or:
one could produce the code to be run as specialized zip files, filled with
some-odd Java class files, and any data files;
the zip file is recieved, unpacked into a local directory, then executed with a
specialized/modified Java VM (mostly modified to keep track of what things are
run);
any changed files are then zipped and sent back.
potentially, the protocol used could be HTTP.
ok, so in this case, the server could probably be a modified version of
Apache...
so, the costs could be:
initial setup costs (costs associated with submitting the work request and
setting up the process);
library linkage costs (many utility libraries could require royalties);
running time costs (both a unit-time cost for processor time, and also a
linear-time royalty fee per-library);
bandwidth costs (mostly based on the amount of data to be sent to/from the
server).
after everything is run, the costs are calculated and billed to the user, for
example:
initial setup cost $0.05;
per-library royalty cost $0.01;
processor-time cost $0.05 per second;
processor-time royalty cost $0.005 per second;
heap usage cost $0.01 per MB;
bandwidth costs $0.05 per MB;
..
all this could be calculated though by the servers depending on load, rent,
costs of electrical power, profit margin, ...
the owners then would have a number of in-house servers, and could potentially
rent time on other peoples' systems as well...
----- Original Message -----
From: Kjell Godo
To: Fundamentals of New Computing
Sent: Monday, December 01, 2008 1:45 PM
Subject: Re: [fonc]( picoVerse-:( LambdaLisp Assembler needs testing ) )
I remember a book on mobile Agents I read once.
These ideas have been worked on in the past but they never
got off the ground. Have things changed? Are the impediments
still there?
The idea is that you put some money into an Agent Object.
I'm thinking a penny or less. And you send this Agent out
and it negotiates with other Objects to try to find ones that
it can use that will fit its budget.
So there has to be some way for Objects to sell themselves.
A server Object looks into your system and sees what it needs
to run. And downloads and installs those things and then
installs itself. It has to sit in some protected space where it
won't be fiddled with. If it gets fiddled with then an international
law is broken and you can get prosecuted for it.
Can an open source thing provide safe places for server Objects
that would protect against theft of services? If those services
were almost free then perhaps the cost of stealing them would
be high enough that people just wouldn't steal.
Perhaps such things would need to be built into the hardware?
Perhaps in order to participate you need to buy a card that
lets you in. People would make cards that let you in for
free. Is there any way to stop this?
So the problem is that people will make a free version of the
Xanadu system. How can this be prevented?
Perhaps in order to use the system you have to let Agents
come onto your system that scan the virtual machine in
unpredictable ways to see if it has been monkeyed with?
And if they don't return with the propper results then a
gestapo comes and kicks in your door and scans your
computer for illegal software and if it finds any then off to
jail you go. But then could those illegal softwares have
been put there by hostile Agents in order to take you out?
It's a problem.
How could a Xanadu system be built?
What is required?
On Sun, Nov 30, 2008 at 7:00 PM, Alessandro Warth <[EMAIL PROTECTED]> wrote:
Xanadu was an idea by the guy who invented hypertext. I forget
his name.
Ted Nelson.
On Sun, Oct 19, 2008 at 7:51 PM, Michael FIG <[EMAIL PROTECTED]> wrote:
"Kjell Godo" <[EMAIL PROTECTED]> writes:
> And some internet way to stop piracy can be used. There needs to be
> an electronic currency. millionths of a cent. each function call
> can be paid for. true cloud computing. And then we incorporate OOP
> hardware. And then we incorporate the ultimate OOP hardware. DNA.
With a clever use of quantum entanglement, you could just alter the
DNA to note the original software authors for each stimulus that a
"user" receives, and report the authors of the received stimuli to a
central authority via neutrino broadband. Then, the authority can
bill people on behalf of the authors based on how much of the system
was actually observed.
So, if you were watching a movie online, then walked out of the room,
you'd be billed less for that time because you didn't receive as much
audio stimulation, and no photons.
It should all be quite simple, and finally close the "analog hole"
that has plagued similar efforts in the past.
--
Michael FIG <[EMAIL PROTECTED]> //\
http://michael.fig.org/ \//
_______________________________________________
fonc mailing list
fonc@vpri.org
http://vpri.org/mailman/listinfo/fonc
_______________________________________________
fonc mailing list
fonc@vpri.org
http://vpri.org/mailman/listinfo/fonc
_______________________________________________
fonc mailing list
fonc@vpri.org
http://vpri.org/mailman/listinfo/fonc
------------------------------------------------------------------------------
_______________________________________________
fonc mailing list
fonc@vpri.org
http://vpri.org/mailman/listinfo/fonc
_______________________________________________
fonc mailing list
fonc@vpri.org
http://vpri.org/mailman/listinfo/fonc
