

But then most of your analysis is statistical.  Without defining "almost impossible", you don't actually know what's successful.  What will hold up in court, what will be good enough to send a threatening letter, and what will allow you to enforce an internal rule are all somewhat different requirements.

Further, you assume random distribution, rather than your point 3

3) The watermark should be robust and able to withstand alterations to the text. Perhaps even from a malicious actor trying to destroy the watermark.

Alteration is adversarial - people are TRYING to remove or reduce the level of proof provided by the watermark. For audio, it took a long time for the knowledge of acoustic modeling and MP3 compression to filter into transcoding tools that removed watermarks.  It remains the case that specific watermarking choices are kept secret for as long as possible, and are easily removed once known.  The balance of not affecting perceptible sound and not trivially removed with a re-encode is still debated, and for most purposes, it's not bothered with anymore.

Text is, at first glance, EVEN HARDER to disguise a watermark - word choice is distinctive, especially if it deviates enough to get to "beyond a reasonable doubt" threshold.  For most schemes, simple rewrites (including automatic rewrites) are probably enough to get the doubt level high enough that it's deniable.

Then there's the question of where the watermark is being introduced.  Intentional, "text came from X edition of Y source" watermarking is different from "an LLM generated this chunk of text".  The latter is often called "fingerprinting" more than "watermarking", because it's not intentionally added for the purpose.  It's also doomed in the general case, but most plagiarists are too lazy to bother, so there are current use cases.

There is also the reverse case, where it is often common practice in math or logic to ignore bizarre and unnatural counterexamples. For example, first-order Peano arithmetic is often identified with Peano arithmetic in general, even though the first order theory allows the existence of highly "unnatural" numbers which are certainly not natural numbers, which are the subject of Peano arithmetic.

Another example is the power set axiom in set theory. It is usually assumed to imply the existence of the power set of each infinite set. But the axiom only implies that the existence of such power sets is possible, i.e. that they can exist (in some models), not that they exist full stop. In general, non-categorical theories are often tacitly assumed to talk about some intuitive standard model, even though the axioms don't specify it.

Eliezer talks about both cases in his Highly Advanced Epistemology 101 for Beginners [? · GW] sequence.

The point is that you can't use the result that there exists a hard function, since all you know is that the function is hard, not whether it's in NP, which is a basic problem for your scheme.

Your counting argument for Turing Machines also likely have this problem, and even if not, I see no reason why I couldn't relativize the results, which is a no-go for P vs NP proof attempts.

Footnotes are good in translated works. I read the 3-body trilogy translated into English, and it was very helpful to have notes from the translator explaining certain points of cultural context that a Chinese reader would be expected to be familiar with.