"yelp" entries

Podcast: ratings, rankings, and the advantage of being born lucky

A conversation with Sean Taylor, Hilary Mason, and John Myles White about how ratings affect our thinking

Outcomes following random exogenous upvotes and downvotes on message board posts. Image via Sean Taylor.

Researchers randomly upvoted some posts and downvoted others on a popular message board. The upvoted posts became substantially more popular over the long run. Image via Sean Taylor.

Is popularity just a matter of simple luck–of some early advantage compounded by human preference for things that are already popular? A paper published today in Science offers some insight into the way that popularity emerges in online ratings. Lev Muchnik, Sinan Aral, and Sean Taylor were able to set up a randomized experiment on a popular Reddit-like message board in which they gave some posts a one-point upvote on publication and others a one-point downvote. Posts that were “born lucky” ended up with 25% higher scores on average than those without modification.

In our latest podcast, Renee DiResta and I are joined by Sean Taylor, Hilary Mason and John Myles White to talk about Sean’s findings and about ratings, rankings and reviews in general. Bits and pieces that come up in the podcast:


Subscribe to the O’Reilly Radar Podcast through iTunesSoundCloud, or directly through our podcast’s RSS feed.

Yelp partners with NYC and SF on restaurant inspection data

A joint effort by New York City, San Francisco, and Yelp brings government health data into Yelp reviews.

One of the key notions in my “Government as a Platform” advocacy has been that there are other ways to partner with the private sector besides hiring contractors and buying technology. One of the best of these is to provide data that can be used by the private sector to build or enrich their own citizen-facing services. Yes, the government runs a weather website but it’s more important that data from government weather satellites shows up on the Weather Channel, your local TV and radio stations, Google and Bing weather feeds, and so on. They already have more eyeballs and ears combined than the government could or should possibly acquire for its own website.

That’s why I’m so excited to see a joint effort by New York City, San Francisco, and Yelp to incorporate government health inspection data into Yelp reviews. I was involved in some early discussions and made some introductions, and have been delighted to see the project take shape.

My biggest contribution was to point to GTFS as a model. Bibiana McHugh at the city of Portland’s TriMet transit agency reached out to Google, Bing, and others with the question: “If we came up with a standard format for transit schedules, could you use it?” Google Transit was the result — a service that has spread to many other U.S. cities. When you rejoice in the convenience of getting transit timetables on your phone, remember to thank Portland officials as well as Google. Read more…

Four short links: 29 August 2012

Four short links: 29 August 2012

NeoVictorian Computing, Participatory Budgeting, Micro Thrusters, and Geopositioning Accuracy

  1. NeoVictorian Computing (Mark Bernstein) — read this! I think we all woke up one day to find ourselves living in the software factory. The floor is hard, from time to time it gets very cold at night, and they say the factory is going to close and move somewhere else. […] The Arts & Crafts movement failed in consumer goods, but it could succeed in software. (via James Governor)
  2. Participatory Budgetingresearch shows participation is more effective than penalties in taxation compliance. Participation is more effective than penalties in almost everything.
  3. MIT-Developed Microthrustersa flat, compact square — much like a computer chip — covered with 500 microscopic tips that, when stimulated with voltage, emit tiny beams of ions. Together, the array of spiky tips creates a small puff of charged particles that can help propel a shoebox-sized satellite forward. You say satellite, but it’s only a matter of time until this powers a DIY RC rocket with a camera payload. (via Hacker News)
  4. Yelp Checkins to Measure Geopositioning Accuracy Across PhonesBy analyzing millions of data points, we can easily see how, on average, different platforms perform. iPhones consistently have the most accurate positioning, with a fairly small accuracy radius. Android phones are often inaccurate, but reliably reported that inaccuracy. And finally, iPods using Wi-Fi positioning proved the least accurate and usually reported incorrect accuracy radii.
Four short links: 21 March 2012

Four short links: 21 March 2012

Data Journalism, Fast Web Servers, Android App Inventor, and Daily Deal Dirt

  1. S0rce — gorgeous infographics. They purport to let you Think for Yourself which is bald-faced bullshit: the choice of which data to present, and the invisible collection and curation practices behind the data, is the choice of what story to tell and what it will say. That said, it’s wonderful to see the numbers (and they are attributed) behind the Republican Primary and Copyright and Piracy Legislation.
  2. Modern HTTP Servers are Fast — I remember when the best web engineering in the world would still fall over if a box got more than 10 hits/second. Yes, yes, I’m writing this on my grandpa box. Check out the hardware specs of the box these numbers are from.
  3. MIT App Inventor — web-based app designer. Does not appear to be open source. There is no long-term sustainability for this kind of development environment: when MIT decide “nah screw it, not going to run this any more” or “hmm, maybe we’ll charge for it”, you’re boned–you can download the “source” to your app in a zip file but AppInventor is the only dev environment which can consume it. I hope it’ll become the awesome and easy dev environment that Android needs, but I hope they prevent it from being a dead end.
  4. Daily Deals: Prediction, Social Diffusion, and Reputational Ramificationswe consider the effects of daily deals on the longer-term reputation of merchants, based on their Yelp reviews before and after they run a daily deal. Our analysis shows that while the number of reviews increases significantly due to daily deals, average rating scores from reviewers who mention daily deals are 10% lower than scores of their peers on average. (via Greg Linden)