Ali Bahreman wrote:
What both PEM and PGP users need is a good user interface. If e-mail
is the application of choice to add security to, then integrate PEM
or PGP to all mailers, but until you do that, I don't think you will
get a serious sized user base....
I think this is an important point, the main reason why PEM
(and also PGP) are not so widely used. I hope that examples
bellow will not turn out as over-simplified ones.
Case #1:
Suppose we have a GM car and a Toyota car. Would you as a
user mind if Toyota uses one (redundant) chip more for its
air-bag system than GM? As long as both provide
functionality, required by state and international
regulations, you won' t care. It is Toyota' s problem if
their air-bag system costs them more... Similarly, when you
buy a ticket from air-company, you don' t care if the
aircraft you' ll fly is a Boeing or AirBus. All that you
care of is your seat, friendly on-board service, etc., which
all represent your "user agent" in this case.
So I think that a real problem of PEM is a lack of an
appropriate user agent and other supporting tools that would
basically do the following (let's assume use of X.500,
although it might be any other data-base system):
* With a click on button a certificate of another party
would be retrieved.
* With a click on another button this certificate would
be checked (could be CA-Browser :-)).
* With a click on a third button, a signature could be
produced and a reply sent out.
In this case, the user will not care if there is a very
complex X.500 system and a complex CA-structure in the
background. As long as all these systems work in line with
certain standards and state regulations (which can be
checked, if required), everything is OK.
Case #2:
Let' s move now from users' point of view to the public-key
infrastructure implementation. Regarding technical aspects,
it is almost the same whether public-key infrastructure is
hierarchical or "flat". But if you provide a CA service and
your service vitally depends on another party that has
nothing to do with certification matters, it is pretty tough
job (this is the case in PEM CA-model used within X.500
because of mixing certification hierarchy with naming
hierarchy). Next, having an opportunity to be IPRA and thus
having a monopoly is quite interesting if you are the lucky
one who operates it and makes money with this job. But it is
not very pleasant fact for the rest 99.999...% of Internet
community.
Conclusion.
It is not a technical problem that PEM (and also PGP) are
not widely used (read as "used for commercial purposes").
Both problems have common root and this is *the nature of
human reasoning*. As I am a technician, I would say that the
reasons for the problems may fall in the area of psychology,
management of human resources, etc.
Solution:
I would suggest to:
* relax current PEM CA-hierarchy model to the minimum;
* define additional supporting systems to support users
when using secure services so they can just push buttons on
their user-agents.
Again and again I am thinking of a genius idea of DNS, where
name space is completely separated from IP numbers, and both
are separated from physical address space. This all remains
hidden to users and this is what we should achieve with
public-key infrastructure.
I wish you a successful work and sorry if I have taken too
much of your time.
Best regards,
Denis
************************************************************* ************
* Denis Trcek O O * * *
* "Jozef Stefan" Institute O O * * *
* Jamova 39, 61 111 Ljubljana, SLOVENIA O O * * *
* e-mail: denis(_dot_)trcek(_at_)e5(_dot_)ijs(_dot_)si
denis(_dot_)trcek(_at_)ijs(_dot_)si O * * *
* Tel.:+386 61 1259 199, Fax:+386 61 1261 029, +386 61 273 677 * * *
******************************************************************* ******