About two decades ago, a bunch of economists at Princeton formed a wine-tasting club. The tastings would be blind. Nobody would know what was in the glasses until the end, so the fancy labels wouldn’t influence their judgment.
The Princetonians liked red wines best, especially those from the French region of Bordeaux, but because the club members were all academics what they really liked was data. They wanted to see if they could articulate some kind of reproducible results that would separate the good wines from the bad. So one of the founders, Richard Quandt, wrote a computer program that would run the appropriate statistical tests on their ratings — agreement among tasters, correlations, and so on.
Quandt is retired now, with several microeconomics textbooks on his CV (as well as a book on racetrack betting and another on how dogs think). He curses a lot, in a soft Russian accent. “Every time we had a wine tasting we’d run the program, have some comments, and post it on the website,” Quandt says. “If anybody wants to test any hypothesis about how wines are ranked, we are now up to 1,030 wines, and every month we add eight more.” The influence of the software and the group — both dubbed Liquid Assets — grew. They formalized their research as the American Association of Wine Economics, and started to publish a journal. The entire endeavor has turned into a streamlined locomotive of skepticism about the vast, lucrative world of wine tasting and reviews. It’s not a train you want to get in the way of.
Quandt wants to connect preference to the qualities of the wine, to its indisputable chemistry. If you know what’s in a wine, you ought to know whether it’s good or bad — objectively. But you can’t. No one knows how to do it. Nobody can even identify, with certainty, all the ingredients, all the molecules, in a glass of wine (or beer or gin or whatever). Nobody understands, exactly, why booze tastes the way it does, and why people like it. And nobody understands, exactly, how human beings actually taste things.
The difference between Quandt’s group and professional wine reviewers, then, is that the pros attempt to disguise their subjectivity with — well, put it this way: Quandt’s paper on the subject for the Journal of Wine Economics is called “On Wine Bullshit”:
Since there are many wine writers, and there is a substantial overlap in the wines they write about (particularly Bordeaux wines), it is important that there be substantial agreement among them. And secondly, what they write must actually convey information; that is to say, it must be free of bullshit. Regrettably, wine evaluations fail on both counts.
Quandt thinks that pros like Robert Parker — or your friend who always makes a big show over the wine list at a restaurant — are essentially making it all up. Or, like some storefront psychics, possibly they think they know what they’re talking about, when in actuality they’ve merely intuited their way into a con.
In defense of wine reviewers — and reviewers of any booze — it’s really, really hard to talk about how something smells or tastes. Discussions of flavor, that combination of taste and aroma, are subject to the frustrating limitations of analogy. Benzaldehyde tastes like bitter almond and cherry. Cherry and bitter almond taste like themselves. Or benzaldehyde. That’s excruciatingly non-helpful if you’ve never tasted either of those things or, more likely, if my perception of cherries isn’t like yours. Because how could it be? Our noses are different. Our brains are different. Your cherry need not be my cherry.
So how do we talk about booze? How do we connect the subjective perception of what we are drinking with the objective knowledge of what’s in it and how it’s made? Perhaps not surprisingly, booze researchers are getting close to solving this problem. Research into the taste of alcohol promises to explain the taste of . . . well, everything, really. Booze, more than any other foodstuff, connects the quantifiable real world to the messy version of it we all create in our own brains.
Talking about taste and smell is not like talking about, say, colors. For these senses, the hard philosophical and scientific work is trying to find words that we recognize in common and can share with each other, so that we all know what we’re talking about.
The first step is to acknowledge just how limited our senses actually are. A study titled, simply, “The Color of Odors,” will destroy your faith in anybody’s ability to taste anything. Here’s how it worked: three French researchers started with two wines from Bordeaux, a white made with Sémillon and Sauvignon grapes and a red made with Cabernet Sauvignon and Merlot.
The researchers first had a group of subjects taste both the white wine and the red, under white light in clear glasses, and write down all the words they could think of to describe each one. In this test it didn’t matter whether the tasters perceived the same things. Inter-rater reliability wasn’t a factor here — the researchers didn’t care if tasters agreed with each other about the wine color and taste, just that each taster would consistently call one “red” and one “white.”
Then the researchers took an odorless, tasteless extract of the grape-skin pigment anthocyanin and dripped it into the white wine, turning it red. And they called the tasters back for a second go-around, asking them to compare the white wine and the colored wine — the same wine, in other words, with red food coloring. The result was a taste-test catastrophe. Almost to a person, the tasters chose to use the same words for the white wine from the initial tasting on the white wine in the second. And they used the same words for the red wine on the red-colored white wine. They simply could not tell the difference.
Color alone — not aroma, not flavor — told them what to expect, and that’s exactly what they tasted.
But what about experts? At nice restaurants, a professional walks you through a wine list, telling you what each one tastes like and which ones might go best with what dish. Surely these sommeliers must have a more highly developed palate. The four-level training to be a master sommelier, according to the Court of Master Sommeliers, culminates in a grueling, three-part final exam including a twenty-five-minute practical test in which the applicant must identify six wines correctly — grape variety, country, district, and vintage. Thousands of people take and pass the first two levels of the testing every year, but only a couple hundred try the fourth level — the one with the practical test — and only eight or ten pass. Today there are just over 200 of these “master sommeliers” in the world.
Tim Gaiser is one of them. He doesn’t work in restaurants anymore; these days he’s a consultant for wine buyers. I go to see Gaiser at his house in San Francisco’s Sunset district because — I’m going to own up — while I have a pretty good palate for brown liquor, when it comes to wine I’m not much better than “yes, that’s good” and “no, I don’t like that one.” I’ve never had the knack.
We take seats at his dining room table and Gaiser passes across a page of guidelines that the Court gives aspiring master sommeliers to use as a framework for tasting. Gaiser calls wine a “shared hallucination” — not far off from how the science fiction writer William Gibson described cyberspace, actually — but still says people can find commonalities among their perceptions. “We can say Cabernet from the Loire tends to have more red fruit than black. It’s got red flowers, pronounced green herbal stem, green tobacco component, there’s a chalk soil, it’s very acidic, it’s not really tannic, and it tends to be very dry,” Gaiser says. “You know all those things, you’ve been told that, but when you have enough experience with it and you can remember it, that determines expertise.”
Like all sommeliers, Gaiser has a preferred method for tasting a wine, which he’s happy to show me. He heads into his kitchen and returns with a bottle of Cabernet Franc from the Loire. Cab Franc, he says, needs food to really come through in its glory, but Gaiser pours us each a glass, picks his up, and tilts his at a forty-five-degree angle about an inch from his nose. He opens his mouth slightly, and inhales.
“Someone like me, when I put my nose in the glass, it all comes pretty quickly. What I’m doing is generating images.” Gaiser takes a sip, spits, and then a grid opens up in front of his mind’s eye, he says, like a big, light-up schedule hanging from the ceiling of a train station. It appears at the lower part of his sightline, and all those flavors and smells he named come through, with associated images. “It’s a little stinky barnyard,” Gaiser says. “Then there’s, like, violets. Like red flowers or purple flowers. There is a dirty, musty, earthy element as well as an inorganic element, like chalk.”
The same thing happens with the wine’s structural qualities — acidity, tannins, and finish. Gaiser’s scale for acidity is particularly visual. “Mine looks like a slide rule that’s about from here to here,” he says, karate chopping the air in front of him to demark an invisible line about four feet long, “with a red button on it. And I just wait for the button to move and stop.” It’s as if Gaiser has trained himself to be a very specific kind of synesthete, intentionally converting one kind of sensory input into another.
Here is what I think when I take a sip of the Cab Franc: Yeah, this is pretty good, I guess. I thought it was a little sour and thin, but I almost always think that about Loire wines. I like Italian wines from Umbria.
Of course, Gaiser had already broken the key rule of Quandt’s winetasting group at Princeton: He knew what the bottle was before he took a sip. Would he have been able to be as articulate — or to identify it at all — blind? “What would my hit rate be? Probably 70-plus percent,” Gaiser says. “If they are really good examples and they are classic wines.”
That seems like a pretty big if. The wine world is full of strange (and often delightful) labels and combinations. Gaiser admits that those could fool anyone, even a master. For him, the trick is finding ways not to eliminate subjectivity in tasting, but to share that subjectivity. “My strongest belief about wine is that it’s not precise,” Gaiser says. “We do everything we can to give structure to the experience.”
More likely, he’s filtering experience through memory and a trained vocabulary. In 2011 a team of researchers from the University of Padua in Italy and Macquarie University in Australia compared the discriminatory abilities of trained sommeliers to amateur wine drinkers and to sommeliers-in-training. It was a brutal test — the tasters were exposed to fifty aromas, ten of which were common household smells like shoe polish or garlic, and forty were wines. And of the wines, they were trying to specifically identify just ten Italians — five red and five white. The other thirty were there to distract their noses. Oh, and they weren’t allowed to drink the wine — just sniff. Like I said: brutal.
The subjects got tested on their ability to accurately describe specific odorants — that is, use the right adjectives for them, as assessed by judges. And then they were asked to identify the wines. As you might expect, the more training people had, the more descriptors they could call to mind. They had a bigger vocabulary to describe the aromas they smelled. But in terms of blind identification of what the researchers call “wine-relevant odorants,” the sommeliers didn’t do any better than the novices. The pros didn’t have noses more sensitive or well-trained than the amateurs. But also as you might expect, the pros and trainees were better at identifying specific wines, even against the background of the distraction wine samples. The amateurs knew they were drinking wine, but couldn’t tell which one. That actually jibes with Gaiser’s description of his own experience. He and other sommeliers (and presumably the kind of people who can identify other exemplars of other drinks) are matching the new aroma or taste against a stored memory of what that class of drink tastes like. The difference isn’t in an innate ability or skill at constructing and synthesizing “wine bullshit,” in Quandt’s construction, but in experience.
The really interesting thing about the Padua-Macquarie sommelier study, though, is that while the pros and trainees did better than the amateurs, they didn’t do a lot better. The novices got an average of 7.5 out of 10 right, and the pros got 8.6 out of 10. “We tentatively suggest that the verbal skills, which are developed around wine, perhaps lead to a somewhat similar overestimation of confidence in expertise,” the researchers write. They’re hinting that knowing many words to describe wine makes people think they’re better at identifying it than they really are. So go strong to the hoop with that wine list next time you’re at a fancy restaurant, because the odds are the sommelier isn’t a whole lot better at discriminating among wines than you are.
Adapted from “Proof: The Science of Booze,” © 2014 by Adam Rogers. Reproduced by permission of Houghton Mifflin Harcourt. All rights reserved.