would it be possible ?

Ba-ka.

Persistence, synchronization, security, and control are the problems.

Persistence means that when a player logs off, the areas that were maintained on his/her computer do not simply disappear or kick off those who are in the area. Synchronization means that if a player is the only one in an area, and thus holds the only copy of the data, what happens when the player logs off? Security has already been addressed. Control means that if the information is distributed, it''s far more difficult to change to add quests or whatever.

The obvious solution to persistence is to have enforced redundency. Give the data of an area to five people if there''s less than 5 people in the area. One problem could be maliciousness -- a bunch of people discover they are the only people with the data of an area (after all, nothing client-side is secure), then log off at the same time just to see what happens. The redundency makes the synchronization problem worse -- if A and B have copies of the data in an area, and A blows a hole in a wall with a rocket-launcher, how does B get this information. There''s another problem with security. Let''s say A, instead of doing something the game allows, like blowing a hole in a wall with a rocket launcher, instead covers the ground with powerups. B has another copy of the data. How does the game decide that A''s copy is bogus, when it could just be a hole blown in a wall with a rocket launcher.

I''ve no reason to think it''s impossible, but I wouldn''t want to have to try.

Technically possible but the results would be unplayable. Distributed processing is fine for data crunching projects like SETI because source data can be chopped up and sent out for processing and it does not matter when it comes back. It does not matter if my pc takes an hour, a day or a week to send back the data because it wont stop the project. The continued opperation of the SETI project does not rely on getting data back because the programme isn''t using the data, just processing it. The same isn''t true for a game.

A MMPOG processes data in order to make the game work. It needs to calculate where people are, what they are doing, who they are doing it to, what object they are using to do it with and in what orifice. If the game doesn''t have that data the game stops. With the unreliability of net connections and PCs you would have to have multiple copies of the same data processed to make sure you got the data back. That would greatly increase the the amount of data being sent back and forth as well as the amount of work being done (cos everything needs to be done five times). You also have extra overheads as the game has to handle splitting data, sending it out, getting it back, cross checking various copies to insure integrity etc.

Technically possible on a perfect system but the net/yourPC/MyPC is far from being a perfect system. The results would most likely run as slow an overfed pig wading through mud with fat Bob the the worlds fattest man riding on its back.



Dan Marchant
Obscure Productions
Game Development & Design consultant