David Derbyshire in The Guardian:
[…D]rawing on his background in statistics, [Robert] Hodgson approached the organisers of the California State Fair wine competition, the oldest contest of its kind in North America, and proposed an experiment for their annual June tasting sessions.
Each panel of four judges would be presented with their usual “flight” of samples to sniff, sip and slurp. But some wines would be presented to the panel three times, poured from the same bottle each time. The results would be compiled and analysed to see whether wine testing really is scientific.
The first experiment took place in 2005. The last was in Sacramento earlier this month. Hodgson's findings have stunned the wine industry. Over the years he has shown again and again that even trained, professional palates are terrible at judging wine.
“The results are disturbing,” says Hodgson from the Fieldbrook Wineryin Humboldt County, described by its owner as a rural paradise. “Only about 10% of judges are consistent and those judges who were consistent one year were ordinary the next year.
“Chance has a great deal to do with the awards that wines win.”
These judges are not amateurs either. They read like a who's who of the American wine industry from winemakers, sommeliers, critics and buyers to wine consultants and academics. In Hodgson's tests, judges rated wines on a scale running from 50 to 100. In practice, most wines scored in the 70s, 80s and low 90s.
Results from the first four years of the experiment, published in theJournal of Wine Economics, showed a typical judge's scores varied by plus or minus four points over the three blind tastings. A wine deemed to be a good 90 would be rated as an acceptable 86 by the same judge minutes later and then an excellent 94.
More here.