tag:blogger.com,1999:blog-4068183698747623113.post6113222424906615207..comments2023-10-29T10:40:34.638-04:00Comments on A CS Professor's blog: Conferences and Arxiv. Utilitarian values.Claire Mathieuhttp://www.blogger.com/profile/10957755706440077623noreply@blogger.comBlogger11125tag:blogger.com,1999:blog-4068183698747623113.post-12981427133775514712012-01-12T03:53:32.236-05:002012-01-12T03:53:32.236-05:00I never consider arxiv publications in my research...I never consider arxiv publications in my researches. But I was hoping to get some info about them and what benefits I can have from them.Mike Warrenhttp://www.themikewarren.comnoreply@blogger.comtag:blogger.com,1999:blog-4068183698747623113.post-13527580673571131952012-01-09T08:07:20.877-05:002012-01-09T08:07:20.877-05:00I guess I don't see whether I should interpret...I guess I don't see whether I should interpret the abstract as being instrumental or intellectual in Oded's sense. I might guess he'd say it was instrumental, while I'd call it intellectual, since it seems to me the only implicit motivation is "this is an intellectually interesting problem".Technical Analysishttp://www.chartsmakemoey.comnoreply@blogger.comtag:blogger.com,1999:blog-4068183698747623113.post-85723106376936448062011-11-30T20:49:00.796-05:002011-11-30T20:49:00.796-05:00Anonymous: a little bit of marketing is not bad. F...Anonymous: a little bit of marketing is not bad. For the outsider, what you wrote sounds pretty flat. Only the expert can realize that there is something interesting hidden behind the bland statement. Trying to convey the excitement is a good thing... but it's hard to do it without sounding obnoxious.Claire Mathieuhttps://www.blogger.com/profile/10957755706440077623noreply@blogger.comtag:blogger.com,1999:blog-4068183698747623113.post-64195701256116383262011-11-30T20:11:25.634-05:002011-11-30T20:11:25.634-05:00To me, the issue with this abstract is not whether...To me, the issue with this abstract is not whether it is "instrumental" or "intellectual"; it is the blatant marketing going on. Do we really need to be told the TSP is "the most fundamental" and "the most studied" problem?. Does it help the reader to know that "for more than 30 years" there was one bound, while "significant progress" was recently made for an "important" special case? What would be wrong with the following abstract?<br /><br /><em>In this paper, we provide an improved analysis for the approach to approximating TSP that was introduced by Momke and Svensson, yielding a bound of 13/9 on the approximation factor. Our techniques also give a bound of 19/12+epsilon for any epsilon>0 for a more general Travelling Salesman Path Problem in graph metrics.</em>Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-4068183698747623113.post-78802669291840477932011-11-30T19:51:22.426-05:002011-11-30T19:51:22.426-05:00Sometimes the constants themselves have meaning: ...Sometimes the constants themselves have meaning: Beating 3/2, even if it was by a tiny 10^{-6} told you that it had to be doing something really new, as in the Gharan, Saberi, Singh paper. The authors would not have needed to explain why this had to use really new ideas. Momke and Svensson's algorithm beats this by a lot but would not have been so interesting if it had not introduced new ideas. <br /><br />One of the best student papers from FOCS by Hertli gave a new best exponential upper bound for 3-SAT of the form something like 1.30704^n which improves the best constant in the base from 1.32065 and does this in a memorable way. (Though unlike 3/2, I would have had little idea whether 1.30704 was an improvement, especially since some of these results are stated using the log_2 of the base.) The title of the paper is "3-SAT Faster and Simpler - Unique-SAT Bounds for PPSZ Hold in General". The title tells you that something interesting is going on (if you know what PPSZ is) but the abstract and intro sure sound as though they would score as "instrumental". I am not sure how one would write the paper differently in these sections.Paul Beamenoreply@blogger.comtag:blogger.com,1999:blog-4068183698747623113.post-90578389440520348022011-11-30T17:55:12.284-05:002011-11-30T17:55:12.284-05:00No, Claire, I disagree. I'm happy with people...No, Claire, I disagree. I'm happy with people improving the constants, and I think that as a general rule important, and well worth announcing in an abstract. (See the current debate on Scott Aaronson's blog on the matrix multiplication result...) It just helps if the constants mean something. In this case, you appear to want the new constant to be associated with a new insight or technique that you don't think is being expressed in the abstract.Michael Mitzenmacherhttps://www.blogger.com/profile/02161161032642563814noreply@blogger.comtag:blogger.com,1999:blog-4068183698747623113.post-41329638112050188152011-11-30T12:58:06.791-05:002011-11-30T12:58:06.791-05:00I think everyone is actually in agreement: we don&...I think everyone is actually in agreement: we don't like the "battle of constants" style of presentation of results.Claire Mathieuhttps://www.blogger.com/profile/10957755706440077623noreply@blogger.comtag:blogger.com,1999:blog-4068183698747623113.post-75329486740084412192011-11-30T12:51:59.827-05:002011-11-30T12:51:59.827-05:00Whoops. Just realized I pointed to a competitive ...Whoops. Just realized I pointed to a competitive analysis blog post, when this is an approximation algorithm. The comments (as well as the post), though, tie into the same issues. Apologies.Michael Mitzenmacherhttps://www.blogger.com/profile/02161161032642563814noreply@blogger.comtag:blogger.com,1999:blog-4068183698747623113.post-1081549737287514022011-11-30T12:43:02.993-05:002011-11-30T12:43:02.993-05:00Claire,
I guess I don't see whether I should ...Claire,<br /><br />I guess I don't see whether I should interpret the abstract as being instrumental or intellectual in Oded's sense. I might guess he'd say it was instrumental, while I'd call it intellectual, since it seems to me the only implicit motivation is "this is an intellectually interesting problem".<br /><br />But really, I think competitive analysis is an odd case all its own.<br /><br />http://mybiasedcoin.blogspot.com/2008/02/conference-reviewing-another-rant-on.htmlMichael Mitzenmacherhttps://www.blogger.com/profile/02161161032642563814noreply@blogger.comtag:blogger.com,1999:blog-4068183698747623113.post-7897381168936161462011-11-30T12:09:56.385-05:002011-11-30T12:09:56.385-05:00Don't get me wrong, Chandra, I am interested i...Don't get me wrong, Chandra, I am interested in that paper, and it is definitely worthwhile! In fact I would have been happy to be an author of this paper. Doing it "right" has value, for an important problem and advance! Moemke and Svensson made an important advance, but they didn't do that part quite right; this paper does it right: that's valuable.<br /><br />I'm sorry that didn't come across in my post.<br /><br />Rather, it's just the style of the abstract that didn't convey that but only the "winning another battle in the war of constants". I only took issue (slightly) with the style, not the result.Claire Mathieuhttps://www.blogger.com/profile/10957755706440077623noreply@blogger.comtag:blogger.com,1999:blog-4068183698747623113.post-52439295803497559152011-11-30T11:23:08.821-05:002011-11-30T11:23:08.821-05:00There is place for all types of research. Improvin...There is place for all types of research. Improving an approximation ratio is one concrete way for some one to understand previous work on the problem. Senior researchers seem to forget some times that students and younger researchers have to publish what they can to get a degree, a job etc. Also, research progress does not happen simply by thinking big thoughts - people play around with ideas and along the way some times we get lucky.Chandrahttps://www.blogger.com/profile/00057069075567569157noreply@blogger.com