Microsoft Predictions for the Oscars in 2016
March 5, 2016
I know that Microsoft has a prediction system. I don’t pay much attention to Bing or other Microsoft technology. I understand that I am an analog brontosaurus.
The point, for me, is that predictive systems need to be based on numerical recipes which perform in a consistent manner. One can fiddle the definition of “consistency,” but when a predictive system is driving an autonomous vehicle, identifying a treatment for death, or identifying the worthy individuals as Oscar winners—the systems have to be pretty darned accurate.
The write up points out:
In 2015, Microsoft Bing’s prediction engine nailed the Academy Awards, guessing 20 out of 24 Oscar winners. The year before that, it did even better, going 21 for 24.
But in 2016, the Bingster, according to the write up:
only guessed 71% of the winners correctly, with 17 out of 24 correct choices.
In the real world, Bing’s predictive methods can chop out some highly probable losers. That may be quite useful for some applications like narrowing down a list of potential contractors.
For certain real world applications involving risk to life and limb in some far off war zone, I am not sure the Bing predictive engine will be number one on my list of systems upon which to rely.
The write up does not share my opinion, describing the result as “pretty okay.” Well, for me, a two thirds outcome is not pretty okay. It is below average, almost C minus or D plus territory.
The consumer angle suggests that Microsoft in terms of search and content processing may be prepping to become the next Yahoo.
Stephen E Arnold, March 5, 2016
Comments
-
- Subscribe to Beyond Search
Feature archive
News archive
-
