A shelf talker says 94 points. The bottle next to it says 89. The 94 costs twice as much, and you assume the difference is five points of quality. It is not. Those two numbers were often given by different people, in different rooms, for wines that were never compared to each other, under conditions that had a large effect on both. A score is a real piece of information, but it is a much narrower one than the way it gets used, and knowing its shape is what stops it misleading you.
What a score actually records
It records one taster's opinion of one bottle on one afternoon, tasted alongside twenty or fifty others of broadly the same type, usually in a short window, usually without food, and frequently when the wine is far younger than anyone would ever drink it.
Every one of those conditions matters:
- Lineups reward volume. When a taster works through forty young reds, the wines that stand out are the concentrated, oaky, high-impact ones. Subtle wines do not shout, and a wine that would be lovely with dinner over two hours can seem thin at position twenty-eight. This is not a failure of the taster. It is what tasting in bulk does to anyone.
- Youth distorts. Many scores, particularly for Bordeaux and Burgundy, come from barrel samples months before bottling. The wine is unfinished. The taster is projecting forward, honestly, and projections are sometimes wrong.
- Palates differ systematically. Critics have preferences — for ripeness or for restraint, for new oak or against it — and those preferences are stable and knowable. A reviewer who consistently rewards power will consistently underrate wines built for the table.
The scale is far smaller than it looks
The hundred-point scale is not a hundred points. In practice almost everything published sits between about 85 and 98. Below that, wines mostly do not get reviewed at all. So the working range is a dozen or so points, which means the distance between 88 and 92 is much larger in market terms than the numbers suggest — and much smaller in the glass than the price gap suggests.
Scores also inflate over time and vary between publications. A 92 from one source and a 92 from another are not the same claim. Comparing scores across publications is close to meaningless.
Where scores break down worst: across styles
This is the failure that costs people the most money.
A German Riesling Kabinett at 8% alcohol, a Beaujolais cru, a Napa Cabernet and a vintage Champagne might all be scored 93. That number is supposed to mean "excellent example of its type", and within a type it broadly works. Across types it collapses, because the four wines are not competing for the same job and the scale quietly favours concentration, weight and oak. Light, high-acid, low-alcohol wines are structurally disadvantaged by a system that rewards impact.
The practical result is that the styles most useful at a Hong Kong dinner table — lighter reds that survive a warm room, high-acid whites that cope with soy, salt and chilli — score lower on average than heavy reds that are hard to drink with most of what we actually eat. Buying by score alone will steer you steadily towards big wines and away from wines that would work better with your food.
What scores are genuinely good for
Three things, and they are worth having.
- Screening out faults and incompetence. A wine consistently in the low 80s from multiple sources is probably not very well made. That is useful.
- Comparing one producer across vintages, by the same critic. Holding the palate constant makes the numbers informative. If a grower's wine runs 90, 91, 89, 94, the 94 is a real signal about that year.
- Finding a reviewer whose palate matches yours. Buy three or four wines a particular critic rated highly. If you liked them, that critic is now a useful instrument for you personally. If you found them heavy and overworked, you have learned something more valuable than any single score.
The advice to ignore, in both directions
"Never buy below 90." This is the most expensive rule in wine. It removes from consideration almost every honestly made, fairly priced bottle in the HK$200 to HK$400 range, because those wines are rarely submitted for review at all. Absence of a score is not a low score. Most of the world's wine is never reviewed by anybody.
"Scores are meaningless, trust your own palate." Equally lazy, usually said by people who have already spent years building the experience that makes their palate trustworthy. If you are still working out what you like, a consistent critic is a reasonable starting point. The goal is to outgrow it, not to pretend you began there.
Who should ignore this article
If you buy wine partly as an asset, or you buy names that need to be recognised across a banquet table, scores are not a flawed guide to quality for you — they are the actual mechanism that sets resale value and status. A 100-point vintage trades higher because it is 100 points. That is circular, and it is also true, and arguing with it would cost you money.
For everyone else, the useful posture is this. Treat a score as one person's shorthand, not a measurement. Use it to avoid the bad and to compare like with like. Then, when you are choosing between a 91 you can afford and a 95 you cannot, remember that you are looking at four points of one afternoon's opinion, and that the room, the glass, the food and the company will move your enjoyment further than that gap ever will.
Where to start
The styles that score lowest are the ones that work best at a Hong Kong table — light, high-acid, low-alcohol wines that a lineup of forty young reds is structurally built to overlook. Buying against the score here is not contrarianism; it is buying for the food you actually eat.
- Beaujolais — disadvantaged by a scale that rewards impact, and far better than its numbers alongside Cantonese food in a warm room
- Germany — Riesling Kabinett at 8 to 9% is the clearest case of a wine the hundred-point scale cannot see properly
- Laurence et Remi Dufaitre Brouilly AOC 2021 — the sort of bottle rarely submitted for review at all, which is not the same thing as scoring badly