spf-discuss
[Top] [All Lists]

the problem of regime change

2004-05-31 11:12:13
For a long time, the standard model of government was the
monarchy.  A lucky monarch could hope to reign for his or
her natural life, and during their reign their country could
expect a reasonably consistent society.  When power passed
peacefully from one monarch to another, social turbulence
was kept to a minimum.  When leadership transitions were
eventful, society tended to be disrupted.

The American system of government institutionalizes change:
presidents turn over every four or eight years.  While that
builds a certain minimum level of disruption into society,
at least that disruption can be planned for and foreseen.
Compared to a monarchy, where the order of things could go
on for fifty to a hundred years and then be massively
disrupted, scheduled transitions of power are more like a
controlled burn than a wildfire.

Perhaps it's necessary, in this period of great change on
the Internet, to adopt a model of regular controlled burns,
to avoid the damage of an out-of-control wildfire.

SMTP has remained essentially unchanged since 1985.  Because
there were no small burns, spam grew and grew in the
underbrush.  SPF, which essentially updates the SMTP model,
is like first controlled burn in a long time.

If, in the future, we wish to minimize disruptions, perhaps
we should make provision for controlled burns; we should set
maximum terms on standards.

I'm fairly sure that one day we will find that SPFv1 has run
out of steam.  When that day comes, progressives will want
everybody to do another round of upgrades.  But
conservatives will be happy with the status quo.
Progressives will find that they will have to fight two
battles: one to overcome inertia and get people to accept
the need for change at all, and one to actually agree on and
implement the change.

If we want to make life easier for the progressives in the
future, we can remove the need for the first battle, by
preparing everyone now for change in the future.

For example, we could put an expiration date into SPFv1.
Because the expiration date would be known ahead of time,
organizations could budget the necessary resources to
participate in the development of SPFv2, and an adoption
schedule could be worked out well in advance.

Of course, that expiration date could pass uneventfully;
people might find themselves quite happy with the way things
were, and conclude that SPFv2 should just be a copy of SPFv1.

If people aren't happy with SPFv1, at least everybody's
prepared for the necessary disruption that v2 will bring,
and we save ourselves the trouble of convincing the
conservatives.

Putting an expiration date into an RFC is a pretty radical
move, though.  I don't know if it's been done before.

But legislators do it all the time: they write laws that
expire, and are either renewed at the time of expiration,
allowed to lapse, or are updated with new language.

Why shouldn't RFCs, the laws of the Internet, do the same?

What do people think?


<Prev in Thread] Current Thread [Next in Thread>