home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!ukma!cs.widener.edu!dsinc!netnews.upenn.edu!netnews.cc.lehigh.edu!news
- From: chess@watson.ibm.com (David M. Chess)
- Newsgroups: comp.virus
- Subject: Re: How to measure polymorphism
- Message-ID: <0002.9301281842.AA17847@barnabas.cert.org>
- Date: 18 Jan 93 21:03:05 GMT
- Sender: virus-l@lehigh.edu
- Lines: 24
- Approved: news@netnews.cc.lehigh.edu
-
- >From: Declan Malone <maloned@ul.ie>
-
- > Now how do you measure randomness in an
- >objective way? You can't really, and the irony of it is that the more
- >you try, and the more objective/detailed your description becomes...
-
- Grep Chaitin has done some excellent work on this (he had a general-
- audience paper in a reasonably recent Scientific American, I think),
- and has perfectly objective measures of randomness that also seem to
- jibe very nicely with intuition. The basic idea is that you can
- measure the randomness of a string of bits by finding the smallest
- program for some standard Turing Machine that produces those bits.
- For a very random bitstring, of course, the smallest program simply
- contains the string itself as data, and prints it out. For very
- non-random strings (0000000...), the corresponding program is
- obviously simpler. (This is a very rough description; if it seems
- wrong to you, it probably is, and you're best off finding the original
- papers rather than arguing with me about it!)
-
- This agrees rather nicely with Bill Arnold's suggestion that we
- measure polymorphism by how many lines of code we have to add to make
- a detector... *8)
-
- DC
-