But QC is quite plausible. There is no theoretical constraint that makes it impossible. It really is just an engineering problem at this point.
57,863 karma · joined January 13, 2008
That is a hard question to answer for two reasons. First, there is no bright line that delineates "cryptographically useful". And second, the exact design of a QC that could do such a calculation is not yet known. It's kind of like trying to estimate how many traditional gates would be needed to build a "semantically useful" neural network back in 1985.
But the answer is almost certainly in the millions.
[UPDATE] There is a third reason this is hard to predict: for quantum error correction, there is a tradeoff between the error rate in the raw qbit and the number of gates needed to build a reliable error-corrected virtual qbit. The lower the error rate in the raw qbit, the fewer gates are needed. And there is no way to know at this point what kind of raw error rates can be achieved.
> Is there some pathway that makes quantum computers useful this century?
This century has 75 years left in it, and that is an eternity in tech-time. 75 years ago the state of the art in classical computers was (I'll be generous here) the Univac [1]. Figuring out how much less powerful it was than a modern computer makes an interesting exercise, especially if you do it in terms of ops/watt. I haven't done the math, but it's many, many, many orders of magnitude. If the same progress can be achieved in quantum computing, then pre-quantum encryption is definitely toast by 2100. And it pretty much took only one breakthrough, the transistor, to achieve the improvement in classical computing that we enjoy today. We still don't have the equivalent of that for QC, but who knows when or if it will happen. Everything seems impossible until someone figures it out for the first time.
---
[1] https://en.wikipedia.org/wiki/UNIVAC_I#Technical_description
Oops.
That might be your problem right there. Deciding you can't do something is always a self-fulfilling prophecy. How hard have you tried?
I learned to turn my weird off a long time ago. It wasn't easy. It took many years. It was painful at times. But I did it. If I can do it, you probably can too.
P.S. You might want to think about whether or not turning your weird off is something you actually want. Being normal comes with its own set of trade-offs. But if you are going to keep your weird you should do it because it's something you decide you want, not because it's something you decide you are powerless to change.
You need to look at this:
Python 3.11.6 (v3.11.6:8b6ee5ba3b, Oct 2 2023, 11:18:21) [Clang 13.0.0 (clang-1300.0.29.30)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> import lisp
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
ModuleNotFoundError: No module named 'lisp'In actual practice what I found was that the principle driver of publishing success has absolutely nothing to do with the quality of the work, it has to do with how much your reviewers think that you might some day be in a position to review a paper of theirs. This is the fundamental problem with peer review when: career success is measured by quantity of papers published, the resulting dynamic is governed by game theory, not scientific merit.
I know this because I've been the beneficiary of this system on multiple occasions, and it makes me sick. This is not the kind of world I want to live in.
That's like saying that Darwinian evolution is simple. It's not entirely wrong, but it misses the point rather badly. The thing that makes search useful is not the search per se, it's the heuristics that reduce an exponential search space to make it tractable. In the case of evolution (which is a search process) the heuristic is that at every iteration you select the best solution on the search frontier, and you never backtrack. That heuristic produces a certain kind of interesting result (life) but it also has certain drawbacks (it's limited to a single quality metric: reproductive fitness).
> Beam search is an example of TTC in this modern era.
That's an interesting analogy. I'll have to ponder that.
But my knee-jerk reaction is that it's not enough to say "put reactivity and deliberation together". The manner in which you put them together matters, and in particular, it turns out that putting them together with a third component that manages both the deliberation and the search is highly effective. I can't say definitively that it's the best way -- AFAIK no one has ever actually done the research necessary to establish that. But empirically it produced good results with very little computing power (by today's standards).
My gut tells me that the right way to combine LLMs and search is not to have the search manage the LLM, but to provide search as a resource for the LLM to use, kind of like humans use a pocket calculator to help them do arithmetic.
> If LLMs are coming up with effective search formulation for "open-ended" problems, that would be a big deal.
AFAICT, at the moment LLMs aren't "coming up" with anything, they are just a more effective compression algorithm for vast quantities of data. That's not nothing. You can view the scientific method itself as a compression algorithm. But to come up with original ideas you need something else, something analogous to the random variation and selection in Darwinian evolution. Yes, I know that there is a random element in LLM algorithms, and again I don't really understand the details, but the way in which the randomness is deployed just feels wrong to me somehow.
I wish I had more time to think deeply about these things.
Thirty-five years ago they gave me a Ph.D. basically for pointing out that the controversy du jour -- reactive vs deliberative control for autonomous robots -- was not a dichotomy. You could have the best of both worlds by combining a reactive system with a deliberative one. The reactive system interfaced directly to the hardware on one end and provided essentially a high-level API on the other end that provided primitives like "go that way". It's a little bit more complicated than that because it turns out you need a glue layer in the middle, but the point is: you don't have to choose. The Bitter Lesson is simply a corollary of Ron's First Law: all extreme positions are wrong. So reactive control by itself has limits, and deliberative control by itself has limits. But put the two together (and add some pretty snazzy image processing) and the result is Waymo.
So it was no surprise to me that Stockfish, with its similar approach of combining deliberative search with a small NN computing its quality metric blows everything else out of the water. It has been obvious (at least to me) that this is the right approach for decades now.
I'm actually pretty terrified of the results when the mainstream AI companies finally rediscover this. The capabilities of LLMs are already pretty impressive on their own. If they can get a Stockfish-level boost by combining them with a simple search algorithm the result may very well be the GAI that the rationalist community has been sounding the alarm over for the last 20 years.
Electing Trump was a big FU to that attitude. The astonishing thing is that liberals are so cocksure of themselves that they have not yet figured out this simple truth and are still carrying on as if Trump were simply an anomaly rather than a predictable response to their own actions. The magnitude of the tone-deafness in the Democratic party is simply staggering. And I'm a Democrat, or at least I was until I realized how utterly incompetent they are.
[UPDATE] Ironically, the fact that this comment is being downvoted into oblivion actually demonstrates the very point I am making.
[UPDATE2] With regards to my saying that Democrats are incompetent, this is manifestly true at least with regards to 1) winning elections and 2) controlling Donald Trump. Maybe they are competent at other things, but that seems like a bit of a moot point to me under the present circumstances.
Management Science & Engineering (MS&E): Stanford’s interdisciplinary hub
I don't have kids so I have no direct experience with this, but I'm pretty sure that parent's don't actually instruct their kids how to walk. Kids figure it out mostly on their own. Parents might accelerate the process by giving guidance and encouragement, but even that probably is just another feedback channel. The heavy lifting is all in the kid's brain.
Contrast that with learning how to play chess. The latter is something that a higher level system really can teach to a lower-level one. Really good chess players eventually develop sub-conscious pattern-matching abilities that are very much like a motor skill, but before that happens a person first has to learn to play chess using slower deliberative methods, and they learn that through language. Other good examples are learning to drive, or fly an airplane, write code. But walking really is different. There's a reason that walking is not taught in preschool. (Physiotherapy is different. And as you yourself point out, some people figure that out on their own too.)
And unless those men are women.
Yeah. It’s called “parenting”.