Wine Judges Scored the Same Wine 91, 87 and 95 — From the Same Bottle, in the Same Flight.

ISSUE 051 · MYTHS & MISINFORMATION · OCTOBER 2026

The story has been told so many times it has worn a groove. A French researcher, a room of wine experts, two glasses. One red, one white. The experts describe the red in the usual vocabulary — jammy, crushed red fruit, the whole performance — and then the punchline lands: both glasses held the same white wine. One had been dyed with tasteless food colouring. Not one of the fifty-four noticed.

It is a perfect story. It confirms what almost everyone already suspects about wine, it humiliates exactly the right people, and it takes fifteen seconds to tell.

It is also, in most of the versions you have read, inaccurate.

The researcher was Frédéric Brochet, then a doctoral candidate at the University of Bordeaux. In 2001 he ran a set of experiments on wine perception as part of a larger dissertation, much of which was concerned with the language tasters use rather than with catching anyone out.

In the experiment everyone remembers, he gave participants a white wine, then a week later gave them the same white wine again — this time dyed red with an odourless, tasteless colouring. The tasting notes changed accordingly. The white wine was described in white-wine terms. The dyed wine was described in red-wine terms.

Brochet’s own conclusion was about perception, not fraud. Colour, he wrote, supplies significant sensory information that misleads judgement of flavour, and what happens is a genuine perceptual illusion. That is a claim about how vision hijacks smell. It is not a claim that wine expertise is a con.

Start with the participants. They were not fifty-four wine experts. They were oenology students — undergraduates studying winemaking at Bordeaux. Somewhere in the retelling, students became experts, and experts became sommeliers.

Second, the claim that nobody noticed. Brochet has said that around two to three per cent of people do detect the white wine flavour, and the published data show some participants giving white-wine descriptors to the dyed glass. Not all fifty-four were fooled. The absolute version of the claim — every single one, without exception — was added by people repeating it.

Third, and least known: the tasting portion of the experiment does not appear to have been published in a peer-reviewed journal. The paper usually cited, in Brain and Language, is the lexical analysis. The most-quoted study in the entire wine-scepticism canon is, in its most-quoted part, not a peer-reviewed result.

And the conclusion drawn from it does not follow either. The experiment never tested whether people can tell red wine from white. It tested whether a strong visual cue can reshape the words they reach for while describing a smell. Those are different questions, and only one of them was asked.

Brochet ran another test that gets less attention and deserves more.

He took one mid-range Bordeaux and served it from two bottles: one labelled as a grand cru, the other as ordinary table wine. Same wine. The tasters produced nearly opposite descriptions — the grand cru glass drawing praise, the table wine glass drawing complaints.

This is a cleaner result, because the manipulation is closer to real life. Nobody dyes wine before a dinner party. But everybody sees the label.

If you want a serious case that professional wine judging is unreliable, skip the food colouring entirely. The damaging work was done by a statistician and winery owner named Robert Hodgson, and it is unglamorous in a way that makes it far harder to dismiss.

Hodgson had noticed his own wines winning gold at one competition and placing nowhere at another. So he arranged, with the cooperation of the California State Fair wine competition — the oldest in North America — to test the judges themselves.

The method was simple. Judges were poured the same wine three times from the same bottle within a single flight, without being told. Their scores moved by roughly four points on the 80–100 scale. A wine scored 91 at one pour could come back 87 at the next and 95 at the third.

Across the years he ran it, only about one judge in ten was consistent. Worse, consistency did not persist: the judges who scored reliably one year were unremarkable the next, which suggests the good years were luck rather than skill.

In a related analysis of 65 tasting panels, Hodgson found that the wine itself was the only significant factor in the judges’ quality ratings in just 30 of them. In the remaining 35, judge bias was doing real work.

This is the finding that matters, and note what it is: not that judges cannot taste, but that they cannot reproduce their own verdict from one glass to the next. A medal is a snapshot of one person’s afternoon.

So wine tasting is nonsense and the experts are frauds. That is the conclusion the internet reached, and it is where this issue parts company with it.

The research on expertise is reasonably consistent, and it says something more specific. Wine experts do not have better noses. Studies comparing experts and novices have repeatedly found no difference in raw olfactory sensitivity, and similar sensitivity to wine components like tannin and alcohol.

What experts have is memory and structure. The same work that found no sensory advantage found that experts are substantially better at explicitly identifying and recognising wine-relevant odours. They know what the smell is called, and they have met it before.

Brain imaging tells the same story from another angle. Sommeliers tasting wine activate regions associated with working memory and deliberate strategy. Novices activate the primary taste cortex and areas handling emotion. The novice is having an experience. The expert is running a procedure.

This shows up in trainable ways too. Follow students through a wine course and their descriptions get longer, more specific and more consistent with each other — trained tasters wrote notes around a quarter longer than novices in one study, using more pertinent terms and agreeing more with one another about which terms applied.

So expertise is real. It is simply cognitive rather than sensory. Experts are not tasting things you cannot taste. They are recognising, naming and filing things you also taste but cannot place.

There is one more finding, and it is the one with consequences for anyone standing in a supermarket aisle.

Researchers looking at wine consumers have documented a Dunning-Kruger pattern: the least competent tasters tend to overestimate their own ability the most. The people most confident about what they are detecting are, on average, the ones detecting least.

Combine that with the price effect — the same wine praised from a grand cru bottle and criticised from a table wine bottle — and the mechanism behind the entire premium wine market comes into view. Expectation does a great deal of the tasting. The label is an ingredient.

None of which means expensive wine is a swindle. Some of it is genuinely, measurably better made. It means the confidence attached to the judgement is running well ahead of the evidence supporting it — which is the recurring subject of this newsletter, in a glass.

The dyed-wine experiment: Frédéric Brochet, doctoral research at the University of Bordeaux II (2001), and G. Morrot, F. Brochet and D. Dubourdieu, “The Color of Odors,” Brain and Language 79 (2001); RealClearScience, “The Legendary Study That Embarrassed Wine Experts Across the Globe” (2014), for Brochet’s own framing of the result as a perceptual illusion.

Corrections to the popular version: “Urban Legends of Science: About That Wine Experiment,” sciencesnopes (2013), documenting the student–expert conflation, Brochet’s own 2–3% detection estimate, the presence of white-wine descriptors in the data, and the absence of peer-reviewed publication for the tasting portion.

The bottle-label experiment: Brochet’s Bordeaux supérieur served a week apart from grand cru and vin de table bottles to 57 participants; reported in The New Yorker and summarised in Gizmodo, “Wine Tasting Is Bullshit. Here’s Why” (2013).

Judge reliability: Robert T. Hodgson, “An Examination of Judge Reliability at a Major U.S. Wine Competition,” Journal of Wine Economics 3:2 (2008), and “An Analysis of the Concordance Among 13 U.S. Wine Competitions” (2009); David Derbyshire, The Observer (2013), for Hodgson’s account of the California State Fair trials, the ±4-point swing and the 10% consistency figure.

Expertise, sensory versus cognitive: R. H. Ashton, “Dimensions of Expertise in Wine Evaluation,” Journal of Wine Economics 12:1 (2017); Parr et al. on equivalent sensitivity but superior identification and recognition memory; Castriota-Scanderbeg and the fMRI literature on sommelier brain activation, summarised in Frontiers in Behavioral Neuroscience; Wang and Prešern, “Does Blind Tasting Work?” Journal of Wine Economics 13:4 (2018), on training effects and tasting-note length.

Overconfidence: C. Aqueveque, “Ignorant Experts and Erudite Novices: Exploring the Dunning-Kruger Effect in Wine Consumers,” Food Quality and Preference 65 (2018); Masset and Raub, “The Impact of Wine Tasters’ Expectations on Wine Quality Ratings and Willingness to Pay,” Journal of Wine Economics 18:2 (2023).

ONE MORE THING

You can read our MANY newsletters for free at our website:

Read our archive for free at OneWildWorld.com →