Issue link: http://read.dmtmag.com/i/764739
9 8 V I N E YA R D & W I N E RY M A N A G E M E N T | J a n - F e b 2 017 w w w. v w m m e d i a . c o m The Delectable app does an amazing job at recognizing wine labels using optical rec- ognition to enter the wine data. It's done mostly by algorithms in a matter of seconds; but when it's stumped (I've tried to do so with low-light photos or Greek wine labels), the deciphering gets farmed out to humans. (Score one for humans!) The community is great, as it's focused on sommeliers to anchor it. Vivino is another app that has similar features and a larger user base. Both apps make suggestions (often too many) and offer purchasing options. Wine Ring is another app that's focused squarely on recommendations. Similar to the previous two, users snap pictures, give the wines ratings and keep a sort of vinous journal. But even as the developers boast about their patented recommen- dation process, the recom- mendations often come up short on quality and volume. In one test, 20 rated wines yielded only one suggestion. The process of discovery, or having algorithms sort through products to offer the best option to a user, is advanced in other parts of the tech world, but it lags in the wine world. Maybe that's OK, since that means recommenda- tions remain a job for humans. But one day, someone will crack this conundrum and unleash the powerful, geo-locating computer known as the smartphone to make practical, tasty recommendations for consumers. Let's just hope they make drone deliver- ies within an hour. Tyler Colman, author of the wine blog Dr. Vino, teaches wine classes at New York Uni- versity and the University of Chicago, and wrote the book "Wine Politics: How Gov- ernments, Environmentalists, Mobsters, and Critics Influence the Wines We Drink." Comments? Please e-mail us at feedback@ vwmmedia.com. END POST TYLER COLMAN wo years ago, Amazon pat- ented "anticipatory" ship- ping. The idea was to predict w h i c h i t e m s c o n s u m e r s might want next and start to ship them in their general direc- tion. Customers then could experience a moment of "Ding dong, here's the item you ordered a few minutes ago." (Hopefully, the plan doesn't now entail a drone hovering with packs of toilet paper, ready to drop on the lawn of whoever orders it.) Netflix also tries to guess what consum- ers might want next. The service has thou- sands of shows and movies in its database, and it wants subscribers to be able to find the most satisfying thing for an evening's viewing. Such was its desire to refine the suggestion process, Netflix even offered a $1 million prize a few years ago for anyone who could come up with a better algorithm than it was using (nobody won the big prize). Google can finish your search for you before you type it; Facebook can filter your timeline to display only updates that are most meaningful to you; Spotify will design a playlist for you; Nest will create a schedule for your thermo- stat; your Tesla will back itself out of the garage and wait for you to head to your next appointment. Machine learning and predic- tive algorithms have never been more pres- ent in our lives. Why then, is it so hard for apps and gad- gets to predict which wine we might like? This is the Holy Grail of wine and tech. Sommeliers and those working in retail have been doing it successfully, face-to-face, for years. A customer says which wine or two he or she has liked recently, and the wine professional generally has no trouble com- ing up with a suggestion from the list or the store shelves. But apps still struggle to do the same. (Opinions expressed in this column do not necessarily reflect those of Vineyard & Winery Management.) Wine Tech's Holy Grail