some arguments for privacy

A couple of weeks ago I asked for some arguments in favor of privacy.  I
pointed out that one person's increased privacy exerts negative
externalities on others by reducing their available information.  I wanted
to know why more privacy might be beneficial to society despite this
consideration.  I didn't really get the answers I wanted, probably because
I wasn't clear about the kind of arguments I had in mind.  Anyway, here I
give some arguments of my own, which I hope offer new perspectives on this
issue. 

Privacy as Insurance

Suppose you are looking for a job.  It seems reasonable to argue that if
you are a better than average worker, you would be able to get a better
offer if you had less privacy because the potential employers would be
better able to distinguish your abilities from your past history.  On the
other hand less privacy would hurt you if you are a worse than average
worker.  If you don't yet know your own abilities, you would prefer more
privacy as an insurance against your own potential deficiencies.  If that
doesn't seem realistic, consider how the argument might apply to your
children.  This line of reasoning also explains why people are troubled by
genetic screening.  Thus privacy might increase social welfare by
providing a sort of social insurance.

Privacy as Restriction on Signaling

"Signaling" is a term used by game theorists to describe the use of
publicly observable actions to provide information to others about one's
private attributes.  The best example comes from biology, where male
peacocks grow extravagant tails to signal their genetic fitness to
females.  Clearly signals must be costly, otherwise they wouldn't be
convincing. They are often also wasteful, as in the peacock example.  (As
a side note, the deposit solution to the junk mail problem I talked about
some days ago is an example of non-wasteful signaling.)  Privacy reduces
the range of actions one can use as signals.  This would increase social
welfare if the wastefulness of the signals exceed the benefit they provide
in the form of useful information.  Consider a possible future where every
room in every house is wired with a camera that continously broadcasts to
the Internet.  Life would certainly be very uncomfortable in this future,
as every trivial action must be carefully considered in order to preserve
one's reputation.

Possible Benefit of Non-Privacy Limited

This is more of an argument for privacy technology, rather than privacy
per se.  Suppose that privacy-invading technology becomes much cheaper
than privacy-enhancing technology.  Given the arguments above it seems
inevitible that governments will pass laws to restrict the distribution of
certain kinds of information about individuals.  But of course this will
not keep the information out of the hands of those governments themselves
and other resourceful organizations.  Thus any possible benefit of
decreased privacy in the form of market efficiency would be severely
limited since only a few market players would have improved information.
This benefit would be easily outweighted by the harm in the form of
governments' increased coercive power.

One of the most compelling consequences of strong privacy is
pseudonymity, i.e. the ability to conduct business and personal
affairs without allowing others to learn your physical identity or
location.  This is a very strong protection against extortion, 
kidnapping and all other kinds of physical violence.


(Its lesser cousin, anonymity, also protects again physical
violence, but it has limited usefulness in business and/or personal
affairs.)


(For example, if a business magnate wishes to manage a vast financial
empire while still allowing her children to grow up in a safe and
open environment, there is probably no better solution than that she
manage her vast financial empire pseudonymously, so that would-be
thugs will not know _who_ to kidnap in order to extort ransom.)


Perhaps Wei Dai should apply his sharp mind to analyzing the net
effect on society of _that_ phenomenon.


The benefit for the individual thus protected is clear, but, as the
Unmentionable Topic shows, it is arguable that the net effect on 
society (as it were) will be less than optimal.


Regards,

Zooko Journeyman

Disclaimers follow:  I am not a cypherpunk.  NOT speaking for 
DigiCash or any other person or organization.  No PGP sig follows.

-----BEGIN PGP SIGNED MESSAGE-----

At 02:15 AM 4/8/97 -0700, Wei Dai wrote:

>Suppose you are looking for a job.  It seems reasonable to argue that if
>you are a better than average worker, you would be able to get a better
>offer if you had less privacy because the potential employers would be
>better able to distinguish your abilities from your past history.  On the
>other hand less privacy would hurt you if you are a worse than average
>worker.

One of the problems with talking about privacy is that people use the word to
mean many different things. Apropos to the current discussion, I see two
different uses co-existing in a non-useful way:

1.	Privacy as a lack of information
2.	Privacy as a right to control the flow of information

where the second usage refers to the ability to decide who ends up with the
first usage. I'm assuming that lack of information can be characterized as
risk, and that risk can be characterized as negative value. The Coase Theorem
would seem to suggest that the second usage is (may be) superfluous, because
the [lack of] information/risk will be allocated to the party which values it
most. However, the right discussed in (2) is still meaningful, because it is
(and will) be used as a bargaining chip to gain or lose other things of
value.

Also, it's useful to think about whether or not there are (or should be) some
things we are unwilling to do in the name of efficiency. For example, we
might consider it "efficient" if people were allowed to sell their children;
but most societies have chosen to consider markets for children unacceptable
on moral grounds, apart from concerns about the efficient distribution of
children. (I've heard that (Richard?) Posner, an outspoken federal appellate
judge, wrote a paper discussing markets for children as a thought exercise;
and that the political fallout from that paper (some 10-15 years later) means
that he's got no real chance of a Supreme Court seat, even if he's otherwise
qualified. Perhaps this message means I'll never get there, either. :(

A society might choose to place privacy on the list of "things for which
there may not be a market"; but adherents to the strong version of
cryptoanarchy will likely predict that a market will exist, anyway. (I
believe there's currently a small global market for children; but they're
still more difficult to buy than, say, cigarettes. And slightly less
difficult to buy than, say, a gun in Berkeley.) 

But (for me) the real reason that privacy is good (even if it's
"inefficient") is that privacy is inherently related to control, and to
freedom. It's very difficult to influence or alter the behavior of a person
unless you can monitor their behavior. If we make it difficult (or
impossible) to learn some things about people, we effectively create zones of
autonomy related to those zones of privacy.

I believe that this is the somewhat elusive link between the various
constitutional rulings on "privacy" grounds, discussing abortion and
contraception and childraising and interracial marriage and private
possession of pornography and gay rights (although the last has failed, so
far.) Those decisions paint an invisible line between "private" and "public"
spheres, whereby the state cannot or should not interfere with personal
choices; and I believe that the close correspondence between those lines and
traditional notions of "intimacy" is not coincidental. 

I believe that a hypothetical privacy-free state would be "efficient" in the
same way that a hypothetical centrally-planned economy is "efficient" - e.g.,
on paper, it sounds good. All of the right things will happen at just the
right times, risks (or economic goods) will be allocated to those who deserve
them, etc. In practice, centrally planned economies have proven to suffer
from human flaws - greed, lust for power, shortsightedness, poor judgement,
and so forth. The initial allocation of all (or substantially all) economic
resources to one economic actor (the state) prevents the operation of the
Coase Theorem and its redistribution of rights/values, because the other
actors (individuals) have nothing to offer in return; and because it is not
possible for the other actors to legitimately retain those rights or value. A
society which initially distributed information rights to the state - or
which refused to recognize them as rights - would, I believe, create a
similar lack of freedom and autonomy. 

It's difficult to discuss privacy in economic terms because we are accustomed
to valuing one side of the choice - risk, or lack of information - in
quantitative terms, but we are not accustomed to valuing the other side -
freedom, or lack of opportunity to exercise control - in quantitative terms.
(I don't think that the other party's valuation of risk is a good measure for
the value of the freedom. But this is turning into a pretty long message
which is somewhat more theoretical than may be of interest to many list
members. So I'll go to bed and see if people feel like continuing in this
vein.) 


-----BEGIN PGP SIGNATURE-----
Version: 4.5

iQEVAgUBM0omo/37pMWUJFlhAQHABQf/QQW4B+Wwq6gGLV/YxvAawYOiGp8VBTqm
EUSyTjw+Wai+vv/RovdgvKtU3JWxFn91tU28kkTDqjQ2DqkapWK8zi42XIIite3/
HXQ3t18BGhq1v9IRvOlOOIKXmeEG+frf59CTczzgwzivDSfmchT8PSOvHJo9+KKp
vzHbwaWGS1pCUeQZNYl5GQ1667jb0oy+wKssiX9Cy4HuQ2z/ZnFYKjQZsGfEJalK
z9yoTyGMCB6KUFuEL+AdDe6HdikSqQgBgJBOCXi/B8OTkKPgfOnsdiueiVq6GIgC
fVEywZo5nCtWDfVW1/QE39O7jBQvLG2pdTE5aaYQzLR2xIN3vAAAow==
=exeH
-----END PGP SIGNATURE-----

--
Greg Broiles                | US crypto export control policy in a nutshell:
gbroiles@netbox.com         | 
http://www.io.com/~gbroiles | Export jobs, not crypto.
                            |

On Tue, 8 Apr 1997, Greg Broiles wrote:

> One of the problems with talking about privacy is that people use the word to
> mean many different things. Apropos to the current discussion, I see two
> different uses co-existing in a non-useful way:
> 
> 1.	Privacy as a lack of information
> 2.	Privacy as a right to control the flow of information
> 
> where the second usage refers to the ability to decide who ends up with the
> first usage. I'm assuming that lack of information can be characterized as
> risk, and that risk can be characterized as negative value. The Coase Theorem
> would seem to suggest that the second usage is (may be) superfluous, because
> the [lack of] information/risk will be allocated to the party which values it
> most. However, the right discussed in (2) is still meaningful, because it is
> (and will) be used as a bargaining chip to gain or lose other things of
> value.

What I meant by privacy is somewhat different from your definitions: "the
ability of an individual to control the distribution of information about
himself".  Notice that I said "ability" instead of "right".

Part of what makes privacy so interesting is that the Coase Theorem
doesn't apply.  If you look at the Coase Theorem carefully, it presuposes
the lack of transaction costs, which in turn means that all relevant
information is distributed symmetrically among interested parties.
(Otherwise the cost of inducing some individuals to disclose private
information to others becomes part of the transaction cost.)
Therefore even to invoke the Coase Theorem implies that either there is no
privacy, or for some reason no one chooses to exercise his privacy.

In fact, this is exactly what I meant by the inefficiency of privacy:
not only does the Coase Theorem not apply to privacy itself, it
undermines the Coase Theorem everywhere else by increasing transaction
costs.

The rest of your article is very interesting, so please keep going.  I'll
just take this opportunity to point out that the most useful measure of
efficiency is a relative one.  Even the centrally planned economy is
not totally inefficient.  Apparently some CPEs seem quite able to keep its
citizens from starvation, even for years at a time.  The CPE is only
inefficient compared to the market economy.  If for some reason human
beings are inherently unable to organize markets (perhaps because of some
deep-seated cultural bias against putting prices on essentials like food
and children), then the CPE might very well be the best alternative.

At 1:15 AM -0800 4/8/97, Wei Dai gave some arguments pro and con privacy:
>Privacy as Restriction on Signaling
>
>....  Consider a possible future where every
>room in every house is wired with a camera that continously broadcasts to
>the Internet.  Life would certainly be very uncomfortable in this future,
>as every trivial action must be carefully considered in order to preserve
>one's reputation.

Or peoples expectation about the range of private behavior will change to
be more in line with the reality.  Those whose behavior is 5 standard
deviations away from the norm will be screwed.


>Possible Benefit of Non-Privacy Limited
>
>This is more of an argument for privacy technology, rather than privacy
>per se.  Suppose that privacy-invading technology becomes much cheaper
>than privacy-enhancing technology.  Given the arguments above it seems
>inevitible that governments will pass laws to restrict the distribution of
>certain kinds of information about individuals.  But of course this will
>not keep the information out of the hands of those governments themselves
>and other resourceful organizations.

Such laws will support monopoly use of personal information, and as such be
very unlikely to contribute to the good of society.

At CFP '97, David Brin gave a lunch talk where he argued against privacy
and in favor of accountability.  The example he used was the cop and the
driver, both with the secure recording TV cameras on their shoulders.  Of
course the crypto-anarchist view is you can use any information you can
get, and I can use strong privacy technologies to keep you from getting it.


-------------------------------------------------------------------------
Bill Frantz       | God could make the world   | Periwinkle -- Consulting
(408)356-8506     | in six days because he did | 16345 Englewood Ave.
frantz@netcom.com | not have an installed base.| Los Gatos, CA 95032, USA