Jeopardy Categories That Would Stump Watson

Like a lot of people I watched the IBM supercomputer Watson clobber a couple big-name Jeopardy champs a few weeks ago and I was quite impressed. The computer was able to use it impressive analytical engine to correctly guess the answers to dozens of questions. Sure, there were a few notable missteps, but they weren’t nearly enough to prevent Watson from laying down a silicon smack-down on to his human opponents. However, after the contest was over I got to thinking, things could very well have gone the other way if only the categories had been a little bit different.

Before & After

One of my favorite categories, before and after requires that the contestant combine two phrases or proper nouns into a single response. It is unlikely that Watson’s massive database would contain any humorous, but nonsensical responses, such as “Abraham Lincoln Towncar” or “Mark Cuban Missile Crisis”

Recent Events

It took years to get Watson programmed and up to speed so that it would have even a slight chance of getting the correct answer in time. According to the IBM programmers dozens of experts contributed uncountable amounts of information for Watson’s data banks. Remember that Watson wasn’t connected to the Internet, it only had access to data that was pre-programmed and this process takes time. Questions about very recent events probably would have stumped Watson as the programmers wouldn’t have had time to update the data. If the competition was filmed today possible answers could be “Won Superbowl XLV” or “2011 Best Picture Oscar Winner”.

Rhyme Time

Another category that requires putting together response into a seemingly nonsensical phrase, this category could (and actually did) give Watson fits.

Backwords and other Word Puzzle Categories

These types of categories require contestants to find words that are included in the answer, but are not necessarily related to the text of the answer. As I understand Watson’s algorithms it tries to find responses linked to certain key words in the answer, but if the text of the answer is only related through a word game, this type of lookup would fail miserably.

Humor or Satire-based Category

I think categories like “The Onion Describes the State“,  “Sort Through the Word Problem“, “Book Opposites” either rely on very strange phrasing or include enough meaningless text to stump even Watson.

Unusually Related Items

Examples of this category include “All Hail” (where the question must be the state from which those listed in the answer derive), “Alphabetically First“, and “We Own You“. Watson is the master of relating words and phrases, but if the relationship requested is not easily deciphered or if the items in the relationships might have multiple connections it would fail and fast.

Follow me on twitter or Add me to a Google+ Circle

Previous Post
Leave a comment

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: