[OpenID] A lesson from history for Facebook

Johannes Ernst jernst+openid.net at netmesh.us
Thu Apr 16 23:10:00 UTC 2009

On Apr 16, 2009, at 11:38, Peter Williams wrote:
> ... a lot of marketing about open systems, but not a lot of actual  
> __multi-vendor__ delivery

This is an excellent observation. But I think the exploding  
interconnectivity that we are all seeing, all over the technology  
universe, may cause a generational change here where the new patterns  
are different from the old ones that were observed so many times. This  
particularly applies to OpenID.

Here's my reasoning:

If all there is is a single mainframe with custom software and no  
connectivity (1950's), the subject of open system does not even exist  
because it would serve no useful purpose.

If you want to connect a few workstations or pieces of software, and  
that's all you do, vendor-proprietary interconnect systems win because  
they are cheaper to build and maintain and have less interoperability  
bugs. Your customers won't like that, because they don't like the lock- 
in, so they complain and you market open systems, but don't deliver or  
don't deliver well. That's the situation that we are all familiar with.

But if you are a relying party on the web, and potentially >>1000's of  
IdPs, and vice versa, the only thing that works is open protocols.  
Your choice choice is between not interconnecting, or interconnecting  
via open protocols. Anything else is simply not affordable because of  
the NxM problem. (Note this argument applies from RSS, in a way the  
granddaddy of open interconnect protocols on the web, to all members  
of the "open stack", whichever way it will turn out to be put together  

This is new. Prior to this generation of the web, we simply did not  
have any place where >>1000's of systems built and maintained  
independently by 1000's of actors needed to interconnect, and needed  
to interconnect so cheaply that they better work out of the box  
(because tech support is unaffordable)

The only counter-argument that I know is that if the #1 IdP had a  
gigantic market share (think Windows market share) in all major  
markets and was not likely to lose it over the next N years, then a  
proprietary approach may win. Some people think Facebook may have or  
gain that -- I beg to differ. And given that Facebook is now an active  
contributor to the OpenID movement, they might agree with me on that ;-)

So I think the signs for actual multi-vendor delivery, as Peter put  
it, in my view are far better at this point in time than they ever  
were in the IT universe. And historical parallels are instructive, but  
not decisive.

Now disagree! ;-)



Johannes Ernst
NetMesh Inc.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: lid.gif
Type: image/gif
Size: 977 bytes
Desc: not available
URL: <http://lists.openid.net/pipermail/openid-general/attachments/20090416/eeab4e81/attachment-0004.gif>
-------------- next part --------------
-------------- next part --------------
A non-text attachment was scrubbed...
Name: openid.gif
Type: image/gif
Size: 903 bytes
Desc: not available
URL: <http://lists.openid.net/pipermail/openid-general/attachments/20090416/eeab4e81/attachment-0005.gif>
-------------- next part --------------

More information about the general mailing list