Comments on: The long tail of science: Good, bad, or ugly? http://blog.oddhead.com/2009/03/21/the-long-tail-of-science-good-bad-or-ugly/ Musings of a computer scientist on predictions, odds, and markets Thu, 09 Feb 2012 08:47:34 +0000 hourly 1 http://wordpress.org/?v=3.3.1 By: David Pennock http://blog.oddhead.com/2009/03/21/the-long-tail-of-science-good-bad-or-ugly/#comment-454 David Pennock Mon, 06 Apr 2009 14:41:11 +0000 http://blog.oddhead.com/?p=442#comment-454 Manual trackback: <a href="http://orgtheory.wordpress.com/2009/03/23/the-very-long-tail-of-science/" rel="nofollow">"A few publications not only dominate the attention of the relevant discipline but an equally few seem to be the most profound in their practical application. What does this mean about the scientific program? How much waste is there in science? Pennock offers several different views on the “(very) long tail” of science, one of which is more or less similar to my own view."</a> Manual trackback:

“A few publications not only dominate the attention of the relevant discipline but an equally few seem to be the most profound in their practical application. What does this mean about the scientific program? How much waste is there in science? Pennock offers several different views on the “(very) long tail” of science, one of which is more or less similar to my own view.”

]]> By: Daniel http://blog.oddhead.com/2009/03/21/the-long-tail-of-science-good-bad-or-ugly/#comment-458 Daniel Fri, 03 Apr 2009 21:11:26 +0000 http://blog.oddhead.com/?p=442#comment-458 Just came across this article http://www.technologyreview.com/blog/post.aspx?bid=354&bpid=23289 and from the original source (physics world) http://physicsworld.com/cws/article/print/38468 "More generally, Taleb’s basic investing strategy — which could easily be translated into research terms — is to put a fair fraction of funds into very conservative processes that will not lose their value, even if they have little chance of producing big gains; and to put a small but reasonable fraction into high-risk, high-reward settings, thereby gaining exposure to the potentially enormous gains from these investments. These may be unpredictable in detail, but the statistics makes the expected long-term pay-off very high. " Just came across this article
http://www.technologyreview.com/blog/post.aspx?bid=354&bpid=23289

and from the original source (physics world)
http://physicsworld.com/cws/article/print/38468

“More generally, Taleb’s basic investing strategy — which could easily be translated into research terms — is to put a fair fraction of funds into very conservative processes that will not lose their value, even if they have little chance of producing big gains; and to put a small but reasonable fraction into high-risk, high-reward settings, thereby gaining exposure to the potentially enormous gains from these investments. These may be unpredictable in detail, but the statistics makes the expected long-term pay-off very high. “

]]> By: Daniel http://blog.oddhead.com/2009/03/21/the-long-tail-of-science-good-bad-or-ugly/#comment-457 Daniel Mon, 23 Mar 2009 20:55:39 +0000 http://blog.oddhead.com/?p=442#comment-457 By your definition it's just ugly. But most human behavior is going to be governed by an extreme distribution with a long tail. What we can do is create a better system which awards good scientists (trustworthy hardworking people) as well as good science (experiments are reproducible) (There surely many more metrics we can use here) I think at a minimum, we need to distribute some money randomly, to try and hit these people lost in the long tail. That is, take some money away from the most popular ideas, and fund more new, off-the-wall, and unpopular ones. Yes, breakthroughs are rare and unpredictable, but their impact may be of paramount significance, as opposed to piling more money onto popular ideas to make incremental advances. By your definition it’s just ugly. But most human behavior is going to be governed by an extreme distribution with a long tail. What we can do is create a better system which awards good scientists (trustworthy hardworking people) as well as good science (experiments are reproducible) (There surely many more metrics we can use here)

I think at a minimum, we need to distribute some money randomly, to try and hit these people lost in the long tail. That is, take some money away from the most popular ideas, and fund more new, off-the-wall, and unpopular ones. Yes, breakthroughs are rare and unpredictable, but their impact may be of paramount significance, as opposed to piling more money onto popular ideas to make incremental advances.

]]> By: Mike Giberson http://blog.oddhead.com/2009/03/21/the-long-tail-of-science-good-bad-or-ugly/#comment-456 Mike Giberson Sun, 22 Mar 2009 14:25:59 +0000 http://blog.oddhead.com/?p=442#comment-456 I'm voting for "ugly." Yes, the cost of production for most academic papers is probably well in excess of the value produced. Maybe one in ten is useful for some small purpose, one in hundred more generally interesting, and one in a million provides a significant breakthrough. Only, it is impossible in advance to know which possible research articles will turn into the one useful paper in ten rather than the nine. I’m voting for “ugly.”

Yes, the cost of production for most academic papers is probably well in excess of the value produced.

Maybe one in ten is useful for some small purpose, one in hundred more generally interesting, and one in a million provides a significant breakthrough. Only, it is impossible in advance to know which possible research articles will turn into the one useful paper in ten rather than the nine.

]]> By: Will Dwinnell http://blog.oddhead.com/2009/03/21/the-long-tail-of-science-good-bad-or-ugly/#comment-455 Will Dwinnell Sun, 22 Mar 2009 08:56:04 +0000 http://blog.oddhead.com/?p=442#comment-455 I am forced to vote "bad" on this. The vast majority of peer-reviewed technical papers I've read are hopeless: they are solutions to non-existent problems or trivial variations on previous work. While the academic and research communities perform an important function in filling out the edges of very large innovations, true innovation (Einstein, Shannon, etc., as you mention) is not a product of the publish-or-perish environment. What benefit, then, does that ocean of paper provide? I am forced to vote “bad” on this.

The vast majority of peer-reviewed technical papers I’ve read are hopeless: they are solutions to non-existent problems or trivial variations on previous work. While the academic and research communities perform an important function in filling out the edges of very large innovations, true innovation (Einstein, Shannon, etc., as you mention) is not a product of the publish-or-perish environment. What benefit, then, does that ocean of paper provide?

]]>