People Like You
August 23, 2007
I was reading one of my regular blogs - fred wilson’s a_vc - and he touched on a subject I’ve been pondering for over a year (and which serves as a driving philosophy in zync). The idea is that with the successes (in terms of traffic) of community review sites like yelp, trip advisor, and citysearch, we now find ourselves facing an interesting paradox - there are too many reviews out there! This has a two-fold effect - first, we don’t know which reviews to trust and which are garbage; and second, the actual numerical ratings trend towards a mean, becoming less meaningful as the number of reviewers grows.
An example: the 99 Restaurant in Boston gets an 8.7 on CitySearch. Not a bad rating. Yet, most of the people I know don’t go there because the food quality and ambiance aren’t…how do I say this…good. So while the 99 may be satisfactory as a local watering hole, for some fried food, and to watch the Sox, it would be poor recommendation for people like me.
These days, I find myself disregarding the ratings and focusing on each reviewer’s comments, trying to assess how similar that person is to me and how much I can trust what they are saying. Ideally, as Fred points out, we’d only see reviews by like-minded people that we can trust! A friend from MIT, Anil Dharni, discusses this issue nicely in his blog post: ”What’s Broken With Yelp“.
I run into examples of this all the time, but another that sticks with me is when I was researching hotels in Bali for my honeymoon on trip advisor. I was looking for something upscale, romantic, quaint and off the beaten path, and not gaudy, too large, or “over the top”. I found one that by most accounts sounded amazing. But one reviewer said “Yes, this is a gorgeous hotel on which billions were spent by the prince of Bali. However, we found it to be ‘trying too hard’, ostentatious, and the food was lousy!…”. The review went into great and articulate detail in justifying these claims, and I found myself saying “this person is like me - the level of detail, values, what’s important - I really should trust this person.” …but I didn’t - they were having a deal yada yada and we stayed there. It *was* gorgeous, but completely off the charts on the gaudy/ostentatious and ‘not subtle luxury’ scale (it was the heart shaped bath tub and 10 dining areas that all served the same fare that put us over the top).
The point I’m making is that there’s a real need to personalize the way information is presented on the web, especially local information. We’ve seen a movement towards “personalization” with netflix and amazon, and I believe this same approach can be applied in other markets, such as travel and “local”, to make it easier for people to find the right information. For a crisp look at how recommendation and personalization systems work, read this post from our friend Joshua Porter’s blog on “Which movie to watch“.
Briefly, that’s what zync is about. We’re making “local” personal - enabling faster and easier dicovery of relevant new stuff to do with your free time. Zync applies personalization algorithms to find People Like You™ and pushes activity suggestions to you proactively. I’ll stop there - you can check it out for yourself at zync.
-brad
September 12, 2007 at 6:34 pm
Congratulations on the launch! The site looks awesome. We’re moving back to SF next month, so you gotta start one up there. Best of Luck, Matt