arXiv vs. snarXiv

[steuard]

Entertainment of the week: arXiv vs. snarXiv. How well can you distinguish actual high-energy physics paper titles from computer generated fakes?

As every theoretical physicist knows, the arXiv.org preprint server is the go-to place for current research. (That "X" is supposed to be the Greek letter chi.) Essentially every string theory paper is posted there long before it's published, so active researchers check the new submission list daily.

The newly released snarXiv is "a ran­dom high-energy the­ory paper gen­er­a­tor incor­po­rat­ing all the lat­est trends, entropic rea­son­ing, and excit­ing mod­uli spaces." It generates titles, author lists, and abstracts (for now). Its creator goes on to explain that "The arXiv is sim­i­lar, but occa­sion­ally less ran­dom." His blog post (linked here) even goes on to suggest good uses for the snarXiv at each stage of your career. This is all absolutely hilarious to those of us who follow the arXiv for a living. For everyone else, it's a chance to laugh at us.

When I tried the arXiv vs. snarXiv quiz, I got to 10/10 and then stopped for fear of embarrassing myself if I eventually got one wrong.

Comments

[spoonless.livejournal.com]

Holy cow, I just got a hilarious one:

"Type IIA Surrounded by a Stack of A-type Branes Wrapped on Superspace as Dark Energy on the Surface of the Sun"

Overall, I was impressed by its ability to generate realistic sounding titles, although there were some flagrant tip-offs, like a lot of cases where it just puts "Towards" in front of some topic. Also, the real ones, on average, tended to be a lot longer than the fake ones. Presumably because when they try to do too long of one, like the one above, it becomes too obviously fake.

[steuard.livejournal.com]

I got one fake title that was, honest to goodness, at least fifteen lines long. Almost equally obvious are the ones that are very short: nobody would write a paper entitled just "Heterotic String Theory". (Well, ok, my advisor, but that was rather a long time ago.)

8 of 10 correct

[beth_leonard]

I got 8 out of 10 correct. The ones I got wrong were:

Models of Black Branes
vs.
A Heavy Higgs Boson From Flavor and Electroweak Symmetry Breaking Unification

I thought that surely breaking unification and the poor syntax on "from flavor" had to be a made-up title...

And the grammar got me again here:
The Reduction of Type IIB Strings Living on \Z_7 Quotients of Hirzebruch Surfaces of Ext^6(\R,\Z) Holonomy
vs.
On Cosmological Type Solutions in Multidimensional Model With Gauss-Bonnet Term

I would have pluralized "Model" and "Term" if I were writing that title, so I thought surely it must have been made up.

I hadn't heard of most of the terms in most of the papers. It sounds like complex stuff you're studying there ;-)

--Beth

Re: 8 of 10 correct

[steuard.livejournal.com]

I believe that the first title you got wrong could be rephrased as "A Heavy Higgs Boson [Arising] From the Unification of {Flavor} and {Electroweak Symmetry Breaking}": the syntax of the original doesn't feel too bad if you're already familiar with which clumps of words naturally go together.

The second title you got wrong is just poor grammar, but one gets used to that. English is only a first language for a limited fraction of the world's string theorists, so I try to be forgiving about such things (though I'll admit that it can be maddening at times). I'm guessing I would probably have included the word "A" twice in this title (rather than pluralizing), but it's not uncommon to omit the second of those even among native speakers (at least in titles).

I hadn't heard of most of the terms in most of the papers.

After I'd been going to the weekly string seminars for a while in grad school, I described the experience as "listening to people make odd grunting noises week after week until I started to recognize clumps of grunting noises that often occurred together". We do have rather a lot of jargon, but I don't think that's entirely different than other complex professions.

In fact, it may actually help that many of our terms are unfamiliar (or used in very unfamiliar ways). Compare this to, say, women's studies, where in my limited experience the jargon often consists of familiar words whose specialized meaning differs in small but absolutely crucial ways from standard usage. There, context isn't necessarily enough to make you realize that you're missing the whole point of the statement.