Log in

No account? Create an account
Liveblogging the DIMACS privacy workshop - Arvind Narayanan's journal [entries|archive|friends|userinfo]

Liveblogging the DIMACS privacy workshop [Feb. 6th, 2008|11:47 am]
Arvind Narayanan
[Tags|, , ]

I know I said I won't be blogging for a while, but this workshop has been a lot more interesting than I thought, so I'm taking some time off to write this.

This is a very interesting crowd of people: cryptographers, database people, statisticians, and one or two people from law, advertising, the census bureau and so on. The person who's speaking now uses words like "construct" and "deconstruct" and "glocalization" and "dialogic perspective" :-) A very diverse group of people who have a shared stake/interest in privacy. It's the crowd I feel most comfortable hanging out with.

The last time I was at such a workshop was in May 2004 in Italy. Much has changed for the better this time around. First, the people from different backgrounds are actually having a productive dialog. In the last workshop, the cryptographers would say their thing, and the statisticians would say their thing, but neither group made sense to the other because our backgrounds, our modes of thinking and our way of modeling the same problem were all very different.

Second, differential privacy has been a huge success. The basic theory has been very well worked out now in several papers and it's clear that it's the right definition for a wide variety of privacy preserving data mining problems. As someone said, it has given rise to a whole cottage industry of extending the definition and applying it to various problems.

Third, I think people generally understand now that most people don't give two hoots about privacy. Someone (I'm not mentioning names to avoid pissing off anyone) asked for a show of hands from everyone who had sent or received an encrypted email in the past week. No hands went up.

It's clear that very few people will take the slightest extra step to get privacy; the only way to have it is to have it by default. Also, there is an increasing (but not enough!) realization of the fact that a lot of the things that young people do is their choice and their prerogative, even if scares privacy advocates shitless. If kids don't want privacy, we shouldn't shove it down their throats.

On the other hand, people do a lot of stupid things that gets them into trouble, such as identity theft; we should of course work on preventing that sort of stuff even if people are too lazy or don't care about privacy. Bottom line: it's not necessarily vital to protect privacy unless it leads to actual or potential damage down the line, as opposed to general unease about other people learning things about you.

On a personal level, this was a very different experience for me than other conferences or workshops I've been to: it was strange to find that many or perhaps most people here know me or have heard of my work. I'm meeting many friends after a long time.

I've really enjoyed my chats with people from other disciplines, especially those work influences public policy. Two years from now, I may or may not be doing research. if I'm not, I still want to be able to make something of my research career, since I'm sort of emotionally attached to my research (I think you need to be to get through a Ph.D.)

Everyone goes to all the talks at this conference (and in fact, at most conferences). I just don't get this. There is no way everyone can be interested in or understand all the talks. Ideally, there should be about half the people on average going to any talk and the other half hanging out outside discussing each other's work or just networking. On a related note, there seriously needs to be a speakerratings.com.

Finally, Rutgers is a very nice place to visit. The campus I'm put up in has a rustic air and a scenic beauty to it even though it's surrounded by industrial wasteland. But it's a terrible place to go to school. Freshmen and sophomores are apparently not allowed to bring their cars even though it's impossible to get anywhere without one. It's really horrible. It's not that easy to get to NYC or anywhere fun either, not to mention the round trip by train costs $40.

Edit: ok, I was wrong about the $40. Sorry about that. NJ still sucks :)

[User Picture]From: medryn
2008-02-06 05:28 pm (UTC)
Hmm, I was basing the $22 each way on a document provided by the workshop, which is turns out is wrong. According to njtransit.com, it's $10.50 each way to NY Penn Station. Somewhat more reasonable.
(Reply) (Thread)
[User Picture]From: arvindn
2008-02-06 05:32 pm (UTC)
the fact remains that nyc is hard to get to :)

i was talking to some undergrads and they don't seem to go there all that often, if at all.
(Reply) (Parent) (Thread)
[User Picture]From: rfc9000
2008-02-06 11:04 pm (UTC)
I wanted to attend some of the talks (esp the tutorials on the first day), but sadly I was so damn busy with work that day :(

It's not that easy to get to NYC or anywhere fun either, not to mention the round trip by train costs $40.
Nope, its super easy to go to NYC, take a train from Rutgers college ave campus, and it takes you right to Times Sq. and round trip costs 16$.
And this week is free NJ Transit week, so if you show your student id, it costs 0$!
(Reply) (Thread)
[User Picture]From: paper_crystals
2008-02-08 06:29 pm (UTC)
I realized that you are in my relationships filter and not my sex filter. Do you want in the sex filter? Or out of the relationships filter? Or in both? Or in neither?
(Reply) (Thread)
[User Picture]From: arvindn
2008-02-09 02:48 pm (UTC)
i guess i answered that one :)
(Reply) (Parent) (Thread)