Matt Brubeck ([info]mbrubeck) wrote,
@ 2008-04-17 17:17:00
Previous Entry  Add to memories!  Tell a Friend!  Next Entry
Entry tags:ideas, transparency

The voluntary transparent society
[info]pmb was talking recently about how the taboo on talking about salary hurts workers at the negotiating table, because the employers have more data. That reminded me of an idea I had a while ago…

I think it would be interesting to create a web site where users voluntarily share information that is normally kept private. This could include details of personal finances, medical history, sex/romance, web browsing habits, and more. It would also include basic demographic information (age, sex, location, occupation).

The site would resemble a social network or web of trust. Some or all of a member's data could be restricted in various ways: available only to friends, or friends of friends; only to other people who share the same data; or only as part of aggregate statistics that can't be tied to specific individuals. There would be several difficult problems of anonymity and confidentiality to solve. (For example, see the section here on re-identification of anonymized data.)

The members of the site would then form a sort of transparent society—or a collection of overlapping societies, if information is shared only between friends. In return for sharing their own details, members would have access to new information about their friends and society at large, both individual profiles (for those who so choose) and statistics about groups.

Is this a worthwhile idea? Any other useful feature or applications you can think of?


(Post a new comment)


[info]amoken
2008-04-18 01:09 am UTC (link)
One other application: research. For example, a lot of groups like medical researchers have access to a portion of some people's medical history, probably anonymized if they have any sizable grouping. But they don't always have the full picture. If you did this in such a way that there were lots of structured fields, it would be a cinch to grab all of that stuff and correlate it. With unstructured data, it's still doable, but the more free-form it gets the more reliability you sacrifice.

(Reply to this) (Thread)

[info]amoken
2008-04-18 01:11 am UTC (link)
Oh, and people like me could make interactive visualization applications to let you peruse and discover your own correlations. :D

(Reply to this) (Parent)

[info]triath
2008-04-18 03:01 am UTC (link)
I would be very interested to see this information, but fairly wary of any accuracy since people will selectively self-report the things they want to share (e.g. share if they have a high salary, but not a low one).

Note: pmb's post that you linked to is friends-locked so probably not everyone on your list can view it.

(Reply to this) (Thread)

[info]mbrubeck
2008-04-18 03:36 am UTC (link)
Yes, the voluntary nature means that you couldn't make valid generalizations from the (self-selected) sample to any broader population. It would be more useful for rough, qualitative analysis. For example, you might learn that few of your friends' salaries fall into the range you had thought was "normal."

[I changed the intro to paraphrase Peter's entry instead of linking; thanks.]

(Reply to this) (Parent)

[info]neonelephant
2008-04-18 03:52 am UTC (link)
I know the horse is, like one of those animals in a relativity problem, mostly out of the barn already as I'm trying to close the door here (I think personally I have enough paranoia to give me all of the stress, but not enough to actually secure myself when it comes to these things), but who would operate such a thing and what assurances would users have that a) the owner/operators would not exploit the data, and b) (and, I would hope, more relevantly) any future owners would not exploit the data?

Being able to compartmentalize and selectively share personal information with ease is interesting (and I'm sure the list of applications you mention is nowhere near complete), but it does (or at least it seems to me that the model you present does) require trust in those running the site.

(Reply to this) (Thread)

[info]mbrubeck
2008-04-18 03:38 pm UTC (link)
That's the sort of "difficult problems of anonymity and confidentiality" I'm talking about. :) I think good security engineering can mitigate those concerns, but it might impose other constraints on the design.

In the most extreme case, users could each generate asymmetric key-pairs, and submit data through a trusted (open-source) client that encrypts data on their own computer before transmission, using each of their friends' public keys. Data for aggregated public consumption could be submitted pseudonymously through an onion-routing proxy network like Tor. Small random errors or holes can be introduced in the pseudonymous data; these perturbations make individual profiles harder to re-identify, but will tend to cancel out in aggregate statistics over large datasets.

Then the centralized site needn't be trusted, since it would receive only encrypted or already-anonymized data. On the other hand, this is a hard security model to implement correctly, and it would certainly hurt usability. You might be able to get most of the benefits through clever tricks, maybe involving pseudonyms and hashing (c.f. bloom filters for public sharing of private friend lists).

(Reply to this) (Parent)

[info]coraa
2008-04-18 05:33 am UTC (link)
I think that would be extremely interesting.

It's something I wonder about, especially as, if I remember correctly, the statute of limitations on legally contesting unequal salary/unfair hiring is very short. The chances of actually getting close enough to one's co-workers to find out whether one has been paid less than people of a different sex/race/etc with similar background/skills before the time runs out seems really slender.

(Reply to this)


[info]paperclippy
2008-04-18 01:33 pm UTC (link)
I think it would be really interesting. There are already sites that do some of that -- I can't remember whether it's salary.com or something like that, but there is at least one where you enter in your salary, years of experience, job title, education, and location, and then you can see average salaries of other people, with no identifying information.

For what it's worth, I'm happy to share my salary with anyone who wants to know. I am not a fan of salaries being kept secret. I would also be willing to share a lot of those other details, but either anonymously or only to my friends (for example, I don't think my coworkers need to know how often I have sex with my husband or how much I websurf at work).

(Reply to this)


[info]istgut
2008-04-18 04:42 pm UTC (link)
The problem with using pseudoannonymous information as currency is that pseudoannonymous information is difficult to distinguish from pseudoannonymous disinformation. While I am generally a fan of the wisdom of the crowds thing, I worry about any system in which you have to provide information to get information. How many people fill out questionaires quickly and without a care in order to get to content?

I've long envisioned a post-privacy world, not in fear but rather in anticipation. Yeah, some of what I do is weird. But some of what EVERYONE does is weird. I want us all to stop worrying about it so much.

********** The Following Possibly Not Safe for Lunch *********
A coworker and I happened upon the topic (don't remember how) of wiping habits after using the toilet. Think about it: This is a process that one goes through at least a couple of times a day, but the only instruction or example most people ever recieve is at a very young age from their parents, or perhaps later on from a significant other if they aren't squeemish about it. What is the best way to do it? How effective are you at it compared to other people? Are you getting yourself sufficently clean?
**************************************************************

And don't get me started on how taboos around talking about sex make it less enjoyable and more unsafe for everyone.

(Reply to this) (Thread)

[info]mbrubeck
2008-04-18 04:59 pm UTC (link)
An Advogato-style trust metric might mitigate problems of disinformation. Input to the trust metric could include both the explicit social graph of friend/contact relationships and the implicit graph formed by Gmail-style invitations. (Carefully-metered invitations would also limit users' ability to submit bad information en masse.) You could achieve some level of confidence in your view of the data as long as you trust your friends to some degree, their friends somewhat less, and so on.

(Reply to this) (Parent)(Thread)

[info]istgut
2008-04-18 07:02 pm UTC (link)
Ah, that makes sense. I assume that you'd do the standard thing of targeted advertising to pay for it? Or would you want it to be donation-driven like Wikipedia?

(Reply to this) (Parent)

[info]istgut
2008-04-18 07:05 pm UTC (link)
I was thinking -- you could also put something into the click-through contract when the person provides the information that limits what the information can be used for, even in the event of aquisition... That might help people be more willing to share accurate info.

(Reply to this) (Parent)(Thread)

[info]mbrubeck
2008-04-18 08:42 pm UTC (link)
Ed Felten points out that privacy promises are difficult to rely on:
Even though a company might make a contractual promise to honor some privacy rules, customers won’t have the time or training to verify that the promise is enforceable and free of loopholes. [...] But even if the contract is legally bulletproof, the company might still violate it.
That's one reason I favor technical measures that minimize the opportunity for abuse.

(Reply to this) (Parent)

[info]neonelephant
2008-04-18 07:49 pm UTC (link)
Bruce Schneier wrote something relevant at Wired.com recently, and reposted to his blog here He does have a link to a response from David Brin (since The Transparent Society was cited as a source for the position he discusses); the response is interesting but I don't find it entirely compelling.

(Reply to this) (Parent)

[info]patrissimo
2008-04-18 04:58 pm UTC (link)
One could also imagine a voluntary society (perhaps on a seastead) which had transparency rules.

(Reply to this)


[info]madduckdes
2008-04-18 05:34 pm UTC (link)
I would be interested in participating in something like that.

(Reply to this)

Create an Account
Forgot your login?
Login w/ OpenID
English • Español • Deutsch • Русский…