If you have a 1 in 60*24 chance of guessing correctly then after 1000 guesses (potentially against different people) you have a 50 chance of being correct on one.
Your chance of being wrong in all 1000 guesses, assuming you guess randomly (not ensuring you never make the same guess twice) is (1-1/(60* 24))^1000
That's about .5
This is based on the assumption that you are guessing for a different person each time, so you can't increase your odds by eliminating any guesses you've already made. If you're guessing for the same person every time, you just need to guess half the possibilities. You can confirm this by realizing:
Your chance of being wrong in a guess is 1-1/(60* 24-x) where x is the number of guesses you have already made.
The product of 1-1/(60* 24-x) from x=0 to x=60* 24/2-1 is .5
Edit: hackernews formatting is weird