home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!noc.near.net!hri.com!spool.mu.edu!wupost!udel!princeton!csservices!volkl!egs
- From: egs@cs.Princeton.EDU.Princeton.EDU (Emin Gun Sirer)
- Newsgroups: rec.puzzles
- Subject: Re: Guessing at random numbers.
- Message-ID: <1992Dec24.102342.6627@csservices.Princeton.EDU>
- Date: 24 Dec 92 10:23:42 GMT
- References: <1992Dec24.015625.3450@csservices.Princeton.EDU> <1QSM!4XMK@linac.fnal.gov>
- Sender: news@csservices.Princeton.EDU (USENET News System)
- Reply-To: egs@cs.Princeton.EDU.Princeton.EDU (Emin Gun Sirer)
- Organization: Department of Computer Science, Princeton University
- Lines: 26
- Originator: egs@volkl
-
- Matt Crawford writes:
- >I assume that we play the game for several iterations.
- >
- >As long as you don't alter the distribution that your generator picks
- >from between iterations....
- >
- >I guess "high" or "low" according to whether the number you give me is
- >higher or lower than the median of the numbers given so far.
-
- A good guess given my incomplete original description of the problem.
- Do not assume that we play the game for more than one iteration (though I
- would like to hear your ideas on multiple-iteration games). The answer
- I have can be "proved" to be better than average even for single games.
-
- Also, I can change my distribution for each iteration. If the two
- numbers I pick are equal, I discard one and pick a new one. Degenerate
- distributions (e.g. rand(){return 1}) are not allowed. This is not an adversary
- problem, however. I will not look at your guess history and modify my
- behaviour (and doing so shouldn't help you in nay way I can foresee) to
- thwart your scheme. In fact, as I said before, we may have no history at
- all and still play with better than average odds (for you).
-
- I hope this covers everything, sorry about the lack of rigor in the
- first description.
-
- Gun.
-