[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: computer-go: Authenticating the identity of a remote go-playing computer program
Forgive me if I seem a little behind the current discussion on
digital signatures, I'm stepping back a little, to the basics.
I don't agree at all with limiting algorithms to be
deterministic. This is too limiting to the developers and
especially the AI field in general as applied to the game of Go. I
think this would cut too many opportunities out of competition.
It would be a shame to have a "limited" tournament become
standard, simply because no-one has to travel. By limited, I mean the
limitation of algorithms. Human nature tends toward laziness - this type
of tournament could become very popular. Do you want a de-facto standard
to embrace this serious limitation of algorithms? It will lead to a
dead-end road, and in the end serious competitors will not be associated
with it.
The idea (as I see it) is to keep people from having to fly around
the country, not necessarily that the internet (which requires a network
card and drivers) MUST be used. If network is eliminated, it's a
*little* easier to eliminate unauthorized inputs, by not allowing any
network hardware or drivers to be installed on the competing computer.
Using the internet is definitely more friendly, but isn't the only
solution. Competitions are still using standard, non-network, modem
protocol for play.
It seems that the core of the problems is to determine if a
program/computer is receiving data from somewhere other than algorithms.
Why not just disable all hardware input ports which are not required
for a Go program to operate in the competition? (keyboard, mouse,
network, com ports, printer ports etc). Isn't there some way to disable
inputs from outside the computer?
Couldn't this be done by using a device driver which controls and
monitors hardware in the computer? (mouse, keyboard, and other ports of
communication) Maybe this requires a standard hardware
configuration, where the driver won't even load if certain hardware (like
network cards and keyboards) are present. Also, the competition
"organization" would need to distribute the device driver(s).
Of course, this requires writing compatible drivers for different
operating systems. Authentication could be included, where two
different drivers are used: one for developer integration, which is
released early, and another for competition which is released just before
competition.
The competition could take place by dial-up modem, where a server
dials the competitors, instead of the other way around. Since long
distance charges are involved, competitions would need sponsorship,
either from sponsors or entry fees. This doesn't seem like a big problem,
especially with the cheap long distance rates these days.
All this could be applied to the official "remote"
tournaments, but practice sessions could still be held over the internet
with no authentication whatsoever. As someone already pointed out
(but not in the same words), you can cheat all you want in practice, but
not at the competition, where "the men are separated from the
boys", e.g. the players from the cheaters.
I'm not intending to open a can of worms here, I just wanted to
throw out an opinion.
-----------------------------------------------------------------
Steve
s_korfhage@xxxxxxxxxxxxxxxxx
korfhage@xxxxxxxxxxxxxxxxx