Players complained.

There were two reactions. The Ultima Online approach was to rebalance the game so that PKers didn't have so much of an edge. The EverQuest approach was to remove player (character) death altogether. EQ quickly garnered more players than UO, and set the precedent. Subsequent online games designed by people who cut their teeth on EQ (which is most of them) are using the EQ model.

Well yes, it's true: too many PKs do empty your game. However, removing player death is only a short-term solution. The same people who initially complained bitterly about being killed in EQ are the same ones who are leaving now for other games because they're bored.

Without PD (it can also mean "permadeath"), there's no sense of achievement in a game. Anyone can reach the top, no matter how dim they are, simply by playing relentlessly. There's no way to prevent them. It's like running a race with a bungee jump rope attached to your waist: sure, it's a lot harder the further you go, and you might slide back a few paces where the surface is slippery, but if the rope doesn't have enough elasticity to yank you off your feet before you get to the finishing line then all it does is delay the inevitable.

People who start to play PD-free games because they see them as a challenge eventually realise that they're no such thing. Anyone with time on their hands is guaranteed to "win". From a game-player's point of view, the whole exercise is therefore pointless. As a result, persistent worlds without PD are occupied mainly by socialisers. This is fine while they have enough to talk about, but without real, human-level events it all too quickly becomes sterile. Social ties will bind players to a game for long periods, but with nothing really happening they'll eventually leave for pastures new. To keep them, you need to have either a constant influx of newbies or the regular addition of new areas so there's some different wallpaper for them to talk about.

Introducing player death helps; indeed, it may well be enough. It has to be "real" death, though – loss of character. If people can be resurrected, death is little more than a slap on the wrist. Note that "player death" is not the same as "player killing": in the former, you lost your persona because the game took it away; in the latter, you lost it because another human being took it away. Players tend to accept player death more readily than they do player killing (although they don't like either), because the root cause is the behaviour of the game's random-number generator. RNGs don't hate you, or mock you, or carry a grudge. They don't add conflict.

PKs do add conflict.

The point is, games - as games - need conflict. Conflict brings out heroes. Conflict drives narrative. Conflict is resolved - things change as a result of it. Artificial conflict doesn't cut it: conflict has to have meaning, and to have meaning it has to involve loss. A game without loss is not a game.

But players won't play if they think they can be killed. It's hard enough to persuade designers that a degree of PKing makes for a better game, let alone persuade the players. Players will go for the games where they can't be killed, then leave when they get bored, without ever linking the fact that they've become bored with the fact that they can't be killed.

A new wave of persistent worlds is on the horizon, whose designers are trying to find ways to introduce PKing without putting off players. The basic idea is a reworking of the UO approach, in that if you don't want to be Pked then you don't have to be. Most areas in the game are safe, with only a few "badland" areas where you risk being killed if you enter. The rewards in the badlands are greater, of course; if you survive, you'll have a lead on people who stay at home. It's PK by geography rather than by server.

Unfortunately, "having a lead" doesn't mean anything if the people who don't put themselves at risk can catch up simply by playing more hours than you. All you're buying yourself is time. Nevertheless, this is how most of the new games are handling it. It's a partial solution, but it still doesn't bite the bullet.

To make it worthwhile to enter the badlands, it has to be the case that you can gain tangible, gameplay rewards that simply can not be obtained elsewhere. If you could reach higher levels in the badlands than out of them, for example, then that would do it; if you could increase your character's stats beyond those of stay-at-homes; if you could buy bigger houses, or get higher skills, or bake tastier pizzas - whatever. Putting your character at risk to gain meaningful rewards that you can't get anywhere else adds the necessary sense of achievement to keep players playing.

In this sense, PD is consensual - you don't have to venture into the badlands unless you want to.

It's this "don't have to" where the crux of the issue lies. In current upcoming designs, it's literal - you really can get everything from outside the badlands that you can get in them, it just takes longer. After all, if you can't, you "have to" enter the badlands if you want to advance, which means that you're going to be PKed "against your will". The game is forcing you into the situation.

Well, no it isn't.

You have a choice: risk all for significant gain, or risk nothing to get almost to the top but not quite there. Do you want to be a hero? You get to decide. The only difference is, unlike in EQ you can't talk being a hero unless you act being one.

There is another choice, of course. If you don't like your risk-free advancement being capped, you stop playing, simple as that. You just tell the designers straight up: if you don't change this, my entire guild will leave and play some other game – one with no cruel, cynical designers like YOU robbing us of months of playing time in an instant.

Except you won't. To get into this kind of situation, you have to have played for months. You're hooked! The game still has interest, because there's still more you can do, you just can't bring yourself to do it. You can complain about unfairness, but you have no ammunition to use against the simple counter-argument: no pain, no gain.

Players complain because they care about the game. They want it to change, to improve. Some changes are good in the short term, bad in the long; some are bad in the short term, good in the long; some changes are universally good, some universally bad; some make little difference but hey, often players like change just for the sake of it.

Major counter-intuitive fact: most players who complain don't actually leave. If they try, they're back within two weeks. The players who really leave are the ones drift away without a murmur, not the ones who orchestrate public demonstrations. Player power may elicit knee-jerk reactions from administrators, but players aren't game designers. They may think they know what's best, but it's for the designers to decide what is best. Some degree of PKing is good: although players may rail against the very concept of it, unless they do actually get PKed on a regular basis they're not going to leave in droves.

So it's safe to implement PD in badland areas that give higher rewards than available elsewhere in the game?

Sadly, it's not. The flaw in this argument is that that although current players may not leave a game that has PD or PKing, new players might not even start it. They could simply head for games where they know their character will be immortal, not realising until it's too late what that means for the game's long-term prospects.

Verant recently stopped eBay from selling characters from EQ on their service, but gave no public explanation as to why. Was it because they felt the practice impinged on their intellectual property? Because they were worried their players were being defrauded? Because they wanted to sell characters themselves? No. It was because players complained bitterly that it gave an unfair advantage to a select few. Your level 40 character doesn't seem quite so impressive if anyone with dollars to spare can buy a level 50 one off the shelf. In a game with PD it wouldn't do the buyer any good, because without the skill to run the character they'd lose it the first time they put it at risk. In a game in which all that happens is you drop a bit of kit and maybe lose a few points, who cares?

Unfortunately, the players don't see it that way until it's too late. Their attitude to PD is akin to the general public's attitude to public transport: if only everyone else did it, my life would be so much better. There's not a lot you can do in the face of that!

So the designers of massively multiplayer games have a dilemma. Should they have no PD (and pay for it in long-term churn), have caps on functionally immortal characters (and attract fewer newbies) or have optional PD (and cross their fingers)?

Over the next few years, we'll find out what they decide...