DOOM 3 will use P2P System?

Ant writes "From Page 6 of FiringSquad's QuakeCon 2002 Postmortem article: John Carmack said something at the end of the Q&A about how the multiplayer will be only four players? Tim: After 2 hours of talking up at the podium, sometimes you leave a few details out. Doom 3 multiplayer will be fully scalable. It will be a peer to peer system. We haven't started working on it yet. Tell everyone not to panic - it will be fine. John just forgot to mention it'll be scalable past four players. It's hard to give a hard number because we haven't started working on it yet. Right now we're focused on making Doom 3 a kickass, over the top single player game."

Comments

DOOM 3 will use P2P System?

[James Foster]

Why is there a question mark after the title of this? It's been stated by id, that DOOM 3 will use a peer-to-peer network architecture. There's nothing uncertain about that.
This has been known for around about an entire week now... it's been stated multiple times.

Also, to clarify, when they say "peer-to-peer", they don't mean a network of users like Kazaa or file sharing applications, they mean that it is client-to-client as opposed to client-to-server.
The best way of thinking about DOOM 3's multiplayer is as being the same as the original DOOM's multiplayer. 4 players, and no such thing as a "server".
The only actual uncertainty is the 4 player limit. It was initially mentioned, but now Willits has said that it is scalable beyond that... This is unclear as we don't know if he means that the game can go beyond that, but the network code is ideal for 4 players, or means that the game will have a hard limit of 4 players, but mods and games based on the engine will be able to scale beyond 4 players.

Also, it is known for definate that once a game has started, additional players cannot join. This limitation is due mainly to DOOM 3's physics engine. Basically, there is so much physics data that would need to be synchronised, that if a player had to "catch up" with the physics data, it would probably be a lot of data to send, and since it's constantly changing data, it is likely that as the player recieves the data, it becomes invalid.

It will be interesting to see how other games deal with the problem of physics data. As physics engines in games become increasingly complex, it will become harder for programmers to cope with players joining a game that has already started. Perhaps if all games employed "rounds" (like Counter-Strike), then player's wouldn't have to wait long until the game restarts and they can start playing. This already has to happen when a player joins a Counter-Strike game that's already in play.

Re:why?

[Reziac]

Probably hard marketing facts:

At the height of the Quake-online frenzy, Doomworld ran a poll asking how many people played in each mode: single, deathmatch, coop. Turns out solo players outnumbered DMers by 4 to 1, and coop players by 20 to 1. (Sample size was several thousand, so statistically significant.)

DOOM (primarily a solo game) outsold all versions of Quake (primarily a multiplayer game) *combined* by at least 3 to 1. And that's even tho DOOM came out when home computers were still a relative novelty, and priced out of many people's reach. By the time Quake came along, most households already had a computer (and PCs cost a lot less too). So -- Quake didn't sell as well even tho more people had PCs by then. Obviously, something went wrong with the spectrum of Quake's market appeal, and consensus is the lack of really good solo play.

I'd hazard a guess that DOOM3 won't really be playable over the net unless you have broadband. Which would artificially limit its market to the small subset of net users who actually have broadband (the last figure I saw was under 20%). Which would be stupid, from a sales standpoint.

In short, the single player market is a helluva lot larger than the multiplayer market. And idSoftware is really in the game *engine* business, which multiplies that market by a factor of however many companies they license their code to.