Happy family

Find a legal form in minutes

Browse US Legal Forms’ largest database of 85k state and industry-specific legal forms.

Rules Lawyer: Watson Only Appears to Dominate Jeopary

I watched at midnight local time last night, and IBM Jeopardy Computer “Watson” appeared to thoroughly trounce its human opponents in the Double Jeopardy round, wracking up over $30,000 in winnings compared to paltry, sub-$6k sums for the humans  (the Final Jeopardy question was a different matter, noted below). My thoughts on the opening round are here.

But something sinister was going on. The crucial “buzz-in” was rigged in Watson’s favor. Read on.

It is somewhat generally known – and easily confirmed on line – that Jeopardy players must wait until host Alex Trebek finishes reading the question before they buz in to answer. The first to buz in gets to answer first. A little more digging reveals that a bank of lights next to the big Jeopardy video monitor board lights up when the buzzers go “hot.” If you click before the lights go on, your click doesn’t count and your buzzer is frozen for .25 seconds before you can try again. Previous champions have noted the importance of buzzing in first, since the contestants speed-read the question before Trebek finishes his reading, and often know the answer ahead of time.

What I haven’t been able to find is the actual mechanism for when the buzzer-enabling lights come on. Is it a Jeopardy employee throwing a switch when Trebek finishes pronouncing the last syllable, or something more technical like a sound wave monitor targeted on Trebek?

I knew that Watson has a buzzer that it clicks mechanically, but I also knew Watson can neither hear nor see (as Trebek stated yesterday). How does Watson know when buzzing in is OK?

There’s the rub. It was evident last night that all three contestants knew the answers most of the time, but Watson almost always buzzed in first. You could see Ken and Brad clicking disgustedly at their buzzers, but not reacting quickly enough.

This morning, the sentiment was all over the web that Watson was only a superior buzzer-in, not a superior Jeopardy player in the ordinary sense (though it was very good, and a remarkable achievement). Then I saw it confirmed in this Esquire interview, that Watson is hard-wired to receive an impulse when the buzzers go hot, and it can react immediately — and I mean near instantly, faster than humanly possible — when it gets the signal the buzzer is hot.

And perhaps more alarmingly for the humans standing on either side of it, Watson will be hard-wired into the game board. When those banks of lights come to life, Watson will be triggered by an electrical impulse. Jennings’s and Rutter’s nervous systems will be racing against a higher voltage, and it’s been a long time since anyone could turn off the lights and get into bed before the room got dark.

But Watson can.

It is not remarkable that a machine can be designed to have faster reaction times than a human. Still, this doesn’t detract too much from the Watson achievement, even if it does detract from bragging rights. To be fair, some sort of variable delay should be programmed into Watson’s buzz-in attempts, based on an average human reaction time. Then we could eliminate that variable and compare the mastery of riddling trivia.

In final Jeopardy, the category was U.S. Cities. Watson apparently didn’t account for this, because it answered the question (about largest airports named for a WWII hero and battle) with “Toronto” – which is not even a U.S. city. Ha. The humans got it right, preserving some dignity for us in the end. I’m sure the IBMers are having a field day analyzing why Watson couldn’t parse that one.

I think they play an entire game tonight, without the long IBM infomercials (which admittedly are pretty neat), and add their scores to their current total. Will be interesting to see if the humans can make up any ground, but if Watson’s quick-draw isn’t dialed down, it isn’t likely!

B