In article <[EMAIL PROTECTED]>,
        [EMAIL PROTECTED] writes:
> Keith did this several times during the Get Even game:
> 
> http:[EMAIL PROTECTED]/msg01257.html
> 
> "If you suspect your solution is unsound, please please do what
> Keith did, and try to find some test data that breaks it.
> At the end of the game, the other players have the right to
> find test data that breaks any solution and that solution will
> then be disqualified, so it is much better to find that out now."
> 
> I admit I was shocked that Ton's winning entry did not
> pass the test program. That would never happen in a
> game run by me. If someone submitted a valid solution
> that failed the test program, I would immediately and
> urgently fix the test program so that it passed.
> That is because I always thought that passing the
> test program was a necessary (but not sufficient)
> condition of each and every entry.
> 
I only started playing with the "-" thing when I started to concentrate
on improving the tie-break for if Eugene would shave one stroke.
(it doesn't save anything, just improves the tie-break by 0.008).
So my 150 only stopped working during the last few hours, and it
was probably too little time to still bother changing the test-program.

When in earlier program versions i found cases where I could abuse
the testcase and I told this, the test program was in fact extended.

So the judges were still following this good old tradition.

Reply via email to