Which fish dish? TopDish, Spork have advice
Beyond restaurant reviews, new services tell you what to order at a restaurant, or where to find just what you want to eat right now.
SAN FRANCISCO--With the restaurant rating and recommendation business being pretty well locked up (by Yelp, OpenTable, Foursquare, etc.), the new game in town is apparently recommendations on individual dishes. Got a hankering for tom kha gai soup? You can check out Spork (live) or TopDish (invite-only beta) to find the best restaurant nearby that serves that particular dish; both companies are presenting in the low-rent "launch pad" sideshow of the Launch conference here.
These two services collect user reviews--ratings and pictures--of dishes to help you make the life-critical decision of where to find the best of whatever you're looking for, or if you're sitting at a restaurant, which dish to order. Both sites let you profile your tastes to help decide for you what you're more likely to like.
Spork is a bit more social at the moment. It connects to your Facebook network to prioritize food ratings from your friends. An upcoming feature will let you gift a dish to a friend via a PayPal credit for the cost of the dish. A future network update may work the credit through restaurants directly.
Co-founder Dan Cheung told me he's also considering adding a "reverse Groupon" feature to the service: If enough users like a restaurant's dish, Spork may ask the restaurant to create a coupon for it, to stimulate demand just a little bit more.
TopDish is a bit newer, still in closed beta. Its recommendations are network-wide, for the time being, and the mobile app isn't out yet. The model is largely the same as Spork's, but co-founder Salil Pandit told me his service's secret sauce will be communication with restaurants: If you run an eatery, you'll be able to see how all your individual dishes rate. This will be a free service for a while, although the value to a restaurant could obviously be quite high. "We just want to help start a conversation," Pandit told me.
The increasing granularity of data in new Web services is an important trend to watch. Highly-specific recommendation databases don't work unless there's enough volume of users and data feeding into them. Without that, you get a lot of empty records and unsatisfied users. But with everyone getting with the program of recommending things to friends, checking in, and Tweeting or Facebooking their every move, it's not surprising that companies like these (and some others, launching tomorrow at this conference) are tying to make sense of these little tidbits of opinion.