[Beowulf] OT: public random numbers?

Steve Cousins cousins at umit.maine.edu
Mon Aug 15 13:59:11 PDT 2011

Hi David,

Can you give us more information about what you are doing? I'm getting 
curious about what problem you are working with that requires these 


> We have not discussed election results (votes per candidate), but those
> are, ironically, really unsuitable for this, even though statistically
> the final set of digits should have a lot of entropy.  Mostly election
> numbers are a problem because they may be revised for long periods after
> the election, and the numbers could almost always be forced to shift by
> a challenge by one of the candidates.  Every recount will come up with a
> slightly different result.  Examples: the Coleman vs. Franken senatorial
> contest in Minnesota, or Bush vs. Gore in Florida.
> So I'm leaning towards sports scores, as those are generated in full
> view of a multitude of witnesses (often numbering in the millions).  It
> would be extremely difficult to rig the absolute final score.  It might
> be possible to rig the winner, or even the point spread, but to rig the
> absolute score in a high scoring game like basketball, would be
> exceedingly difficult, and would likely be obvious to even the casual
> observer.  To rig every digit in the final score of every game played on
> a given day should be pretty close to impossible.

More information about the Beowulf mailing list