[cap-talk] Capability accounting

[Norman Hardy]

On Jun 25, 2006, at 5:15 PM, Sandro Magi wrote:

> Norman Hardy wrote:
>> I agree with all your points. Agents for some situations are easy
>> however.
>> I can instruct an agent in my browser to pay up to 1 cent for any
>> link that I explicitly click on but 10 cents for any NY Times page I
>> select.
>> Other demands by web servers will require my explicit attention.
>
> I think the problem, is that metered resources like water and
> electricity have:
>
> 1. sufficiently course-grained controls, power switches and  
> faucets, so
> the user can mentally track/estimate how much these resources were
> actually used (should he so choose).
>
> 2. sufficiently stable, or at least slowly-varying, prices (so there's
> no shocking "surprise" after the fact).

If prices had actually risen during the recent California power
crises it would have been less painful than the blackouts.
Since the PUC has a reaction time expressed in years instead of minutes
we had blackouts, which is like the price going to infinity.
I don't want to start a political discussion here but this bears  
directly
on the nature of meters that we chose over a more common ideas.
Keykos CPU meters were designed to support CPU futures
(otherwise known as reservations) which is a hedging
strategy that users can use to protect against price spikes.
Engines <http://www.cs.indiana.edu/~dyb/pubs/engines.pdf> have
been implemented in the language Scheme to allocate time
but I think you can't build reservations using engines.

Many institutions use CPU metering as a way to allocate the cost
of the machine that they bought.
Others use meters to control an internal market to insure that
the machine is allocated to its most valuable use and any time.
The later plan leads to fluctuation prices and futures ameliorate
the pain.

........