Network & Multiplayer Games 96 Hat

Developing Multi-Player Games from Scratch
Adapting Existing Single-Player Games
for the Multi-Player Environment


          Single-player games came first.
          All the major games companies in the world got where they are today by producing single-player games. There are, of course, a number of relatively successful companies which have concentrated on multi-player games, the most obvious of which is Kesmai Inc.; even so, the top 20 - perhaps the top 50 - computer games development companies all have a background in single-player games.
          Things are, of course, changing. Most of the major games releases you can buy in the shops today, for the PC at least, have multi-player options (whether by LAN, modem head-to-head, Internet, or all three). In many cases, the games have even been designed so they're great fun to play this way. However, in the main they're basically the same, old, single-player games with their AI modules replaced by comms modules. Lack of a multi- player option reduces the scores that games are awarded in the all-important magazine reviews, and, since it's not too so difficult to add a multi-player capability, developers include such an option; whether the game is any good multi-player is not especially important (although there are honourable exceptions).
          My contention is that the way forward for multi-player games is to design them from the ground up to be uncompromisingly server-based, multi-player, online games: the multi-player games of the future should not simply be single-player games that allow extra players. (Corollary: single-player games should not simply be multi-player games that have AIs instead of other players, although that isn't what I intend to argue).
          Note that I make a distinction between online multi-player games (which are what we'll get) and peer-to-peer multi-player games (which are what we have).

Player Psychology

          Online multi-player games make different demands on players to those made by single-player games, or peer-to-peer multi-player games:

          Players play for different reasons.
          For single-player games, the chief source of amusement is usually puzzle-solving or beating the game's AI (or, if it doesn't have one, its random-number generator and the clock). For peer-to-peer games, the fun comes from beating other players; there may be team play, but the goal is the same.
          For multi-player, online games, the player wants a more rounded experience; people are as important as gameplay. Some of these people will be enemies, and others will be can't do this in peer-to-peer games.

          Players stop playing for different reasons.
          Single-player games lose their lustre for a number of reasons. They will usually be played intensively after being purchased, but the player will quite often begin to find flaws in the gameplay, or in the way the game's reality measures up to its pre-launch hype. Those games which are well-designed enough to satisfy the player will typically occupy them for some time afterwards, until the gameplay gets samey enough for the player to find playing a chore. This will normally happen a few months, or, rarely (e.g. Civilization) one or two years after purchase.
          Peer-to-peer games have a fad-like quality. Everybody spends their lunch hour playing multi-player Doom until they get bored, then they stop. Multi-player Quake will probably go the same way. A group of people must all collectively want to play; it's an aspect of the group's sociability, but it doesn't moderate their society. Online, multi-player games hold on to their players for much, much longer (if the games are any good - or, often, even if they're not!). Players will either leave after a few hours because they can't get into a game, or they'll stay for ages. I have people playing my online games who have been doing so for 9 or 10 years. When players do leave, it is often due to external factors (new job, moving house, huge phone bill), but there is a fair degree of churn due to the attitude and actions of other players in the game. If an environment ceases to be an attractive place, the players will move out. If another environment appears which is more attractive, they may also move out.

          Players play at different times
          Single-player games have no timing problems; players can play them whenever they have the time. Peer-to-peer games have big timing problems; players can only have a proper game when all the players are ready, and if they all want to play rather than go down the pub for a game of darts.
          Multi-player, online games have players coming in at all times. People can have already been playing in a long session when someone else arrives, and unless the game is designed to anticipate this the new player could be seriously disadvantaged. This enforced variety is also a strength, however, in that it increases the depth of the game if components of it have been especially created with this in mind.

          Players play for different times.
          Single-player games can be played in sessions lasting from minutes to hours. Peer- to-peer multi-player games are usually of a fixed duration implicitly agreed upon by the participants at the outset.
          Multi-player online games are never-ending, and players play for as long as they are having fun. Many will plan on having a session of a certain length, but how long exactly they do play for varies depending on what's going on. Usually, they will either "pop in for a few minutes" or have a definite goal in mind beforehand which could take them several hours to complete. The provision of such goals is an important but frequently-overlooked aspect of multi-player game design. Quite often, though, players will drop by just to see what's going on; when this happens, you know your game is a success. death. For single-player games, this does not matter; competition is only verbal, of the "your top score is only 115%, ha ha ha" variety. For peer-to-peer games, it does matter; the inexperienced player will be ritually slaughtered.
          For multi-player, online games, it has not to matter. If it did matter, the game would not appeal to new players. There are a number of strategies for ensuring that newcomers aren't put off by the show-off antics of other players, ranging from off-line training sessions to hard-coded newbie protection schemes. As a cautionary note, it must not be the case that experienced players are handicapped to an extent that gives novices a huge advantage; rather, the aim is for old-timers either to take new players under their wings, or to be completely unconcerned by them.

          Different numbers of players play.
          It may seem obvious, but there are differences in the way that players treat games which have more than one player. Cheating in a single-player game is fine if you're the sort of person who can justify it to yourself; it is not fine in multi-player games, and must be dealt with ruthlessly.
          In a big, online world there could be 300 other players playing at the same time; this gives the world a bustle that leads to an entirely different gaming experience, and players are likely to have to choose their activities in the light of what else is going on, rather than simply what they want to do.

Constraints on Games

          The above issues are reflected in a number of constraints which must be borne by multi-player games.

          As has already been mentioned, there is a distinction between those multi-player games which are server-based, and those which are peer-to-peer. The latter are the logical extensions of single-player games, and are the type of multi-player game we see on the market at the moment. They don't need a massive amount of computing power, because they don't deal with the kind of numbers of players that online games do. Also, because they use game engines and libraries developed for the single-player market, they can be put together much more easily.
          Multi-player, online games do need to be fast; when a command comes in, the player wants it processed now. Sometimes, there are periods when few commands come in (during which time background tasks can be performed), but at other times a whole bunch of commands can come in simultaneously and the machine has to be able to deal with them expediently. Fortunately, the grunt graphical work can be done by a client at the user end, although unfortunately that does mean two separate pieces of software are required to talk to one another, and only one (the client) uses a great deal of software already in developers' toolboxes.

          Robustness and reliability.
Single-player games can crash, which is just as well because many of them do. LAN-based games can crash; in doing so they cause groans among the players, but they're easy enough to restart, and it's not as if anyone would be losing the fruits of hours of effort as a result.
          Multi-player, online games can't crash. If they do, many people will be inconvenienced, and they'll all complain together. People may have been setting up their position for an hour or more, and although their score and character stats will normally be saved as the game tumbles over, their overall situation will not be, nor could it ever be because it's relative to the other players. People are seriously annoyed when a multi-player server-based game goes down, unless they have been sufficiently warned beforehand that it is going to happen.

          Depth and lifetime.
          Multi-player, online games are played over a very long period of time, principally because of the ties of the community which develops. However, they must have enough in-built gameplay to support this; talkers have a much greater churn than games, as there are no "events" happening. Because of this, online games typically have to be more open-ended than single-player ones: they need to have more options available, and to be able to address specific types of users such as newbies, old-timers, people who want to play all evening, and people who want a 15-minute diversion over their lunch hour. As mentioned earlier, though, it is often the case that by facing up to the issue of having different player types, an increased game depth follows; this in turn can lead to a more satisfying gameplay.
          The main point where many single-player puzzle-based games fall down is replayability. It would be perfectly possible to program a How to Host a Murder style game online, but people could only play it once. Worse, measures would have to be taken to stop people who had played it before from playing it a second time, so they wouldn't spoil it for everyone else. Puzzles in an online game are perfectly OK, but they should be comparatively numerous, of varying degrees of difficulty, and very much secondary to the problem-solving required as a result of the activities of other players.

          Certain types of game cannot be successfully implemented in a multi-user environment, particularly those involving time travel or the destruction of shared resources. Many games of a risqué nature are also often best played in private... Complicated narratives don't work well in multi-player games, which tend to thrive by offering a free choice between many courses of action. Also, by virtue of the fact that many people want to play at once, the genre has to be popular: SF and Fantasy dominate.

          Size of world.
          The playing area of a single-player game does not have to be huge; even a few hundred squares or hexes or whatever can seem vast. In multi-player games, it's the population density which matters: too few locations per player and it feels impossibly crowded; too many, and it ceases to spark enough interaction to be truly multi-player. This isn't too important in peer-to-peer games, where the number of players won't exceed some fairly low value like 8. It is important in online games, though.
          If the number of players in a game is likely to fluctuate wildly, it's best if the game is self-regulating, allowing areas to open up or close down depending on how many people are in it at the time. Thus, when it's 2am and only 20 people are playing, the quests
          Single-player games have a winner. Peer-to-peer games may or may not have a winner, but they're certainly competitive. Multi-player online games do not have a winner. There may be some status level beyond which progress cannot be made, in which case the player can be said to have "won", but that doesn't mean other players can't win later. Consideration of the final "goal" of a long-term, multi-player game is something which needs to be built in to the design at the very start.
          It is possible to have winners for session games, which are usually of the Space Opera "conquer the universe" theme. These would typically have a lifetime of, say, 3 months, and many of them would run in parallel with different start dates. This works best with play-by-email or "batch play" games, however, and suffers badly from player dropout if they see their situation as untenable. Thus, teams of players tend to be encouraged, so lots of people "win" when it finishes.

          Player pool.
          Online games each want a slice of the player cake. However, if one game takes a big slice, the other games have to lose a piece off theirs, and some may get none at all. It's made worse in that games need a critical mass of players: below a certain number, everyone will stop playing and the game will be dead. This is not a problem with single-player games, of course. Although multi-player games can provide interesting things for people to do if they are alone, rarely is this enough to keep them occupied for hours (because if it were, the game would effectively be two games combined, one single-player and one multi- player).

Legacy Design

          The way that games are written at the moment, players and network providers unwittingly conspire with developers to maintain the single-player paradigm. Players do get multi-player options for the single-player games they buy, but they don't get the kind of multi-player experience which causes all the fuss that made them want a multi-player option in the first place.
          There are several levels at which this operates:

          What players say they want doesn't necessarily mean they do want it.
          People will passionately advocate something in order to seem hip and trendy (or to show that they remember the old days when games were great and computers were 8-bit). The former, bandwagon effect is the one which developers pay attention to, because if they jump on it then they can be seen as a progressive outfit. Hence, developers hear what gamers say they want, then give them it whether they really want it or not. The worst offence is to be old hat; this is what killed off text adventures.

          Players will say they want something because its very premise is interesting, although when they see the finished product they may decide otherwise.
          Multi-player games fall into this category, in that the idea of playing other people is very attractive, but it may not necessarily turn out to be quite so nice when the only people you can beat are aged under 12. The trick is to isolate what it is that people like about an idea and implement that, rather than implement something which fits the general category. Even then, the results can be discouraging: people may genuinely like the idea of seeing a game in stereoscopic 3D, but when they try on the headache-inducing glasses they may defer buying it.

          There are some things that players really do want, but which they won't believe they actually want until they are forced to experience them.
          In many respects, MUDs are like this. Whatever the Internet killer application finally turns out to be, the chances are that at the moment people don't know they want it.

          What the developers can realistically provide, 2 years down the line, is based on what they can do now. Now, they can write single-player games, and design them in such a way that they can be played by a limited number of extra players instead of AI modules. That is therefore what they provide; however, it's still basically the same, single-player paradigm.
          The problem is that writing single-player games is not like writing multi-player, online games, and the peer-to-peer games we're getting at the moment, while good in their own right, are nothing like the online games that are going to make pots of money.
          To use an analogy, we're asking people who design speedboats to design passenger liners. There are many similarities, but in general you don't design a passenger liner by taking a speedboat and adding more seats. What's more, if you did that, you'd lose a lot of what's fun about a speedboat.
          Some of the features which give single-player games an advantage over multi-player ones (turn-based strategy games, narrative-driven adventures, high-tech defender against hordes of low-tech attackers) are nowadays ditched at the design stage so that a game can be made multi-player. This means that in the quest for multi-player capabilities, sadly we're no longer getting the variety of single-player games we used to get.
          The way to design a passenger liner is to design it from scratch as a passenger liner.

Product Support

          Online games are a commitment.
          With a single-player game, you finish it as close to the deadline as possible, you get a distributor to sell it, you take your money, and you reluctantly send out patches to correct major bugs. You have a help-line and a web page for people who wish to bother you, staffed by trained personnel who can answer questions about all the games you carry while play-testing your new ones.
          Peer-to-peer games follow exactly the same path.
          With an online, multi-player game, you get your basic, stable version working, and you make it available. You take your money, and you continue to do so while people play it. However, you have to ensure that bugs are fixed very quickly, and you have to improve the game gradually so as to add the depth it will come to need. Your support staff are people managers who spend all their time logged in to the game ensuring that the players are behaving as they are supposed to behave, and acting as cheerleaders to encourage people to play more often than they otherwise might.
          These people-managers are also dedicated players. This is good, because it means there is always someone playing in the game. Nothing puts people off more than an empty multi-player world: there may well be 50 people waiting to play just as soon as someone else appears, but, if no-one does, they don't.


          The main sources of income for a game are:
  • selling the software needed to play it
  • charging people to play it
  • merchandising
  • sponsorship/advertising.
          The key factor which multi-player online games have that single-player or peer-to- peer don't is the ability to charge people money every time they play. In the publishing world, this would be like printing a book and taking a royalty every time someone read it. With peer-to-peer games, the per-game charging is either non-existent, goes to the telecommunications company, or goes to a gamer introduction service. This is fine if you are the telecommunications company, but if you're not then you're going to the trouble of making your games multi-player but someone else is making all the money that follows from it. In fact, it's worse: if people are still playing your last game when you bring your new one out, will they want to spend money on this new one?
          Note that games which are only played once are no good as online games. Returning to the How to Host a Murder analogy, it would take you forever to recover the development costs for an 8-player game, even if the players paid premium rates (£10+ a session) and the game engine could deal with a large number of pre-written scenarios.

          Multi-player games can still sell as boxed units. When Air Warrior was launched in the UK, it made practically all of its money from sales of boxes. Roughly speaking, for every 200 units sold, 1 player actually tried the game out. That was four or five years ago now, though, and things have changed somewhat. Nevertheless, Meridian 59 still manages to bring out client software at £39.95 a pop. Whether they'll make money from their online charges remains to be seen, but the chances are they will do rather well if the game has enough depth.
          Server-based games will therefore win out in the long term, because they have a revenue stream unavailable to single-player games: pay-to-play. Like pay-per-view TV, if the quality is there, people will indeed reach for their credit cards. The market will eventually become saturated, of course, but that will only serve to increase the quality of what's available as true competition takes over. It will take a long time to reach full capacity, though: the current list of publicly-available MUDs has 675 games on it, whereas last year at this time it had 538; the actual total is probably 50% more than this, and these games aren't even professional quality.
          There'll still be a market for single-player games, because they're a different breed; the single-player market may even remain a bigger market than that of online games. However, will I be able to find someone in 5 years' time who will want to play multi-player Warcraft II with me? Probably not. Will I be able to play Meridian 59? Probably. Will I be able to play a text-based MUD? Definitely.


          The multi-player games which are going to make money are the ones that have a server, because they are the only ones which can keep players playing and paying for years and years and years. Peer-to-peer facilities in games may sell marginally more boxes, but they're one-off payments by the customer. It would really, really help if Wireplay and their ilk could pay royalties, but can they?
          Single-player or peer-to-peer games will still be around and will still sell like they always did.
          Those developers who think they will gain all the benefits of a multi-player online game by merely adding multi-player capability to games designed for single players are mistaken. Such a feature may be all well and good as a means to stop writers from games magazines complaining about if it's missing, and it may sell you a few more units, but it won't generate a revenue stream.
          Those developers who think they will gain all the benefits of a multi-player online game by designing in a multi-player option into their game from the beginning are mistaken. It's not an option, it's a defining quality! It might give you better head-to-head play, which could keep people interested in the game for some time, but will it sell you more units? Or will the fact that people are still playing your old games mean that they'll see less reason to go out and buy new ones?
          In the long term, server-based games are going to make some people very rich. In the short term, developers are going to continue to add multi-player options for "Internet play" to their products and wonder what all the fuss concerning Internet games is all about.

          Two things will remedy this:
  1. The success of risk-takers. Meridian 59 is a first step, but it's not a marvellous game. Ultima Online will probably be the turning point.
  2. Time. Developers will eventually work all this out for themselves, especially those who attend conferences such as this one and have it all pointed out to them..!

Copyright © Richard A. Bartle (
7th May :\webdes~1\ nmg96.htm