I drank this

From InfoCamp

Built by an Amazon engineer, who's very open to feature requests.

Want it to be easy to import, rather than interrupting the drinking experience by opening a computer.

Using Twitter as an input mechanism (rather than uploading a spreadsheet).

Take a picture of it with your smartphone, then Tweet with the photo and the #idrankthis hashtag (and optional hashtags to show that you liked it or really liked it).

Creates a job on Mechanical Turk, where someone gets paid 3 cents to see the label in the photo and tag it appropriately.

The ultimate goal (the holy grail) is a recommendation engine: no good recommendation engine exists for wine, and wine tends to be something that varies by personal taste.

If you don't rate it at the time, you can go back to the site later on, sign in with OAuth and rate it.

http://twitter.com/jonesabi http://twitter.com/idrankthis

Particularly good for wine or beer because the individual details (the year or the particular winery) makes a big difference to the taste and is really hard to remember after the fact.

Problems with recommendations: too sparse a matrix? Could we use the properties of the wine (the vineyard, the winemaker, the sweetness or dryness) the way that Pandora does?

Recommendations from the audience: social tools label gallery food pairings more ratings experts define wine, then you vote it up or down (pandora) how do the netflix matrix issues show up in wines? could you rate by winery? family trees field blends partner with a winery or a tasting Roshambo Tweetup at wine store Historical overview - data viz, suggestions - tell someone if they've been drinking more chardonnay/zin/ Consider an automated version of the Smuggler's Cove wine club