Hacker News Reader

Comment by SeanLuke | Hacker News Reader

We published a paper a while back in a genetic programming conference, and before the paper had been published (only the title was announced) Schmidhuber guessed that we had not cited his work. He very publically put us through the wringer for not citing him as the inventor of the concept were were examining. In fact we had not cited him: but among our dozen or so previous examples, we had cited two others which long predated his work and were the actual seminal papers. He had failed to cite them himself.

View on HN

But it is possible that he ought to have been cited as well? It doesn't matter if he's a hypocrite. The justification ultimately has to be, "Our paper did not use your work in any substantial way therefore a citation is not needed."

I can see how citations can get very complex for the purposes of scientific credit, traditionally there's a self-policed aspect to it in every academic community; however, in this century as advanced research gets more complex and globalized, it might be time for a more rigorous and neutral process of doing so.

Whenever you cite papers just because you're "politically supposed to", even though you don't think it's really intellectually useful for anyone to be aware of the cited paper, you're making the citations less useful for everybody.

Schmidhuber is only the worst of many abusers of the citation system. Everyone knows that anonymous reviewers are more likely to approve of people who cite their work. Someone is politically powerful? Better cite their paper, even though you don't really think it's a good paper and you didn't use it.

Schmidhuber is cited far more often than he should be, because people just don't want to deal with this crap, and the cost of sticking in one more citation is very low.

> Whenever you cite papers just because you're "politically supposed to", even though you don't think it's really intellectually useful for anyone to be aware of the cited paper, you're making the citations less useful for everybody.

The audacity of deciding on behalf of readers whether information is valuable to them when publishing an academic paper is quite an interesting concept to me.

I'm not published in journals, but I do have a "professional" writing background. There's only a few ways I can imagine to read into your statement. Either you think you're above citation, or so much smarter than the reader you're qualified to decide on their behalf whether it's relevant, or simply hiding it for nefarious reasons.

Any of these options, or even the perception of them, is precisely the type of behavior fostering anti intellectualism and prejudice for academia.

It's also quite ironic you repeatedly mention "politics" while your entire argument seems to hinge on seeking to deny empowering further those already in power. Isn't that exercising political capital?

> Isn't that exercising political capital?

These are very good points.

> ... the worst of many abusers of the citation system.

Is this ad hominem argument by user "lacker" meant to distract from the omissions of the awardees Bengio, Hinton, and LeCun? Especially the first two got tons of citations for work that should have credited Schmidhuber's lab: the analysis of vanishing gradients in neural networks, the principle of generative adversarial networks, attention in neural networks, distilling neural networks, speech recognition with LSTM neural networks, self-supervised pre-training, and more.

The disputes with LeCun are more recent and of lesser magnitude IMO.

My godfather is an academic that does some arcane nonsense (I mean that in the nicest way possible) involving topology and number theory. He told me a story about how he and his coauthor had a paper in review for 3 years because the reviewer insisted on adding citations to what they assumed were his own work because they were not particularly relevant. My godfather after a round or two didn’t have the patience for it and said whatever let’s just put them in by his coauthor refused hence the three years of back and forth.

>> Schmidhuber is only the worst of many abusers of the citation system. Everyone knows that anonymous reviewers are more likely to approve of people who cite their work. Someone is politically powerful? Better cite their paper, even though you don't really think it's a good paper and you didn't use it.

The reason that this is not the norm in academia is because there is a substantial number of researchers who care very, very much about academic integrity and even more so about getting their work properly recognised, just like good old You_again.

Btw, "everyone knows" is a running joke in Game of Thrones, it's what the Dothraki always say to show that they're pre-scientific barbarians who believe whatever they like, or at least that's how I get the joke.

Whether Schmidhuber is a bad actor taking advantage of the "system" in general is separate from the specific case I was responding to (not sure if that was your comment or somebody else's.)

But as I said before, we should take the long view. Globalized scientific research ought to have a more neutral way of assessing citations. The 20th century way of doing it may not be the right way any more. I don't see anyone else suggesting a problem of this scope.

> Schmidhuber is cited far more often than he should be

You haven't even read the paper, have you? Otherwise you'd see that it's Hinton and Bengio who are cited far more often than they should be. Just look at disputes B1, B2, B5, H2, H4, and H5 to see how they republished parts of his work again and again without citing it. No honest scientist can approve of something like that.

He really should be cited. The paper is supposed to give context to the reader, so to not cite a notable paper that is relevant is doing a disservice to the reader and it does seem pretty sketchy to me (I mean only if it's notable, idk who Schmidhuber is).

Sometimes this just happens by mistake. The literature is vast and people rediscover things fairly frequently that they honestly didn’t know existed. The related work section is a good faith effort to contextualize your work in the space of extant works. You will not know every work even seminal ones (it happens, especially if it came out a few decades beforehand), and you don’t have space to include everyone. Sometimes you’re coming from an adjacent field that doesn’t literature from other sources. I study a lot of control theory, and optimal control theory is something that shows up in engineering and economics. An economist writing on the topic may not be familiar with all the engineering literature. Recently, with the machine learning explosion, I’ve come across reinforcement learning papers from time to time where they rediscovered something from optimal control and are proposing it as a whole new thing because, well, they don’t know a lot about the control theory literature. Personally, I don’t think it’s a large problem because it brings the idea to a new audience and it’s impossible to know everything.

Another aspect that makes it difficult is that different people even in the same field use different terms for the same thing. As an example, engineering optimal control folks like to optimize cost functions, economists might refer to that as the utility instead, and RL folks like to use rewards (which is just a negative of a cost and functionally equivalent). That makes it difficult to run a search. I’ve come across a whole slew of new papers to read on topics just by changing out key terms for synonyms.

A reviewer may suggest other works to include in your discussion. Sometimes the recommendations are appropriate, other times it’s a backhanded way to fish for citations and the works suggested aren’t actually relevant. I would say harping over not including one is poor form unless it is so similar that is important for you to explain the novel aspect in your paper.

"The literature on this problem is extensive, widely scattered, and not always aware of itself."

Is that quote from something? It's funny. I couldn't find any Google result for it.

an article on Malfatti circles in geometry https://en.wikipedia.org/wiki/Malfatti_circles#History

These are good points but Bengio, Hinton and LeCun do not ignore Schmidhuber's work, they just don't cite it.

Our paper was an analysis of an already established technique used nearly identically by a number of previous papers. We dug up about a dozen IIRC, including the seminal ones, but we missed his. He didn't go after us for not citing him (which he had divined). He went after us for not acknowledging that he invented it (which he did not).

Are you talking about the 1985 Genetic Programming paper by Cramer? Unlike Hinton and Bengio, Schmidhuber has corrected himself:

> BTW, I committed a similar error in 1987 when I published what I thought was the first paper on Genetic Programming (GP), that is, on automatically evolving computer programs[GP1][GP] (authors in alphabetic order). At least our 1987 paper[GP1] seems to be the first on GP for codes with loops and codes of variable size, and the first on GP implemented in a Logic Programming language. Only later I found out that Nichael Cramer had published GP already in 1985[GP0] (and that Stephen F. Smith had proposed a related approach as part of a larger system[GPA] in 1980). Since then I have been trying to do the right thing and correctly attribute credit.

Source: https://people.idsia.ch/~juergen/deep-learning-miraculous-ye...

No.

So which papers are you talking about?

> He very publically put us through the wringer for not citing him as the inventor of the concept were were examining

He put you through the wringer for not citing him in the announcement of the title?

Reply on news.ycombinator.com