From: jim bell <jimbell@pacifier.com>
Date: Fri, 5 Jul 1996 11:47:16 +0800
To: "Mark M." <markm@voicenet.com>
Subject: Re: ecash thoughts
Message-ID: <199607050111.SAA17969@mail.pacifier.com>
MIME-Version: 1.0
Content-Type: text/plain


At 08:26 PM 7/4/96 -0400, Mark M. wrote:
>-----BEGIN PGP SIGNED MESSAGE-----
>
>On Wed, 3 Jul 1996, jim bell wrote:
>
>> At 05:53 PM 7/3/96 -0400, Simon Spero wrote:

>> Low, maybe a tenth of an American cent.  But probabilistic payment should be 
>> used to allow the minimum average payment to go way below this, perhaps to 
>> an unlimited extent.  The reason is simple:  The cost of providing net 
>> transactions, and electronic transactions in general, can be expected to 
>> drop exponentially, just like the cost of telecommunications and CPU power 
>> do.  Any arbitrary limit to how low they can go will act somewhat akin to 
>> the minimum wage:  It will deter development of any product or service whose 
>> perceived value is less than this arbitrary minimum.
>
>If the value of a Turing is one tenth of an American cent, then it would
>actually just be a pseudocurrency backed by U.S. dollars. 

I should have said, "about a tenth of a cent."  I didn't mean to imply a linkage.

>The inflation of
>ecash would be the same as the inflation of U.S. money.  However, I do agree
>that the value of one unit should be low.
>
>You use Moore's Law to state that the cost of electronic transactions drops
>exponentially.  However, this is only true if the electronic transactions
>use the same amount of bandwidth.  As chip processing speed and transmission
>bandwidth double, the cost of building the equipment also doubles.

At any given time, that's true, but over time the cost of that processing 
(per unit transaction) will drop, probably in some exponential fashion.  

For an optical fiber transmission system, the cost of the fiber does NOT go 
up with the speed, since it's nowhere near its limiting capacity.  
End-termination systems will be more expensive, but I suspect that's a 
relatively small fraction of the overall cost.  

CPU cost will be significant, but then again the Moore's law trend will 
predominate.  CPU's probably have 1000 times the power, per unit cost, than 
they did in 1980 or so.  It would probably be over-optimistic to think that 
they'll drop the same ratio over the next 15 or so years, but it'll be 
enough of a reduction so that whatever costs appear to be limits today won't 
be then.

Jim Bell
jimbell@pacifier.com