Ed Felton

speaker
18 appearances 1 recordings 1 series first heard Aug 2013 last heard Aug 2013

Ed Felton’s voice in public audio — every appearance, attributed to the second.

Trend

recordings per month · last 12 months
No recordings in the last 12 months.Older appearances are listed below; set an alert to hear about the next one.

Appearances

newest first · ▶ plays the moment
There are different methods.
In all cases, it involves assembling small pieces of data from different sources and putting them together to make a kind of mosaic from which an identity emerges.
It is possible to anonymize or clean data sets so that they are safe to release.
But the difficulty is it can be very hard to tell whether you've done that successfully.
And we're just now starting to see the emergence of a sort of science of anonymization and de-anonymization so that we can know with mathematical confidence that
a particular data set might be safe.
Here's an example of how this potential problem can become reality.
Well, let's tell the story about Netflix.
Netflix is a service that sells people movies, and they want to be able to recommend movies that you're likely to like.
And a few years back, Netflix issued a challenge to the public to try to help them improve their prediction or recommendation methods.
And the way they did this is they took a bunch of data about what their real users did and liked,
And they scrubbed the names and user identifiers out of that information, and then they released it.
Well, some researchers took that data, and what they showed was that it was possible in many cases to identify which users had corresponded to which records by combining the data set with external data sets.
For example, IMDb, which is a site that publishes information about movies, and some users of the IMDb site corresponded to users in the Netflix data set.
That's right.
For the most part, citizens making decisions about their data practices and what to allow are not thinking carefully about what they are able to do, and companies are not very transparent about their practices.
I think we could do better than we are simply by
more transparency and a more honest conversation about what we're giving and what we're getting.
Showing 1–18 of 18