First your math is right for 1-bottle of wine tasted by one critic. You say above that the critics are only reliable within ±4-points with the Kosta Brown. That’s about right I suspect, we should ask Princeton’s Orly Ashenfelter.

Second your logic is wrong for 100-bottles, with the same ±4-points. The bivariate regression confidence limits show critics are right within 1-point with 70% confidence if you give them a set of 100-bottles. Critics will get the Kosta-Brown within 1-point most of the time.

The Point—The glass is half-empty for you because you have one-bottle to sell, whereas it is half-full for the retailers who have hundreds of bottles to buy.

As I prepare to bike home I realize this is a cool blog, that a wine blog is considering the math of flavor is wonderful.

Again though, you’re assuming that the error that McCloskey estimates (30%) is accurate and that Quality itself can be quantified.

You assume that by averaging scores we can get closer to the “truth”. Or am I mis-reading you?

McCloskey seems to be claiming that he already knows the “truth” and can somehow compute the standard error of the critics’ scores based on their deviation from it.

I however believe that the “truth” (at least in terms of wine quality) is subjective and different for different groups of people. I don’t buy the argument that intensity=quality, and neither do a whole slew of wine drinkers. Subjective quality is based on individual physiological differences in people’s palates, their preferences and other completely irrational factors.

Totally agree that consumers want better information about wine though. They are uncertain in their palates and feel like they aren’t qualified to judge quality for themselves. They want to be confident they are buying good wine, even if it’s just someone else’s personal preferences.

I wrote about the problem for winemakers here: http://www.pinotblogger.com/2007/06/15/bashing-the-100-point-scale/

Even made a cool flow chart. This blog (and the spread in influence of other wine blogs) as well as the rest of our marketing is my attempt to hedge against ever receiving a bad score.

I’m confident, but only time will tell if we’re successful.

And thanks again for the conversation!

Nice to have a banana slug as a reader! Thanks for the comment.

I guess my response would be two-fold.

Markets are efficient in that they incorporate information quickly, and this info is reflected in prices quickly. Totally agree. But what if the information is wrong? Garbage in, garbage out. The entire market mis-priced risk on mortgage bonds very badly for the past few years, for instance.

Additionally, if prices and scores always move together, why does DRC command such high prices when it receives pedestrian scores of 90 points and lower? And if prices and scores aren’t linked, then what does this say about either’s correlation with wine quality? Clearly there’s something else going on, and it probably isn’t rational.

Now if I were to open a shop like Zacchy’s, none of this would interest me particularly. All I need to know is what will sell. And what sells is what Parker and Laube say is good. But this has nothing at all to do with (capital Q) Quality, which, I contend, is unquantifiable much less definable.

Thanks again for contributing Mathman!

And Morgan – excellent comments and observations. I’m a little more accepting of the quant movement in general, but I share an equally deep skepticism for people who think that we live in a gaussian world where we can accurately estimate everything with formulas. Still, quants can be very useful. Just need to know when and where to apply their skills.

Markets are efficient, using information to make money. Scores attract my attention because the markets use them. Mathematics of scores is real. Before we can talk about why scores are real, lets consider why we sense scores are not accurate. Writers more famous that we are equally confused, See http://www.jancisrobinson.com/articles/jr429

Josh gives the example of Kosta Brown, measured 2-times. Laube and Parker have an Error of 8-points. So, the single wine has an error of ±4-points.

So why do markets use scores? Someone is using scores to make money, but they can only do it if they have some reliability.

Retailers like Zacchy’s are buying and selling 100 wines, not a single Kosta Brown. An accuracy of 3-4 points is better than guessing says our retailer. When you buy 100 wines there is a 70% chance that the score is accurate to within 1 or 2 points, pretty good for business purposes

i love Leo for exactly that. In the same way that I love Clark Smith. I want to sucker punch them both in the stomach on a personal level. I feel that they endeavor to take every romantic aspect of vineyard and place, and of the ineluctible greatness of wine from a certain piece of soil, that they can. they have the will to actually think they can quantify human emotion.

As Socrates said though. Your strongest friend and enemy is the gadfly. It is that thing that throws your most fervent beliefs back in your face and makes you reconsider. that is what leo and clark are. there is no doubt that their egos would grant them much more.

that said, i personally believe that their necessity to quantify wine in matrix form comes from a combination of freudian impulse and an inability to smell….

I find that to be the most disturbing observation of the entire article. The fact that everyone spends their time chasing the palates of a few overly influential critics is leading toward a homogenization of wine, the ultimate dumbing down.

Basically, he’s suggesting we do what Bordeaux did to their region, which (IMHO) is the most confusing and elitist appellation and grading system known to mankind. I just can’t see it happening (or at least, I’m hoping I don’t have nightmares about it happening).

Of course, with AA reporting that the TTB is suspending new AVA designations indefinitely due to the Calistoga issue, who knows what might be coming. Gives me chills.

OK then, so glad I came by to read this before going to bed. HA!

–R