?

Log in

No account? Create an account
Abstract length - Arvind Narayanan's journal [entries|archive|friends|userinfo]

Abstract length [Mar. 18th, 2007|10:24 am]
Arvind Narayanan
[Tags|, ]
[Current Mood |caffeinated]

Over the last week, I met many people who'd come across one of my papers. Most had only looked at the abstract, which is quite natural -- reading pdfs on screen is not fun, and you generally don't print out a paper that's only peripherally related to your area of interest.

This got me thinking: currently, abstracts are typically written in a hurry 5 minutes before the submission deadline, and are two paragraphs long. This is bad; it would be in the best interest of the authors if those who run into your paper actually learn what you're doing, and so abstracts should be about a page long and should have some thought put into them.

Good idea? Bad idea? Anyone already doing this?
LinkReply

Comments:
[User Picture]From: theswede
2007-03-18 06:41 pm (UTC)
Two paragraphs is perfect; however, just like with the cover letter for a job application, a LOT of work should be put in the abstract. It's what sells the paper. A lousy abstract will bring down an excellent paper, and a great abstract will rescue a lousy paper - just like cover letters with CV's. Of course, nothing will make a professor show mercy, just like a CV ultimately also has to stand on its own ... but with a great cover letter/abstract, the foot is in the door.
(Reply) (Thread)
From: nicertry
2007-03-18 09:28 pm (UTC)
i agree with the above comment .. the solution is not to increase the space limit for abstracts but force authors to think about what exactly is the gist of the paper .. advocate the more thought rather than more space bit. im guessing the concept of needing to write long abstracts to convey your main ideas doesnt quite exist in theory papers?
(Reply) (Thread)
[User Picture]From: ephermata
2007-03-19 03:42 am (UTC)
More thought good, more length bad!

The abstract also helps you define a target audience. For example, an abstract that uses several different complexity classes and talks about relations between them may advertise the paper as being directed at experts in complexity theory.
(Reply) (Thread)
[User Picture]From: arvindn
2007-03-19 05:40 am (UTC)
Hmm.. everyone seems to be against longer abstracts. Maybe I'm just obtuse, but would you mind explaining what the downside is, other than the scarcity of space in a dead-tree medium?
(Reply) (Parent) (Thread)
[User Picture]From: ephermata
2007-03-19 05:47 am (UTC)
It's the scarcity of attention. I will read a two-paragraph abstract, but I won't usually read a full-page thing unless I'm bored or have some other reason. Two paragraphs is about the right length for sending as part of a talk announcement or putting up on a feed of new papers, too, where the emphasis is on quickly deciding whether to spend more time or not.
(Reply) (Parent) (Thread)
[User Picture]From: forvrkate
2007-03-19 04:00 am (UTC)
\begin{cry}
An abstract! Something else I must write before due date!
\begin{freakout}


(Please type a command or say `\end')
(Reply) (Thread)
[User Picture]From: skthewimp
2007-03-20 06:54 am (UTC)
completely agree wiht theswede, nicertry and ephermata.

write it better but two paragraphs is just perfect i think! enough to tell the reader what the paper is all about. and they'll read the whole thing also!
(Reply) (Thread)
From: fixious
2007-03-25 08:05 pm (UTC)
Just for perspective, humanities (and perhaps also social science) conferences require submission of only the abstracts, which are generally 1-2 pages long.
(Reply) (Thread)
[User Picture]From: vinodv
2007-04-03 11:47 am (UTC)

I have quite the opposite point of view -- inspired by abstracts that I find forceful and compelling. Such abstracts typically come with strong, focussed papers and force me to look at the intro (at least) despite the "inconvenience" of downloading the paper. Examples (from cryptography) I can think of include papers of Silvio Micali -- for example a paper called "CS proofs" (http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=365746), and to be a little extreme, "Primes in P". If I had some more patience, I could find you many such examples.

I belive the abstract should be as short as possible, not at all rambling, and if it gets to more than two paragraphs, it should be part of the intro.
(Reply) (Thread)