Wine tech algorithms are often useless. here’s why

Wine and the Internet go hand in hand.

For many wine lovers, there is something at the heart of these two businesses that seems contrary to why they drink wine in the first place. Inevitably, they say, these databases end up favoring mass-produced wines, as these will inevitably generate more reviews and be available in more stores. Unique wines from small, already obscure producers are becoming even more difficult to discover in these large online tangles.

Another argument against such zoomed-out searches is that they can reduce the complicated and unknown matrix of an idiosyncratic wine to a few basic qualities. Instead of learning about the history of this family estate, or the traditions of the region, or the method of making, you are only told if the wine tastes like green apple or cherry. Pix had hoped to perfect the art of the flavor search algorithm, allowing you to search for wines based on the flavors you’re looking for, though he never really made it work.

I sympathize with these anti-wine-tech sentiments – to some degree. I’ve always been wary of, for example, online personality quizzes that try to associate you with your “perfect” wine based on other foods you like. (Apologies to long-time readers who have endured my rants about this before.) These quizzes mistakenly assume that people will want to taste the same type of wine over and over again. And they seem to place way too much emphasis on individual flavors.

All of these technological efforts attempt to make wine easily intelligible to the masses. To simplify. But maybe the problem is that wine is not simple.

As one reader summed it up, emailing in response to Pix’s article: “The real reason wine technology keeps failing is because there’s nothing that technology can offer to wine”. (It bears repeating that Vivino, which has raised hundreds of millions of dollars in venture capital, is not failing.)

It happened in a Twitter feed last week by Aaron Ayscough, the veteran natural wine blogger who writes the excellent newsletter “Do not drink poison.” He explained why “a search algorithm for wine flavor descriptors is a hellish idea”: primarily, these individual flavors – say, peach – are “the most fleeting aspects of a wine”. Ultimately, these are distractions from the things that really make us like one bottle over another, Ayscough said, like its balance and harmony.

Ayscough’s antidote to this dumb approach to wine recommendations is to get to know your winemaker. Although “this is often taken as obscure information”, he tweeted, knowing personally the makers of the food and wine we consume “this is how humans have sourced much of their diet until good in the 20e century.”

It’s a nice thought, but it’s idealistic, divorced from reality.

There will always be a market for those like Ayscough and me, who are interested in learning more about the stories behind the wines we drink and who appreciate bottles made with integrity (however we choose to define that). . No matter how passionately wine writers promote this way of thinking, most people who buy wine will never ask these questions. Yet, I still believe that there should be effective ways for these people to communicate what they want. Allowing them to seek out a wine that tastes like “blackberry jam” is clearly not the solution. Still, there has to be a way to make the complicated task of finding a wine to drink a little more accessible, right?

Will that solution come in the form of a new wine technology company? Maybe. I’ll wait.

Sharon D. Cole