Short answer: the question is what the 92 on the shelf talker tells you about the bottle under it, and the answer is that one person, on one afternoon, at a table of eighty wines, was impressed by it, and that this now describes most of the wines in the store. It says nothing about the price, which was the only question you were asking. A score is a compliment. A worth is a number with the price inside it. Use the score to rule out the bottle that has none, use the critic’s name to know whose mouth it was, and then do the arithmetic the score was built to make you forget.
Where the number came from
Robert Parker started scoring wine out of 100 in the late 1970s because the 20-point scales of the day had gone soft, and he wanted a number with room in it. He always said it was “really a 50-to-100 point scale”: every wine starts at 50 for showing up, and a 90 meant something rare. He and the friend he built it with picked it, in his words, because it seemed “much less likely to result in the inflated wine ratings they saw all around them.”
It worked for about twenty years. Then the stores noticed that a card reading 90 under a bottle sold the bottle, every other critic adopted the scale to get on the card, and the number stopped being a measurement and became a sales tool with a measurement’s face.
What happened to it
In 2013 Wine Spectator’s James Laube wrote that his editors had reviewed 17,000 wines the year before and given none of them 100, and then wrote the sentence that mattered: “Critics are giving more 100-point scores than ever before.” He was already late. A year earlier, one critic’s first pass at Napa had 1,061 new wines, 123 of them at 95 or above, nearly one in eight, with the midpoint of the whole set at 92. In 2015 Tim Atkin tasted 389 young Bordeaux and averaged 91.1 points across them, and wrote, “The easiest thing in the world is to give everything 93 points.” By 2024 Jamie Goode was reporting that “95 is the new 90” and that a 90 “now feels a bit like a fail” to the winemaker who gets it.
Why it happened is not a mystery. Critics compete for wineries’ samples and readers’ attention, a high score gets quoted and a low one gets ignored, and, as one of them put it, these critics like being liked. The store finishes the job. When five critics have scored a bottle, the card under it shows the highest. On wine.com this week there is a $14.97 Pinot Noir wearing a 94, and nothing about that is false. It is just not information.
How much a palate agrees with itself
Here is the part the card never prints. A statistician named Robert Hodgson spent years slipping the same wine, from the same bottle, three times past the judges at the California State Fair without telling them. Only one judge in ten gave it the same medal each time; another one in ten swung from gold to nothing; the rest wandered in between. These were trained judges in a quiet room. The critic who scored your bottle was tasting eighty that afternoon.
The critics do not agree with one another much better. A 2024 study of red Bordeaux found that professional critics’ scores correlate with each other about 63 percent, and that the crowd on Vivino correlates with the critics about 40 percent, and with one critic, Decanter’s panel, about 16, which is to say barely at all. Vivino was already a hundred million ratings from thirty million people by 2018, nearly all of them three or four stars, which tells you the crowd likes wine and not much else. Neither the one palate nor the thirty million is measuring the thing you think they are.
What the number is worth in dollars
Somebody checked. A study of 266,301 bottles scored by Wine Spectator and sold in American stores found that the score moves the price only above 90. Below that line, a 79 and an 89 sell for about the same money. The market does not price the wine; it prices the threshold, because the threshold is what fits on the card. Which means the one place the score is doing real work is the one place it is least reliable: the two points on either side of 90, where a judge agrees with himself one time in ten.
What to do with the score
Treat 90 as “nothing wrong with it.” That is what it has come to mean, and it is useful. A wine over $20 with no score at all from anyone is the one to wonder about.
Read the name before the number. A 92 is not a fact; it is a person. If you know the person likes big, sweet, oaky wines, and you do not, his 95 is a warning. The card hides the name in small type for a reason.
Ignore the difference between 91 and 93. It is inside the noise of one afternoon. The difference between 85 and 95 is real. The difference between 91 and 93 is lunch.
Then ask the question the card will not. What is this bottle worth, against what the shops charge for it and against what the same money buys one shelf over? That is a number with a price in it, and no critic prints one, because a critic who said “93 points, and $20 too expensive” would not get his samples next year.
I am one palate too. The difference is that I say it in a sentence, with the price in it, to you and nobody else, and no winery sends me a case.
Straight answers
What does a 90-point wine mean?
That one critic, tasting many wines in a session, found nothing wrong with it. Scores have risen so far that 90 is now a floor, not a distinction; critics themselves say “95 is the new 90.”
Are Vivino ratings reliable?
Somewhat. A 2024 study found Vivino’s crowd scores correlate with professional critics about 40 percent, while critics correlate with each other about 63 percent. Nearly all Vivino ratings cluster at three or four stars.
Is a 95-point wine worth more than a 90?
In price, yes: a study of 266,301 bottles found scores raise prices only above 90. In the glass, the difference between adjacent scores is within one taster’s own day-to-day variation.
How consistent are wine judges?
In Robert Hodgson’s California State Fair experiments, about one judge in ten rated the same wine the same way when served it repeatedly; one in ten swung from gold medal to no medal.
Why do critics use a 100-point scale?
Robert Parker adopted it in the late 1970s as effectively a 50-to-100 scale, intending it to resist inflation. Retailers’ shelf cards made 90 the selling threshold, and scores have drifted up since.