Re: Reboot: Strategic goals for GNOME

[Richard Stallman <rms gnu org>]

I wrote:

    > Let's not be in a rush to invite users to use servers -- even our own
    > -- instead of their own computers.  That is the wrong direction to go.

I chose those words carefully.  They do not say we should eliminate
all servers; I don't think that.  For some purposes, servers are the
only solution.  For some purposes, there is nothing wrong with using a
server.  The most obvious case where servers are no problem is for
publishing things (such as blogs, photos, and essays), but there are
many others.

By contrast, the clearest problem case is where you use a server
instead of a program running on your own computer.  In other words,
when the job you are doing is a job that can be done on your own
computer, your dependence on someone else's server to do that job is
evitable.  That case is Software as a Service (SaaS).

Another problem case is sharing data with specific people privately.
The obvious solution is for these people to connect to a single common
server.  That is not SaaS, since the job is one you can't do within
your own computer, but it is still a risk.  And the server is not the
only solution.  It could be done peer-to-peer.

Exchanging the data through email is one peer-to-peer solution.  That
uses servers -- plural -- but it does not require all the users to use
one common server.  Some knowledgeable users do run their own mail
servers, and this method permits them to use those.  Other users don't
run their own mail servers (I for one don't know how), but they can
use mail servers that they have some reason to trust.

Also, this method allows the data to be encrypted except within the
individual users' machines, so that none of the servers through which
the email passes has access to the real data.

There are other ways of implementing data sharing through peer-to-peer
methods.  "Use a common server used by thousands of people" should be
the fall-back choice, not the first choice.