10 ms·
while yes human reaction times are nowhere near as good as those achievable with computers I think the physical pressing of the buzzer was enough to level the p
by patrickod 16y ago
while yes human reaction times are nowhere near as good as those achievable with computers I think the physical pressing of the buzzer was enough to level the playing field. I can't remember where I read it but it seems that at most Watson had 100ms of an advantage if he was confident of his answer before the buzzers opened. I also can't think of a consistent way of getting rid of this time gap.
- gojomo 16y agoParts of the game were eliminated to give Watson a chance: audio/video clues, and categories that required extra explanation. If the exact timing of buzzing in – and not 'first' but 'first after a light goes on'– gave an overwhelming advantage to Watson, that could be eliminated, too. For example, why not just 'first to buzz in'? (That is, no penalty for being early?) Let Watson's parsing of the question text race the humans' sight-reading. It looked to me like the humans usually won the buzzer on short questions, which strongly suggests it was the slowness of reading long questions that put it on par with the humans in knowledge, and then it crushed them on buzzing-after-the-light precision.
- alanfalcon 16y agoYou can program a computer to wait until it has high confidence before buzzing in, but would you trust a human to wait to buzz in until he has already chosen his answer? In many situations, a contestant will buzz in knowing they're likely to be able to determine the answer in the five seconds they're given after being called upon. You'd be trading one player's perceived advantage for another.