Tired of snotty wine writers? Tired of indecipherable wine reviews? Tired of trying to figure out whether a 90-point wine is actually better than an 88-point wine?
Have no fear. The computer is here.
Researchers at Northwestern University, the Wine Curmudgeon’s alma mater, have developed software that allows computers to write stories for publication. Their company, Narrative Science, has developed an algorithm that takes the facts of an event and turns it into a short story. The process was developed using sports events, which would actually be easy for computer software to write — something I know from having done it for more years than I care to admit. All you need is the final score and who did what, and even a 15-year-old high school student in suburban Chicago can write an effective paragraphs-long story detailing who beat who and how they did it.
I contacted Narrative Science last week to ask them about what they do and how they do it, and especially if the algorithm would work for wine. No one got back to me, but if they do, I’ll update the post.
Because it seems that the process should work for wine — allowing for the tasting problem, since computers don’t have palates. All a winery has to do is to supply the basic facts about the wine, like reviews, scores, tasting notes and technical details, and the algorithm should be able to do its stuff. Writing a wine review, the actual writing part, isn’t all that different from writing a 250-word story about a high school basketball game.
Think of the possibilities — ComputerWines.com, an entire web site devoted to computer-generated wine reviews, with thousands and thousands and thousands of reviews completed in an instant and updated for every vintage, all searchable and sortable and adjustable. Talk about getting on the front page of a Google search.
And the fact that computers can’t taste wine might actually be an advantage. There would be no need for a reviewer to ponder about whether a wine has berry flavors; instead, the algorithm could parse one blackberry, one blueberry and one black cherry from the submitted information and come up with whatever it would come up with.
Better yet, the cost to do the reviews would be negligible, and the business model would call for the wineries to pay for the reviews. Which they would no doubt be happy to do. It would be far cheaper than than sending out thousands of dollars worth of samples, and far less taxing than sucking up to the Winestream Media. Throw in a disclaimer in small print — “These reviews were computer-generated and no one actually tasted the wine” — and the Wine Curmudgeon sees big-money possibilities. Real big money. Moving to Burgundy and being snotty to people big money.
Ah, but many of you are wondering: Is the Wine Curmudgeon being serious? Or is this another example of his well-known crankiness? Let’s put it this way — given the uselessness of so many wine reviews, and given how little respect the industry has for the consumer, how much worse would ComputerWines.com actually be?