Software-Generated Paper Accepted At IEEE Conference | Hacker News Reader

Software-Generated Paper Accepted At IEEE Conference

entertainment.slashdot.org

19 pointsnaish11 comments

So it was post-modern cultural studies first with Alan Sokal (http://en.wikipedia.org/wiki/Sokal_affair), then theoretical physics with the Bogdanov brothers (http://en.wikipedia.org/wiki/Bogdanov_Affair), and now it's the turn of computer science. As good CS people, the meta-authors of this article used a random generator rather than faking the text themselves. I guess the discipline is growing up. :)

The abstract was craptastic but it is no different from the craptastic abstracts of several human-generated papers.

I think part of the problem is the people expect academic papers to be complicated/cutting edge/forward thinking, and therefore not immediately comprehensible.

Now I have to admit, the first sentence of the abstract sounded really good but it was all downhill from there. Again though, it was downhill in the same way as many other abstracts I have read.

Maybe this means that academics need to stop trying to sound so smart in their papers and just get the point across...or maybe it just exposes one (or several) of the reviewers as utterly incompetent.

When I was writing my first paper, I legitimately tried to write it so anybody could understand. My test was to have my girlfriend read it and tell me what she thought it was about (not that my girlfriend isn't smart, but she had absolutely no clue what I was working on). When I showed it to the PI, he said I had to revise it all because it was "too juicy."

More specifically, he said there were two reasons against understandable writing:

1) Scientists are lazy, and do not want to read long passages. Coupled with the fact that they are usually pretty smart, they can understand the terse language.

2) Scientific (read: incomprehensible) writing removes all biases. There is no room to interpret something incorrectly if the precise words are used.

I agree though, it's not a very inviting culture that promotes these ideals.

That's my point though. Even extremely smart people can stumble on a dense abstract. It is the nature of language- just because I know what I mean does not mean that you do. Even if you write something that can only be interpreted correctly in one way, there is still room for misinterpretation or total lack of understanding.

I was really trying to say that if a paper with no meaning made it through screening (no matter how light) then maybe the papers are too dense.

In response to 1), I have always found that scientific papers provide more verbose explanations for even simple concepts. I have reread entire paragraphs three times only to discover that I already have a firm grasp of the concept being discussed and can explain it in a single sentence. But it probably just depends on the author and the verbosity could be explained by your 2): removing all uncertainty by explaining exactly what you mean.

I think that this is a failing of modern science. It leads to fragmentation between fields. Of course most chemists can read a chemistry article, but a brilliant physicists will stumble on the same article without the common language shared among chemists. This makes interdisciplinary collaboration difficult if not impossible. Each field develops their own version of major concepts. I've taken quantum mechanics courses from both the chemistry and physics departments at the same university. They might as well have been speaking different languages the notations and presentation of concepts were that different.This also makes life tough for those entering a field such as grad students and scientists who switch fields.

Magazines such as Scientific American used to help by featuring article geared towards scientist outside their specialty, but now SciAm seams more geared toward the lay reader.

I think many researchers have a deep-seated insecurity that they aren't smart enough. They hide behind jargon and incomplete explanations to make their research appear more complicated than it really is. Most concepts in science and especially computer science are pretty easy when explained well and next to impossible when obfuscated with unfamiliar jargon.

</rant by grad student sick of reading journal articles>

It was actually accepted as a "poster" which is not a rigorous thing. I still think it's an embarrassment. This has been done before, and accepted as a full CS paper (search for "scigen").

The best example is the "Sokal affair" (search for "Alan Sokal") which is awesome because it doesn't just demonstrate sheer negligence with a randomly generated paper: it was carefully crafted to sound like it made sense.

Computer science has failed the pseudoscience test:

http://unqualified-reservations.blogspot.com/2007/08/whats-w...

Very interesting article. Should be a submission here.

http://news.ycombinator.com/item?id=43074

"In this work we better understand how digital-to-analog converters can be applied to the development of e-commerce." LOL :)

people don't like admitting that they don't know something. Especially that they don't know something when they're supposed to be very knowledgeable about it.

But come on! They could have rejected it as "incomprehensible" or "needing to explain what it was talking about better" or "not staying on one topic".

The abstract started talking about the Internet and OOP and ended up talking about digital-to-analog converters and e-commerce.

Just because you're smart and you can't understand it doesn't mean that it's smarter.

Reply on news.ycombinator.com