I'm not sure which FPS's you play on which servers, but I play Counter Strike Source games all the time on 44 player servers that get less lag than Diablo II games. In fact, I only play on servers with pings of 70 and less.

The thing that I'd like to point out is that it's as foolish to assume that a vaguely defined system will fail as it is to assume that the same system would work like magic. But, I suppose folks here are jaded...everyone and their grandmother thinks that they'll develop the next great MMO.

My aspirations aren't that high, of course.

In any case, I plan to decrease the problem of lag by using this system:
(Bear with me, I'm writing this at one A.M. and I haven't slept in 18 hours)

Essentially, the servers record only player commands, just like almost every other MMO. I’ll assume that you all already know the basics of netcode, so I won’t burden you.

The difference between this concept and the system used in a game like World of Warcraft, lets say, is that combat isn’t handled in the same way. In WoW, the server can run a combat loop and just check for distance between you and your opponent, and where you are relative to them (in front, behind, etc). In the system I propose, combat and actions aren’t tied to objects. You have server clusters with each server dedicated to performing a certain task. The internal network between the servers could use network equipment on the caliber of what was used for the Virginia Tech G5 supercluster, just without the 1100 computers… In any case, one server handles collision detection, or, depending on load, multiple servers could be assigned to more processor intensive tasks. Interface commands, and various sections of game play are separated in a similar fashion, and each server sends its packets across the network unless:

A.) New data will specifically interfere with data waiting to be processed by another server

B.) Obvious example: Someone’s cheating, and an operation is illegal. This is a given.

This means that each server can operate independently of the others, but this would only be used in certain cases. For instance, the interface doesn’t have to lag because the collision detection algorithm is handling a lot of data. Distributing the processor load reduces latency by removing various bottlenecks in the system. I don’t know much about systems currently used in online games, so it’s likely someone’s already doing this.

So you have your gaggle of computers, processing in tandem. Since internal network bandwidth consumption isn’t an issue, the servers can chat with each other as fast as they can process. The servers will constantly be relaying information, and a series of security servers will analyze the network data from every angle, independent of the other tasks unless a problem is discovered. So, what you have is this:

1. Player requests to attack
2. Clientside, the player begins to attack their target, predicting the outcome of the server’s decision
3. Server decides whether or not to veto the client’s request
4. Assuming the request to attack is valid, the operation proceeds
5. The server has received a call to perform the action “attack”, from a specific player
6. The server gets the players information from internal storage
7. The server tells all clients that are within range and/or can see the character that an attack has been generated for the character that called it
8. Server runs the outcome of the situation
9. Server checks to see if clients had the same outcome

This boils down to relaying basic network messages. When a player enters the system, that player’s information (character information) is relayed to all the applicable clients (whoever’s in range). The clients aren’t trusted to do anything on their own, the server always runs a simple check, to make sure that the clients have arrived at the same outcome as the server has. If a client is out of sync, the server will correct them, just like in every other online game out there. The key is to allow parallel processing of tasks, and realize that a real-time MMORPG doesn’t require a drastically higher amount of network traffic than a more turn based version. If parallel processing was taken advantage of to its full extent, server side lag issues could be severely reduced, since it would be rare that all of the servers in the system succumbed to the same problem simultaneously. If you can play a forty player game of CS Source on one server with 70 ms pings, this is just as doable. It’s highly unlikely that an area would be filled with an inordinate amount of clients, requiring serious bandwidth in one area, but in that case, a parallel processing system is scalable. The servers that decide which clusters get which data could simply account for bias in bandwidth and processing consumption, and dedicate more energy to high population areas.

My thoughts are that most MMO’s already use these systems, and that they aren’t very effective. However, I’m not an industry insider, so I wouldn’t know. Thoughts?

By the way, Orron, thanks for a post that said more than just: “Everyone’s done this before and you’re going to fail.”

Edit:

I also factored in the fact that revenue for ads viewed is in the $60-70 per thousand views. That still puts you at a lofty (in relative terms) $6500 per 100,000 views. Assuming that you had four million views a month (Lineage has almost that many users. That'd be assuming that each logs on once or twice a month) that would equate to revenues of about $260,000 a month. Not much on a corporate register. But, if you account for the fact that it's likely that most of the users will log in at least once per month, and serious players will often log in five times per day, you can figure much higher than $260,000 a month. How do you think Google makes their money?

Quote:

Original post by Nytehauq I also factored in the fact that revenue for ads viewed is in the $60-70 per thousand views. That still puts you at a lofty (in relative terms) $6500 per 100,000 views. Assuming that you had four million views a month (Lineage has almost that many users. That'd be assuming that each logs on once or twice a month) that would equate to revenues of about $260,000 a month. Not much on a corporate register. But, if you account for the fact that it's likely that most of the users will log in at least once per month, and serious players will often log in five times per day, you can figure much higher than $260,000 a month. How do you think Google makes their money?

I don't know if $70 per 1000 views is accurate, but presuming this is correct let's do some basic math:

* monthly subscription fee for MMO is ~$15 (using WoW here)
* with the ad viewing payment rate of $70 per 1000 views, a player needs to log in at least 215 times in single month to just break even with their 'flat' monthly subscription fee.
* this means 7-8 logins every single day from the same player.

Now note, even hardcore players in your example aren't expected to login as frequently... and casual players are expected to log in 5-10 times less often. (1 login every 1-2 days) This means, it's not unreasonable to expect the revenue from this kind of payment model would be 5x lower than the 'flat' monthly subscriptions model. Maybe even less.

Of course, if the lack of subscription fee will attract 5 times or more players than your typical MMO, and if you have the architecture to support them all... (that doesn't cost you significantly more than hardware you'd need to support just 1/5th of such playerbase a 'regular' MMO has) then you're all set. ^^

Indeed. But also, remember the fact that chances are, more than one ad will be viewed per login. Also, if you were to put ads on the character selection screen, or essentially all menu's outside of the game world, you could increase revenues. You could, theorhetically, make more than a subscription based RPG. However, I suppose that most companies don't want the hassle and would rather stick to a tried and true method, with gauranteed pay. They charge you for time even when you don't play, and you have to pay thrity days at a time, so they inevitably get money even if you don't play and don't cost them anything.

PS: I didn't want to post a long treatise on the technical side of the engine on the GAME DESIGN forum, for obvious reasons. But there it is, for all to see.

Fair point, i was thinking more of the current game launchers which are crammed with info as it is and don't leave much room for possible ads. ^^