Nice article from NYT. In a sense, turning wine analysis into LC/MS based optimization is the obvious thing to do, once wine has already been reduced to a single-axis, 100 point score.


The thing is, as a consumer, I do feel a need for this kind of thing. I bought some wine yesterday, and despite having a little interest, and despite the vastly reduced variety available at my local liquor store, I still felt baffled. I read somewhere that there are estimated to be somewhere between 13,000 and 15,000 brands of wine available on the US market. That's too many. Choice is good, but studies of satisfaction show that people like around 3-5 choices, not 15,000. The rest is just noise. Ditto for microbrews, but at least in beer, people know they're primarily purchasing a funky label. Collect the whole set.