Google’s announcements of Chromebook and subscription pricing offer a great opportunity to reflect on where we are in the history of computing.  It is all too easy to think of this as something radically new, when it is simply the latest entry in a long historical settling down of the locus of computation.

I started my ongoing love affair with computing at a fascinating time.  In the early 80s businesses were almost entirely dominated by centrally controlled mainframes with dumb terminals while at home I was able to program freely on my Apple II in a type of splendid isolation.  What we have seen in the intervening (gasp) thirty some years is a progressive melding of the two.  We now have lots of local compute power (including increasingly in our pockets) connected to powerful centralized services (not servers) via high speed networks.  This allows us to have the locus of computation sit where it makes the most sense: local for low latency stuff, eg game play, and central for big data stuff, eg filtered feeds.

With the locus of computation problem largely solved, it is fun to think about what is potentially still missing.  I see two areas in particular where we have not yet reached what I would consider some kind of “end state” of computing (with the caveat that “end of history” arguments are always a bit dangerous). 

First, today’s largest centralized services (e.g., Facebook, Twitter, Google, Skype) will eventually be replaced by federated services.   Someone might be quick to point out that Skype is a P2P system, which is true, but this is not an observation about technical architecture (because Google’s search service for instance uses a decentralized system also) but rather about whether there is a single corporate entity that controls a service as compared to a protocol that allows different services to interoperate.

Second, we will see the eventual rise of a new portable and federated identity standard for individuals.  The recent string of password breaches is a sign of an ancient system of identity that has been pushed past the breaking point by the explosion in the number of endusers and services.  I have written a bunch of posts about this problem previously (including a series on the namespace problem for individuals).

Both of these are about shifting the locus of control towards smaller groups and even individuals (and Google’s Chromebook does none of that).  I phrased them both in the affirmative (“will” as opposed to “may”) because I believe that the fundamental thrust of the Internet is to shift power away from institutions.  Neither federation nor identity are likely to require some profound further technical breakthrough (although there might have to be some new crypto stuff involved).  But they do form a kind of mutually recursive loop.  Nobody will want a centralized (as opposed to federated) identity system and federated services require a universal identity system.  I don’t know yet how that mutual dependency will be broken open but I am confident that it will.

Google’s announcements of Chromebook and subscription pricing offer a great opportunity to reflect on where we are in the history of computing.  It is all too easy to think of this as something radically new, when it is simply the latest entry in a long historical settling down of the locus of computation.

I started my ongoing love affair with computing at a fascinating time.  In the early 80s businesses were almost entirely dominated by centrally controlled mainframes with dumb terminals while at home I was able to program freely on my Apple II in a type of splendid isolation.  What we have seen in the intervening (gasp) thirty some years is a progressive melding of the two.  We now have lots of local compute power (including increasingly in our pockets) connected to powerful centralized services (not servers) via high speed networks.  This allows us to have the locus of computation sit where it makes the most sense: local for low latency stuff, eg game play, and central for big data stuff, eg filtered feeds.

With the locus of computation problem largely solved, it is fun to think about what is potentially still missing.  I see two areas in particular where we have not yet reached what I would consider some kind of “end state” of computing (with the caveat that “end of history” arguments are always a bit dangerous). 

First, today’s largest centralized services (e.g., Facebook, Twitter, Google, Skype) will eventually be replaced by federated services.   Someone might be quick to point out that Skype is a P2P system, which is true, but this is not an observation about technical architecture (because Google’s search service for instance uses a decentralized system also) but rather about whether there is a single corporate entity that controls a service as compared to a protocol that allows different services to interoperate.

Second, we will see the eventual rise of a new portable and federated identity standard for individuals.  The recent string of password breaches is a sign of an ancient system of identity that has been pushed past the breaking point by the explosion in the number of endusers and services.  I have written a bunch of posts about this problem previously (including a series on the namespace problem for individuals).

Both of these are about shifting the locus of control towards smaller groups and even individuals (and Google’s Chromebook does none of that).  I phrased them both in the affirmative (“will” as opposed to “may”) because I believe that the fundamental thrust of the Internet is to shift power away from institutions.  Neither federation nor identity are likely to require some profound further technical breakthrough (although there might have to be some new crypto stuff involved).  But they do form a kind of mutually recursive loop.  Nobody will want a centralized (as opposed to federated) identity system and federated services require a universal identity system.  I don’t know yet how that mutual dependency will be broken open but I am confident that it will.