Friday, January 12, 2007

Re-Post: 200 Player Levels..How Close Are We?

Another Re-post, again from almost 10 years ago. It is great to think about how far multiplayer gaming has come since this point in time, especially looking at the MMORPG (Massive Multiplayer Online Role Playing Games). As an extra special treat for you all, once you have read the wondering article I wrote, scoll on down a little further for an interview I did, shortly after writing the article, with Charlie Wiederhold, who was a games designer at Ritual Entertainment at that time and had spoken a lot about multiplayer games with lots and lots of players.

I am having a lot of fun reading through my old website, it's bringing back a lot of things I had forgotten. For example at one point I was getting a couple of thousand hits a day and had moved the site to be part of QuestGate (a family of games related websites). Still more Re-posts to come, but for now enjoy how we once thought 200 players in one game would be amazing!

200 Player Levels..How Close Are We?

Written by Matt Large (31/10/1997)

"How many Quake players does it take to change a virtual many can you get in the room?"

The word went around a few weeks ago that because of the way QuakeWorld is coded it is theoretically possible to have up to and above 200 players on a level. This happens because it is the server that takes the network traffic hit. So to support this you would need a massive SMP server with truck loads or RAM and at least one T1 connection to the internet.

Currently there are many limitations that prevent this from being done, firstly the client graphics engine could never cop with that. We've all seen even GLQuakeWorld chug along at busy times. Secondly the required size of level cannot be supported.

Even though it won't happen soon, this news prompted great debate in the online gaming community. Some thought the idea was great, others thought it was terrible. However, following in the great tradition of previous online debates, the majority realised the current impracticalities and debated the maximum current size and future developments.

Serial-Cable Killers to Serial Killers.

Multiplayer games have evolved very quickly during the 90's. Previous to this home gamers had to use serial cables and play two payer games. The lucky ones had access to LAN technologies, but these could take hours to setup, and then there were few games that supported them.

The first popular LAN enabled games were in the EBS (Electronic Battlefield System) series of Falcon 3/Mig 29/Hornet. The other big proponent of networks, some may say visionary, was Peter Molyneux and his company Bullfrog. Syndicat:American Revolt (the first of the Syndicate series to be networked) was an immediate success. Not because of its expansion of the levels available, these were handled badly, the learning curve being too steep, no it was the addition of network play that sold this pack.

While all this was going on with the hardcore gamers, one company was about to take networks to the masses. Id gave the world Doom in 1993 and immediately offices around the world had their networks besieged by games of Doom. While slow to react, the rest of the industry soon caught on and a slew of LAN games followed, arguably the best being Rise Of The Triad which was designed around the idea of office games, there was even a site license version!

Network games had truly moved from the Saturday serial game to the out of hours office tournaments.

Go 4th and Multiply.

Doom and its impersonators gave the world 4 player games, Syndicate:American Revolt increased this to 8 player games but beyond that there was nothing. This wasn't surprising. The network code in Doom had made even these small games seem as if they were eating bandwidth and you were unlikely to find more than 8 players at once in most buildings.

Then iDoom was released, this allowed Doom to be played over the internet, again it terrible network code meant that the minimum access was ISDN, but it did start the ball rolling.

With the internet as a forum for arranging and playing games it was no problem finding 3 like minded people. The problem now became that of frustration at only being allowed 4 players.

Id came to the rescue again, rumours of 8 player plus games being played with their soon to be release 'Quake' engine were ripe. Then in a surprise move Id released 'Qtest', a demo of just the network code, to test the internet play. Players the world over went mad for it. Not only did it allow up to 16 player but it provided a dedicated server which reduced latency and improved it interaction with other network resources.

Immediately clans were formed, 24 hours a day servers went up and using the innovative QuakeC new styles of play were added, including the now infamous Capture The Flag (CTF). The population of the world of online games "praised be" to John Carmack, he'd shown them the way forward. But many forgot that this was a man for whom a completed game held no interest at all. While gamers enjoyed the freed of 16 player games he worked on the next generation.

New Worlds.

the proliferation of internet Quake games again spurred the industry on. Many, at first, forcing their current projects to accept internet play, a mistake that would see many fine games fall at the wayside. other companies promised all and delivered little.

Many people rested there hopes on the first bastion of network play, Peter Molyneux and on his rumoured tour de force game 'Dungeon Keeper'. Rumours had promised a game that would allow constant play in player created dungeons, a system that would learn your style of play and keep your dungeon for you when you weren't there. First one, then many slips, caused by his decision to leave Bullfrog after the takeover by Electronic Arts, forced this game to be held. When it was released mid 1997 it was acclaimed from all around, even without some of the advance AI functions promised.

While others worked, Lucasarts missing the boat completely with Dark Forces and then producing lackluster play and performance with X-Wing Vs Tie Fighter, John Carmack had a pet project. He called it 'QuakeWorld'.

QuakeWorld changed and added to the original. It was multiplayer only and required a dedicated server, but promised lower latency and improved facilities for arranging games.

When used QuakeWorld allowed automatic downloading of skins, patches and levels. Interfaces were provided to allow game info such as max and current player numbers, levels being played etc to be viewed with out the need to join. This was to be use by 'QuakeSpy', a program that checked against a list of servers and offered a simple graphical way to join games.

However, it was the innovative predictive routines that made QuakeWorld so good. With one command, 'pushlatency', a player could invoke a client side system that predicted what was happening at times when the network information was delayed. With this system 32 players, with decently low latency, could play in one arena.

Beyond QuakeWorld.

So back to where we started. Currently the online gaming community waits with baited breath for the release of Quake II. We know that the network code itself is better than the original bit it seemed for a while as though, in the retail version at least, there might not be any of the QuakeWorld predictive routines. Now John Carmack has told us that they fit into the new code nicely.

With new network code, predictive routines, a better rendering engine and larger levels the future for mass online game of Quake looks good.

However, we must not forget what I said about John Carmack earlier. A finished product means little to him, so what's he working on now?

The Holy 'Trinity'.

Id's next game engine is codenamed 'Trinity' and is generations ahead of the Quake II engine in all respects. On the graphics side we will see an engine that can only be played on accelerated systems, but it is the network play that we are interested in. The news about this is sketchy, and untrustworthy. Rumours say that levels will not only span into each other as can be seen to some degree in Quake II (follow the view of the big gun as you progress) but also into other server. This would allow you to look out over levels on other servers from the level you are on. Finally, we will see players running around holding the weapons they're using as opposed to a generic gun, thus allowing you to play different tactics.

Another change to remember will be how all of this is interfaced to the user. We now have GameSpy, which is a version of QuakeSpy that is expandable to deal with all Quake engine games. How will this system work with servers that interact with each other? Data feedback in to live web pages will also enhance tournaments and clan play. How about a Java applet in a web page that provides a lurkers view of a running match before entering.

One thing is for certain, the future of network gaming looks good. Now id only Lucasarts would release a dedicated server for X-Wing Vs Tie Fighter, with a drop in, drop out deathmatch mode!

Week Begining: 29/12/97

Thur 1st January:

Charlie Wiederhold on all things deathmatch:

I sent Charlie Wiederhold, games designer at Ritual Entertainment, two e-mails on the 16th and 20th of December, last year now, and tonight I recieved his replies. Ritual are currently working on the Quake II engine game SiN which is set in the year 2097 and at last is a game that features a female lead villan. She is the brillian biochemist that is the source of the drug U4 that the players HARDCORPS law enforcement agency, privatised of course, is trying to stop. Not only is this vile, but probably beautiful, woman releasing this drug but she also plans use it to force evolutionary changes to the human race, why I don't know but she does. Well that's enough plot, onto what Charlie has to say;

I wanted Charlie's opinions on some of the key elements that are/will be shaping deathmatch play in the near future, firstly I wanted to know how he felt about the current trend to have more and more players in one arena.

> I agree that we are some way off having the infrastructure in the community
> for regular 100+ player games, but do you think that moves towards larger
> arenas is a good thing? Would you still prefer to play a select few
> individuals in a confined level?

"I definately think that playing a selected few will always be the
most popular way to play. However, I think there is always a place
for large amounts of people. It's like a bunch of kids going to the
park to play football or baseball. You may have way to many people to
play a real game... but you make two teams and everyone plays and has
a good time. I definately like the idea of eventually watching large
scale *virtual* battles. Like with a complete command structure and
everything. Talk about a simulation that could be a lot of fun.
Unfortunately we are a loooong way of from something like that really
working well."

So how will these "large scale *virtual* battles" be acomplished in level designing terms? I asked Charlie about his recent comments on designing these big arenas.

> I read your .plans including the ideas about having large
> vertical maps and shutting off lifts or wide ones and shutting down
> teleporters. These would make 100+ player levels feasible but what do you
> think could make them "great" to play in.?

"Well, it is more of just losing some of the creative aspect. If you
are forced to make sections and then have looong elevators you are
bound to that design concept. I don't like that and don't really like
it as an option. It wouldn't affect play too much except for the fact
that the map would be in definate "chunks"."

> Both the 32 player project and the CTF expansion project created some nice
> 32 player levels but these took a long time to tweak until the played well
> with that many players. How long would it take to create and then tweak a
> 100+ player map?

"I really couldn't say... since there hasn't been one done yet. But
the logical guess is at least a bit longer than the 32 player maps."

Finally a subject which I seem to be fascinated by at the moment, for some unknown reason, and that is Game Logging. I asked Charlie about the recently proposed standard for DeathMatch logging.

> An additional question to the last lot I sent you (sory to bother you
> again), but are you planning to support Fritz's (the author of GibStats)
> Logging Standard Proposal?

"I have no idea, that is something that would most likely have to be
decided when we start pulling all of the deathmatch elements

I'd like to say thank you to Charlie for taking the time out to reply to my questions, I know what a bad time coming back off holiday and trying to get back to work can be without someone firing questions at you.