[SumZero]

I think there is a time to go for those gold mine (or as I like to call them, longball) type questions, but generally only after your opponent hits their goldmine question. That is binary search is the best to do if you can. But if you opponent does a non-optimal search much of the time you can stick with binary search and win. Think of picking numbers between 1-100 where you can ask "is it higher/lower than X" OR you can ask "is it X?". If you start out and say "is it higher than 50" and get yes. Then your opponent says "is it less than 20?" and you have to say yes. Now your opponent has only 19 numbers left to pick from while you have 49. And because each time you pick your midpoint question you aren't also winning if they get the midpoint it may be time for you to launch a "is it less than 61" or "is it more than 90" type question in order to try to out longball the longball strategy. But 80% of the time their longball will fail and you are way ahead and can stick with your binary search. And at some point when they get really close to the number when they are ahead of you it pays to just start guessing is it x? rather than trying to narrow your range. In esscence because all you are doing is trying to beat them or else you lose. You don't get credit for being close to guessing the right answer if you lose.