Re: poly: Why so much anonymous virtual reality?

David Friedman writes:
>1. It isn't clear that you need lots of personae. If all you want is
>privacy, freedom from regulation by government, etc., you only need
>two--one online and one in realspace. And if you are a good person, there
>is much to be said for linking your online personae, on the theory that the
>good reputation of each will help all the others. That is one of the
>reasons my web page has all four of my worlds on it, instead of my setting
>up separate pages not linked to each other.

When you link all your online persona into one, you are most at risk to
being discovered. Someone trying to uncover who you "really" are has the
most data to work with, and the largest payoff from finding out. They
could be state police or private extortionists. And the longer you want
these persona to last, the worst it gets. You'd better be damn paranoid
sure of your security, and of near-zero observable correlation between
your online actions and your realspace actions.

>2. This also answers Robin's point about people being suspicious of
>anonymous dealings. I'm not anonymous online--I'm just not linked (in the
>hypothetical future) to a realspace body. The decision to keep that link
>cut only clearly signals dishonesty if one believes that government law
>enforcement is the only good way of enforcing contracts and that there are
>no strong, common reasons to want government unable to observe your
>behavior.

I'd say it depends on the fraction of the population that makes substantial
use of such anonymous persona. Sure, if most people use them a lot, then
use doesn't signal much. But if only a small fraction do, there's likely to
be a substantial selection effect of who does so use, and so signaling.
If only a small fraction of ordinary people switch to more use of anon persona
upon acquiring some feature they'd rather hide, this can swamp the otherwise
motivated users. This is one of the reasons I keep focusing on what fraction
of the population would find the benefits of anon virtual persona outweigh
their costs.

>3. The more of your life occurs in cyberspace, the harder it is for the
>sort of time matching that Robin describes to work. After all, some of your
>time in cyberspace involves no persona at all, since you are just reading
>information, not sending it. And you can set up some unattended sessions
>when you aren't in the room--as long as no serious interaction is involved.

I'm not sure this is true, but even if true is a relatively small effect.

>A further point is that the snoops don't get to observe all, or nearly all,
>of your interactions. Most of them, after all, are encrypted conversations
>with particular people who don't work for the bad guys. The only
>interactions they can use to figure out who you are are the ones they are
>party to--entrapment, "public lectures," and the like.

I find it likely that most people will come to record most of their interactions,
at the very least to be able to rebut accusations by others. Given that these
records are sitting around, and that people regularly get offers to sell copies
of their records about other people to "snoops", the question is how often such
sales would take place. I find it likely that when four or more people who aren't
really close get together, records of the event would be sold, if the value of
the info were worth much more to snoops than the cost to move and process the bits.

I'll try to post soon a simple way for such snoops to cheaply buy such records.

>4. Robin's point about personal style is an interesting one. It sounds as
>though we have an arms race. On the one side are the ID programs, trying to
>pick one person out of a few billion on the basis of voice rhythms, walking
>patterns, etc. On the othe side are the cloaking programs--deliberately
>modifying those things, presumably using observations of other people's
>voice rhythms etc., so as to keep content but randomly vary form. The limit
>seems to be pure personality, whatever that means--the things that make me
>me. My suspicion, perhaps mistaken, is that IDing by that is a hard enough
>problem so that by the time it shows up it is going to be a very different
>world--complete with uploads and intelligent AI's.

My intuition is the opposite. I can think of lots of relatively easy ways
to go looking for good personality signatures, given I have the data to sift
through and the cycles to do it. The task of erasing such signatures seems
much harder to me, and also seems to require expensive conscious attention
to have a chance of working; you have to learn to *act* well. And think of
the money extortionists might have to gain by finding you out.

Robin Hanson
hanson@econ.berkeley.edu http://hanson.berkeley.edu/
RWJF Health Policy Scholar, Sch. of Public Health 510-643-1884
140 Warren Hall, UC Berkeley, CA 94720-7360 FAX: 510-643-8614
Received on Fri Feb 27 18:11:28 1998