Posted by: Chessack | Saturday, April 19, 2008

Alternatives to level systems

On the Champions Online forums, I have had a debate with some of the posters about free-form vs. level-based gaming. They keep saying that free-form game design cannot be made workable in an MMORPG, and I of course, think it can be. Here is a post from that thread that I thought I’d save here, because one never knows how long a forum post is going to be kept up.

The objection to a point-based, free-form system is that it is free-form. The basic argument, as I see it, is that with a free-form system it is possible to gimp yourself — to create a toon that can’t do enough damage to an enemy to actually defeat him. We can imagine a character who can only generate 5 damage per second, but his enemy regens at 6 per second, meaning you can’t win the fight.

One answer to this objection is simply, “So what?” Yes, it is an acknowledged weakness of free-form systems that you can severely gimp yourself. People like myself who prefer, and even advocate for, such systems, believe that the benefits (allowing one to have precisely the character that one wishes) outweigh this cost. Such a belief is a matter of opnion. You don’ t have to hold that opinion, but arguing against it based on some sort of supposed “facts” is not viable, precisely because it is an opinion. You object, Mr. O, that a free-form point-based game “cannot work”, but your definition of it NOT working equals my definition of it working! A game that is free-form enough that I can completely gimp myself is just what I want — because it means that, gimped or not, my character is MINE, 100%, and not just someone else’s. Because, in the end, that is what a level-based system like WOW or Vanguard or EQ is. The designers decide what skills a Paladin, say, will get, and at what level, and by leveling and “buying” the level-appropriate skills, what I am doing amounts to little more than checking off the boxes of a pre-generated sheet. That may do it for you, but it doesn’t make it even halfway up the flagpole for me to salute.

Let’s remember after all, that class/level systems are not without weaknesses. It is an acknowledged weakness of those systems that the player is locked into pre-defined design templates and has less freedom to create his own character setup. Some people, like you, consider the easy balance that results from such rigidity to be a benefit that outweighs the cost of the system being rigid. You are certainly entitled to your opinion, but I do not share it.

Another answer to this objection, however, is that one can easily mitigate against the possible gimpage by providing a wide array of “pre-generated” characters that are set up specifically NOT to be gimped. So, you could easily allow a player to pick a “Ninja” type character, and have all the stats and powers pre-allocated for him, and allow him to just “accept recommended” powers at every single level. But you can also allow a person to say “let me customize” and let him pick EVERY power. Now no one can complain, because people who are (a) unsure of the system, or (b) too lazy to figure it out, or (c) concerned about being gimped, can just take the recommended builds and be completely assured of a character and an advancement scheme that is tailor made to be workable within the system. So, all potential gimpage is curbed by providing pre-made character templates for people to use, and even slightly tweak.

But there is an even better answer to your objection, Mr. O, because you are resting the entire argument on the assumption that spawns and mission generation are going to remain the same in CO as they always have been. This assumption may not be far off the truth, but I hope that it is, because the way missions are generated right now in games like COH is very coarse. You might have a bit of a point if spawns are generated the old way, because they are taking a pre-set NPC based on level, with HP and stats based on level, and just spawning it for you.

But, imagine a more dynamic situation. Imagine, instead, if when you enter a mission, as it loads you into the instance, the computer scans your character and figures out just what stats you have, and dynamically generates a set of enemies who have stats tailor made to give you a challenge. So, if you have chosen to have only 3D6 of attack, that’s no big deal — it can generate enemies with 3 PD/ED to compensate. If you happen to have put all your points into attack and have 20D6, you won’t wipe the floor with enemies… they will have some damage reduction and higher PD/ED to compensate. In fact in Champions, the PnP game, it is actually very easy (I have done it) to create some simple algorithms that can generate very balanced, properly challenging NPCs for any given hero or set of heroes. You can just do things like, “NPCs have 1D6 of normal damage per every 3 PD/ED of the hero entering the zone.” If you do this for attack damage, OCV/DCV/ECV, defenses, special powers (e.g., flash defense) and so on, you can easily balance bunch of enemies for an entire group.

As I say, this is not mere speculation — I have done it. I have created “random NPC generation tables” for Champions (in my youth, when I thought Champions needed such things — I do not, now, at least in PnP). It really didn’t take me very long — a few weeks over the summer one year provided me with just the tables and charts I needed to generate whole groups of agents appropriately challenging for any mix of superheroes you might care to present me with. I did this because the players were changing a lot during the summer and we had no fixed groups. As with an MMORPG, I could not predict ahead of time who would show up. So, people would arrive, and I would use my tables and algorithms to, by hand, generate some agents appropriate to the battles (obviously, mega-villains were made ahead of time). For people who might remember Champions II, the basic idea here was the Turtle Armor agents, but I took it to the next level.

Now, it strikes me that if a high school student on summer vacation can come up with a workable “dynamic power, skill, and stat generation system” for NPCs using pencils, paper, and dice, and some rudimentary math skills, a team of highly paid professional computer programmers ought to be able to create such a system in a computer in the year 2008+.

Your whole argument rests on the basic premise that I can “gimp” myself because the game expects something like 10D6, and makes up villains with PD/ED based on 10D6, so if I have 5D6, I am hosed. But if the game is designed right, then if I only have 5D6, the enemies should have the right PD/ED to compensate, and the battle will still be a challenge, but not impossible. They could (and should!) also make a difficulty slider that lets players play with the generation, giving more XP to people who buff the difficulty level and less to those who lower it.

The key here is dynamically generated content rather than, as we have seen before, statically generated content that was made presuming a set of builds on the part of the player. Let’s see them stop assuming my character will have power A, skill B, and stat C, and start generating a proper challenge for the powers, skills, and stats, that I *do* have.

Responses

I agree that it is possible to have a non-level-based character advancement system in an MMO.

But I think you are *seriously* oversimplifying the issues with dynamically-generated content.

There are certainly some systems where you can easily generate monsters of an appropriate difficulty for a PC of any given stats–but if you want the game to contain any tactical depth at all, you can’t use one of those systems. Because if the system has tactical depth, then you can’t easily predict how well a given character is going to perform, because it will be dependent on the player using insightful tactics–and anything that a generalized computer algorithm can predict you’re going to do is, by the fact alone, probably not very insightful. There’s a reason we play a game instead of flipping a coin, and watering down the system to make dynamic content generation easier tends to undermine that purpose.

The fact that a high school student made some tables that seemed to work does not convince me otherwise. Firstly, I know nothing about the system you were using. Secondly, I would be surprised if the generation process was 100% formalized–I suspect you chose some initial parameters or modified some results based on your instinctive “feel” of the situation. Third, you have a very small sample size–there’s no guarantee it would have continued to work as well if it were applied to thousands of players over a period of years, especially if those players have an opportunity to reverse-engineer your system and try to break it.

But even if we COULD algorithmically generate 100% perfectly balanced challenges for the player, there are at least 3 important problems remaining:

1) Groups. Not much point in an MMO if you’re playing solo. If you’re not playing solo, then even if the overall difficulty of the battle is appropriate to the overall power of your group, some members of your group may still be much more powerful than others, and thus have a greater ability to affect the game. Players will still feel frustrated and useless if their allies are killing ten times as many monsters as them, even if the overall difficulty is perfect.

2) We want monsters to be interesting, not just level-appropriate. In (well-designed) games with static content, not all the monsters of a given level are identical; they have different advantages and disadvantages, abilities and tactics, and so forth. Those differences are important to keep the game interesting. But the tank, the glass cannon, the crowd controller, etc. all need to change in completely different ways to adjust their power levels–setting the armor of all monsters to a single value just isn’t going to cut it. So at a minimum, we need static, manually-created “classes” of monsters, each one of which will probably need a completely different set of tables and algorithms for scaling to player power.

3) The entire point of having a free-form character customization system is so that you can choose the strengths and weaknesses of your personal character. If every challenge your character faces has its difficulty finely tuned to your character’s ability in that category, then the gameplay becomes exactly the same, no matter what your character is. You double your damage and the monsters double their health–you observe no change. You might as well let the players write anything they want on a character sheet, and then completely ignore it for purposes of playing the game–it would be an easier way to achieve the same result.

Level systems have some drawbacks, but they remain an elegant solution to a number of key problems in game design. There will always be some games that try something differently, and I’m in favor of offering the player more choices as regards character customization, but I think level-based systems will be the industry norm for a long time to come.

Thanks for the comments.

Since all of your objections have been thoroughly canvassed on the Champions Online forums, I will refer you to that thread (linked at the top of my post) rather than trying to do a point by point discussion here, particularly since there’s no good way to do multi-part quoting with the wordpress comment software.

C

Leave a response

Your response:

Categories