Ask HN: Which recent research paper blew your mind? • Hacker News Board

Re: Ask HN: Which recent research paper blew your mind?

#161

Post by tracker1 » Tue, Jul 25, 2023, 2:58 PM UTC

darkclouds wrote: post #60

edent wrote: post #17

"Overview of SHARD: A System for Highly Available Replicated Data" it's the first paper to introduce the concept of database sharding. It was published in 1988 by the Computer Corporation of America. It is referenced hundreds of times in many classic papers. But, here's the thing. It doesn't exist . Everyone cites Sarin, DeWitt & Rosenb[e|u]rg's paper but none have ever seen it. I've emailed dozens of academics, libr…

Still searching then? https://shkspr.mobi/blog/2021/06/where-is-the-original-overv... I can only find the Oracle reference to Sharding, which might be the same thing or not. https://docs.oracle.com/en/database/oracle/oracle-database/1... Along with the wikipedia reference. https://en.wikipedia.org/wiki/Shard_(database_architecture) And a Science Direct reference. https://www.sciencedirect.com/topics/computer-science/…

It's funny in that it was such a pivotal work looking back at it. It's possible in some filing cabinet at Xerox or elsewhere (from what I gathered in the thread). That said, it may have simply seemed somewhat obvious to many at the time and nobody bothered to hold on to or archive it. I suspect it's true of a lot of now commonplace software design strategies.

Re: Ask HN: Which recent research paper blew your mind?

#162

Post by GregarianChild » Tue, Jul 25, 2023, 3:27 PM UTC

lmeyerov wrote: post #150

GregarianChild wrote: post #134

Earlier quoted context omitted.

The paper claims that there is no neural / deep learning based solver that performs well on regular expression inference. Calls to Coq and Z3 will be very slow and not competitive with GPU compute.

Ironically, one of its few comparisons was FlashFill / Excel.. which had been going the way of NNs and now LLMs for years now. The paper largely skips measured and non-superficial comparisons, so I wouldn't draw comparative conclusions based on their writeup. RE Coq/Z3 vs GPUs, I think about them as different kinds of things, and thus wouldn't compare directly. A solver like z3 encapsulates search tricks for tasks wh…

Could you point towards a paper that does precise and minimal regular expression (or finite state automaton) inference?

Re: Ask HN: Which recent research paper blew your mind?

#163

Post by lmeyerov » Tue, Jul 25, 2023, 3:54 PM UTC

GregarianChild wrote: post #162

lmeyerov wrote: post #150

Earlier quoted context omitted.

Ironically, one of its few comparisons was FlashFill / Excel.. which had been going the way of NNs and now LLMs for years now. The paper largely skips measured and non-superficial comparisons, so I wouldn't draw comparative conclusions based on their writeup. RE Coq/Z3 vs GPUs, I think about them as different kinds of things, and thus wouldn't compare directly. A solver like z3 encapsulates search tricks for tasks wh…

Could you point towards a paper that does precise and minimal regular expression (or finite state automaton) inference?

That's a great yet narrow request, which is great for POPL and less for PLDI. As a reference, I believe starcoder does worse than GPT4 and has taken time to publish results on program synthesis that subsumes regex. The flashfill team has academic PL roots and would also be good to compare to, as I mentioned.

With prompt engineering being so easy, and naive search/refinement over them so easy too, not comparing via benchmarks, vs brush-off related work section, to these general solvers seems strange.

Re: Ask HN: Which recent research paper blew your mind?

#164

Post by pyuser583 » Tue, Jul 25, 2023, 5:38 PM UTC

“On the frequency and severity of interstate war” by Aaron Clauset.

https://arxiv.org/pdf/1901.05086.pdf

Analyzes a claim made in the 1950s by a prominent statistician: the frequency of interstate wars follows a simple Poisson arrival process and their severity follows a simple power-law distribution.

He was right, but it’s not clear why.

The paper is interesting because it shows how a new bit of knowledge creates a large number of known unknowns from previously unknown unknowns.

It also shows statistical magic was very much possible prior to computers.

Re: Ask HN: Which recent research paper blew your mind?

#165

Post by darkclouds » Tue, Jul 25, 2023, 5:50 PM UTC

tracker1 wrote: post #161

darkclouds wrote: post #60

Earlier quoted context omitted.

Still searching then? https://shkspr.mobi/blog/2021/06/where-is-the-original-overv... I can only find the Oracle reference to Sharding, which might be the same thing or not. https://docs.oracle.com/en/database/oracle/oracle-database/1... Along with the wikipedia reference. https://en.wikipedia.org/wiki/Shard_(database_architecture) And a Science Direct reference. https://www.sciencedirect.com/topics/computer-science/…

It's funny in that it was such a pivotal work looking back at it. It's possible in some filing cabinet at Xerox or elsewhere (from what I gathered in the thread). That said, it may have simply seemed somewhat obvious to many at the time and nobody bothered to hold on to or archive it. I suspect it's true of a lot of now commonplace software design strategies.

> it may have simply seemed somewhat obvious to many at the time and nobody bothered to hold on to or archive it.

I suspect this has been the case for many especially last century, everything was moving so fast and there was so much choice, it really was a free for all with no major players dominating the market place and some people didnt know the significance of what they were doing or building. People then retire or get moved to different projects and the knowledge gets lost.

Re: Ask HN: Which recent research paper blew your mind?

#166

Post by codethief » Tue, Jul 25, 2023, 8:39 PM UTC

the-mitr wrote: post #142

codethief wrote: post #105

Earlier quoted context omitted.

> that is, although, for example, chemistry is subject to the laws of physics, we cannot infer the field of chemistry from our knowledge of physics I mean we could , with infinite computing power and enough time to look into every interesting phenomenon (and to evaluate the corresponding multi-particle Schrödinger equation numerically) but there are simply too many such phenomena, Schrödinger equations are tough to s…

no, that is the point of the paper, even with infinite computing power you could not predict. it is a very strong case against reductionism in science. The phenomenological examples he considers in the paper cannot be derived from basic principles of quantum mechanics.

Huh, I find that very surprising. Looks like I need to have a deeper look at the paper then.

Re: Ask HN: Which recent research paper blew your mind?

#167

Post by smokel » Tue, Jul 25, 2023, 8:51 PM UTC

not-my-account wrote: post #98

dargscisyhp wrote: post #21

I thought the AlphaZero paper was pretty cool: https://arxiv.org/abs/1712.01815 Not only did we get a whole new type of Chess engine, it was also interesting to see how the engine thought of different openings at various stages in its training. For instance, the Caro-Kann, which is my weapon of choice, was favored quite heavily by it for several hours and then seemingly rejected (perhaps it even refuted it?!) near th…

totally! MuZero is my favourite[1] The super cool thing about MuZero is that it learns the dynamics of the problem, i.e. you don't have to give it the rules of the game, which makes the algorithm very general. For example, DeepMind threw MuZero at video compression and found that it can reduce video sizes by 6.28% (massive for something like YouTube)[2][3]. Curious if anyone else knows examples of MuZero being deploy…

> you don't have to give it the rules of the game

To be fair, it uses MCTS, which requires many simulations of the game. For this, it needs to know which moves are valid, and when a player wins or loses the game.

So it does need to know the rules of the game, but it doesn't need any prior knowledge about which moves are better than others.

Re: Ask HN: Which recent research paper blew your mind?

#168

Post by smokel » Tue, Jul 25, 2023, 9:16 PM UTC

the-mitr wrote: post #142

codethief wrote: post #105

Earlier quoted context omitted.

> that is, although, for example, chemistry is subject to the laws of physics, we cannot infer the field of chemistry from our knowledge of physics I mean we could , with infinite computing power and enough time to look into every interesting phenomenon (and to evaluate the corresponding multi-particle Schrödinger equation numerically) but there are simply too many such phenomena, Schrödinger equations are tough to s…

no, that is the point of the paper, even with infinite computing power you could not predict. it is a very strong case against reductionism in science. The phenomenological examples he considers in the paper cannot be derived from basic principles of quantum mechanics.

Hm, I can't say I really understand most of the paper, so I'm not sure if the paper actually implies this, but even then, I find this unconvincing.

Quantum mechanics allow for many things to happen at a macro level, such as magnetism. If you start from pure quantum mechanics, and try to predict something like that emerging, you'd have to be pretty lucky to find the correct initial conditions to lead to such a phenomenon. But that does not mean that it is impossible to predict it at all.

Another take on this might be that understanding phenomena at different levels requires different abstractions and different modes of understanding. In that sense, the knowledge at one level often does not help us to understand things at a higher level. But I think that says more about our limited intelligence than about the phenomena and emergence in themselves.

Re: Ask HN: Which recent research paper blew your mind?

#169

Post by marifjeren » Wed, Jul 26, 2023, 10:38 PM UTC

mindcrime wrote: post #140

mindcrime wrote: post #132

Earlier quoted context omitted.

I also sent a note to another former Computer Corporation of America employee that I found on LinkedIn. I don't know them personally, but we live near each other and have some common connections, so maybe they will at least receive my unsolicited message with some favor.

FWIW, I did get a response from the individual mentioned above. However they were unable to help with finding a copy of the paper. The search continues...

Someone who isn't me is getting in touch with a computer science librarian at a top 10 university about this paper. I'll update here with whatever I (don't) learn.

Re: Ask HN: Which recent research paper blew your mind?

#170

Post by marifjeren » Thu, Jul 27, 2023, 4:50 PM UTC

marifjeren wrote: post #169

mindcrime wrote: post #140

Earlier quoted context omitted.

FWIW, I did get a response from the individual mentioned above. However they were unable to help with finding a copy of the paper. The search continues...

Someone who isn't me is getting in touch with a computer science librarian at a top 10 university about this paper. I'll update here with whatever I (don't) learn.

No dice. The search continues..