1. Apps that do need to know me as a human, e.g. my bank. I am
associated with a real customer account, so I can't be anonymous to
them.
2. Sites where I buy stuff. It would be hard for them to ship
something to me if they couldn't associate my personal info (name,
CCard, address) with my order. That means they need to store my data.
3. Apps that on which I want to build a social network, ala LinkedIn,
Facebook, delicious, etc. My username/data identifies who I am so my
friends can find me. LinkedIn wouldn't be too useful if everyone had
profiles like user123, user423, user994.
4. Apps that provide community driven ratings. These apps are moving
to be less anonymous, like AMZN now can associate the user's profile
to comments. You might think this sucks but in the world of rampant
comment SPAM I think it helps.
5. Sites where financial transactions between users are taking place,
like eBay. I would feel more comfortable knowing something about the
user on the other end of the transaction.
It seems that Clipperz is not suitable for sites that are engaged in
social computing or any kind of finance. For the rest, like news sites
that want a login, you can already just provide a stupid username for
yourself "bozotheclown123". So net effect - in the places where you
want it to work it can't, and for where it will work there is already
an easy solution.
There seems to be a lot of people interested in this solution, so
clearly I am missing something.
I think that your main concerns are about the "zero-knowledge web
architecture", more than the Clipperz application itself.
For some of the examples you have listed, our proposed architecture is
probably not very suitable.
But the problem we are raising, is that everybody is taking the
current architecture (the server knows everything about everybody) as
a matter of fact, and not a design decision you can choose to subvert.
Many times, the choice to keep all the data in their plain format on
the server is just the easiest way to go, not necessarily the "best"
(whatever this could mean in any given situation).
A very simple example we have been reasoning about lately, is the
online poker applications; the current implementations keep the full
state of any macht on the server, and this as already caused a few
frauds.
We are trying to point out that it would be possible to implement an
online poker application that could keep things going as usual,
without any privileged insight on what is going on. It would still be
able to know who won and who lost, but not much more.
Again, consider "zero-knowledge web architecture" as a new conceptual
framework to think about how problems can be solved online available
to designer; it may not be always the best fit, but it is definitely
something that could be quite helpful in many occasions.
Hope this answer some of your concerns.
Best regards,
Giulio Cesare