Wine by the numbers

by Jeff Miller of Artisan Family of Wines (Seven Artisans, Sly Dog Cellars, Red Côte)

jeff-smFor those of you who aren’t political junkies, the name Nate Silver may not mean a whole lot.  But for those of you who are, he has become something of the seer for election forecasting.  You can find his blog on the New York Times website, at

Somewhat amazingly, Silver predicted that Obama would carry every state that in fact he did, with the exception of Florida, which he rated as 50/50 (and which turned out to be very close to a tie).  None of which, you might note, has very much to do with wine.  But Dr. Vino also noted Silver’s accuracy in predicting election results and then went on to ask:

“Will Silver’s sabermetrics extend to the wine world? Will sommeliers start saying, “Based on our algorithms, there’s a 91% chance you will like this wine.”

Dr. Vino ends up answering this question in the negative:

“In all seriousness, there’s a lot of subjectivity in wine that masquerades as objectivity under the false precision of point scores and distant drinkability windows. Just because you see a number in wine, doesn’t mean that it’s empirical.”

But I’m not at all sure that a algorithm couldn’t be devised that would do a pretty good job of predicting which wines a particular consumer would like.  What is hard to fathom is that most wine critics are so out of sync with what really determines whether someone will like a wine or not.

I found this tasting note on

Fantastic Temecula Valley AVA Merlot with a soft balance and full palate coverage. Jammy dark fruit notes on the bouquet, light american oak on the finish with hints of plum and bosenberry [sic]. Lacks the usual Merlot or Cab Sauv strawberry notes and spice which is nice for this Merlot.

As tasting notes go, this isn’t bad, but it’s pretty typical in missing the important points.  The main emphasis, as is the case with pretty much every tasting note, is on the flavors.  There’s lots about jam and plum and boysenberry and strawberry and spice.

But for me, and I’m willing to extrapolate from me to most people, flavor is just one of several elements that go into figuring out why I like, or don’t like, a wine.  And I think flavor is less important than the other factors.  And unlike flavor, those other factors are more objectively quantifiable in ways that a person can actually use to pick out a wine.  Or plug into an algorithm.

I’m going to list a few of those factors.  As always, I’m really concentrating on red wines.

Tannins:  I have a friend that hates tannin.  So I won’t spend a lot of time tasting him on Petite Sirahs.  What’s the point?  I, on the other hand, love tannin.  Tannins are something most people will agree about (i.e., is this wine tannic, are the tannins soft or hard?).

Acidity:  I think all wines should be relatively high in acid.  But I’m not in the majority on this.  Most people like a “soft” (i.e., low acid) wine.  But whether you like high acid wines, or low acid wines, chances are pretty good you’ll be able to identify which wines are high and which are low acid wines, and figure out which one you like.  It is pretty objectively quantifiable by palate, even if it’s more quantifiable by laboratory analysis.

Concentration:  Pinot Noir is a less concentrated wine than Petite Sarah. My wife, who likes light-bodied wines, will almost always opt for a Pinot Noir.  I like both (though I probably veer slightly toward concentration), but I think most people trend pretty clearly one way or the other.

Alcohol:  Higher alcohol wines tend to have a sweetness and body that lower alcohol wines lack.  That’s not to say low alcohol wines are inferior.  It’s all a matter of taste.  Sweetness and body may not be your cup of tea.  But, again, alcohol content is objective.

Flaws:  “Flaws” is a nebulous concept, since many people like things that others consider flaws. I generally like my wines pretty clean.  I don’t want Brett.  I don’t want canned corn.  I don’t want rubber.  Some people like some of those things, but, for present purposes, I’m talking about what I like.

So if you set up an algorithm that gave points for tannins (not too harsh), acidity, concentration, lack of flaws, and low to moderate alcohol, you’d probably be able to come to a pretty good prediction as to whether I, Jeffrey Miller, would like a wine that scored high.  I don’t think the algorithm would perfect.  There would be wines that it would predict I would like that I wouldn’t, and vice versa.  But I think it would, on the whole, be pretty accurate.

Unfortunately, tasting notes are the way they are because it’s what readers want to read, even if they don’t shed a whole lot of light on whether you’ll like the wine.  A tasting note that said:  “Medium but soft tannins, good acidity, low concentration, low alcohol, slight Brett” isn’t the type of tasting note that’s going to show up in the Wine Spectator anytime soon.  But it’s the type of note that really would shed a whole lot more light on whether you, or any other consumer, would like the wine.

Share and Enjoy:
  • Print this article!
  • Digg
  • Facebook
  • Google Bookmarks
  • Blogosphere News
  • E-mail this story to a friend!
  • LinkedIn
  • MSN Reporter
  • Twitter
  • Yahoo! Bookmarks
  • Yahoo! Buzz

Leave a Reply