tag:blogger.com,1999:blog-8890204.post6487642315192103052..comments2024-03-10T05:26:42.148-04:00Comments on My Biased Coin: What's Important in AlgoirthmsMichael Mitzenmacherhttp://www.blogger.com/profile/06738274256402616703noreply@blogger.comBlogger8125tag:blogger.com,1999:blog-8890204.post-3849951421785679802014-06-01T10:03:47.561-04:002014-06-01T10:03:47.561-04:00Anonymous X: I was talking about Algoirthms. Wha...Anonymous X: I was talking about Algoirthms. What are Algorithms? (Just kidding.)<br /><br />Anonymous June 1: I'm not surprised by the "pragmatic" comment. I think TCS should have room for a wide spectrum of researchers, certainly including those that aren't interested in practical implications at all. The "T" is for "Theoretical". But I think I'm on record as noting that TCS has, culturally, tilted pretty far in the "non-practical" direction, so that a comment like Anon May 26 is both unsurprising and, perhaps to many, uncontroversial. Michael Mitzenmacherhttps://www.blogger.com/profile/06738274256402616703noreply@blogger.comtag:blogger.com,1999:blog-8890204.post-33626782506748168102014-06-01T00:39:02.911-04:002014-06-01T00:39:02.911-04:00"Pragmatic considerations are of course impor..."Pragmatic considerations are of course important but academic TCS has to advance the frontiers."<br /><br />how did TCS get to this point...Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-8890204.post-33284828023649516152014-05-28T11:30:28.556-04:002014-05-28T11:30:28.556-04:00Anon #5: I think I disagree with pretty much ever...Anon #5: I think I disagree with pretty much everything in your comment. Since that's rare, I'll respond. <br /><br />"...progress on fundamental and difficult problems in algorithms and theoretical computer science is not going to happen by carefully implementing existing algorithms"<br /><br />I just don't think this is correct. Often implementing existing algorithms is what gives you insight allowing for improvements -- including significant theoretical improvements. I could spout examples from myself and students, but I'm sure you can find some of your own. The relationship between theory and practice is a 2-way street. Sure, implementation isn't the ONLY way to make progress, but it's also a way that should not be ignored. <br /><br />"Pragmatic considerations are of course important but academic TCS has to advance the frontiers."<br /><br />In specifically algorithms, advancing the frontiers should definitely include "pragmatic considerations". "Use" has always been at least a criteria for successful algorithmic work, making pragmatic considerations important. <br /><br /> Michael Mitzenmacherhttps://www.blogger.com/profile/06738274256402616703noreply@blogger.comtag:blogger.com,1999:blog-8890204.post-14674162731266673392014-05-26T20:34:34.172-04:002014-05-26T20:34:34.172-04:00Math envy or not, progress on fundamental and diff...Math envy or not, progress on fundamental and difficult problems in algorithms and theoretical computer science is not going to happen by carefully implementing existing algorithms. Pragmatic considerations are of course important but academic TCS has to advance the frontiers.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-8890204.post-35820660120795991832014-05-24T20:36:39.357-04:002014-05-24T20:36:39.357-04:00It's important to tell that 'Algoirthms&#...It's important to tell that 'Algoirthms' is incorrect.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-8890204.post-15402690320987193972014-05-22T19:19:02.791-04:002014-05-22T19:19:02.791-04:00this is a great high voted question on that "...this is a great high voted question on that <a href="http://cstheory.stackexchange.com/questions/4491/powerful-algorithms-too-complex-to-implement" rel="nofollow">"powerful algorithms too complex to implement"</a> & rjlipton has a great blog on the subject also re "galactic algorithms". gotta blog on this sometime & include your blog!Anonymoushttps://www.blogger.com/profile/10048739391910999672noreply@blogger.comtag:blogger.com,1999:blog-8890204.post-82987314229319589322014-05-22T14:57:10.911-04:002014-05-22T14:57:10.911-04:00Thanks for the link to the Knuth Q&A - some ve...Thanks for the link to the Knuth Q&A - some very interesting material there. The topic of the missing implementations of published algorithms has been discussed before (on this blog I think). <br /><br />It seems to me that this is the theoretical CS version of the non-reproducibility crisis currently occurring in disciplines such as psychology. <br /><br />Have you seen any progress over the last 5 years?<br /><br />Maybe <a href="http://mozillascience.github.io/code-research-object/" rel="nofollow">giving DOIs for code</a> will allow people to get citation credit and go some way to improving the situation. It seems that the reward system of scholarship is badly broken, when obscure "improvements" of algorithms that could never be implemented get points, but careful comparison of actual running times of useful implementations of more important algorithms does not. Math envy is a major problem in many disciplines!Mark Wilsonhttps://www.blogger.com/profile/16382201587698895101noreply@blogger.comtag:blogger.com,1999:blog-8890204.post-58958359724503642882014-05-22T01:11:10.884-04:002014-05-22T01:11:10.884-04:00I think that both theory and practical algorithms ...I think that both theory and practical algorithms should be encouraged, for different reasons. The main problem is in the middle: a new algorithm that is simpler that all already-published algorithms but not at the level of being implementable faces several problems when trying to publish it.Anonymousnoreply@blogger.com