My Golden Globes ballot experiment/drinking game came to a close with the announcement that Three Billboards outside Ebbing, Missouri took the Best Drama award.
The ceremony itself will be written about by many for the next few weeks- the #MeToo solidarity, the power of Oprah’s speech, Natalie Portman’s best director shade. It was a largely quiet event, a Hollywood soiree in the aftermath of horrible occurrences within the industry (and, let’s be honest, within the world as a whole.) Powerful women-driven works took top prizes- The Handmaid’s Tale, Big Little Lies, Lady Bird– demonstrating the empowering talents of the sex so often relegated to second place. It was inspiring, and important, and I can only hope that this wave of feminism continues so that we have a new, positive normal again.
But now I’m going to shift to something far less serious- how a random number generator was able to correctly guess just as many Globes picks as a human being. Or more even, in the case of one of my friends.
Four of us hunkered down with a few bottles of Prosecco and ballots scribbled with our picks and the picks of our intelligent computer overlords. The drinking game itself remained largely intact as I developed it, with one addition. If your pick won you got to give a drink (and if your pick matched the computer and won, you take three but give three).
The game itself probably would’ve worked better with more people playing, but it was a fun diversion for the #DrunkOscars at any rate.
The more interesting of the outcomes was how the computers fared. Based on just pure statistics, the computer should have gotten 5 of the 25 correct. Our results were 1, 7, 7, and 9, placing the average at 6 correct, just a point higher than anticipated. How they fared against the human selections was more interesting. I had my best Globe pick year in ages, managing 17 of 25. Another friend, whose computer had only one correct guess, selected 14 of the recipients on her own. So far humans 2, robots 0.
Person number three drew an exact tie- 9 right for him, 9 for the computer. I should note that he generally fares better on those ballots, a demonstration of how odd the Globes can be.
Person four, the admittedly least in touch with film critics awards, got only three correct, while her RNG picked seven.
For the Golden Globes, the silliest of awards, it is completely possible for a computer to choose better than a human being. But in the end, the human average of 10.75 outscored the 6-pick robots. I should, of course, note that my sample size is excessively small, and my human test group is (in most cases) quite in touch with the goings-on of the awards season, so maybe those numbers would be closer in a larger and more random test, but it was a fun exercise none-the-less.
Should we carry the experiment forward to the Academy Awards themselves? I’m not so sure. I would need to get individuals involved who know little about those particular awards, as the Oscars tend to be much more predictable than the Golden Globes- DGA, WGA, and a handful of critics choice circles tend to be incredibly strong indicators on who will win the ultimate prize.
Either way, Oscar nominations arrive in exactly two weeks, and then we can let the next level fo film fun begin.