Use and Abuse.

posted: July 19th, 2006

Raph Koster wrote an article talking about use-based skill systems in MMO’s and about how they worked out in Ultima Online and StarWars Galaxies. Just another example of synchronicity since I planned on following up A Distinct Lack of Class with a post about skill systems.

As I stated in my earlier post I’m really not a fan of class based systems. I find them artificial and I feel that too often they lead to odd methods of skill progression (beating up orcs to increase your ability to pick locks, as an example). There are some good points to class based systems, as I stated earlier, but I would rather try to find some way to bring as many of those advantages over to a skill based system than remain in a class based system solely for those advantages.

As I see it there are three main ways of handling advancement in a skill based system. The first is to have players earn experience that they spend as they see fit to advance their skills. A fair amount of tabletop RPG’s use this method since it is simple and easy to keep a track of. The biggest problem I have with this method is that it still leads to odd methods of skill progression (beat up orcs to earn the experience points you use to increase your lock-picking skills). This can be avoided somewhat by breaking experience into categories instead of a general pool, so the experience you gain from beating up an orc can be used to increase combat skills but not crafting skills. With a large number of pools this can quickly become too cumbersome for a tabletop game but with servers keeping track of a character’s stats it isn’t as much of an issue.

The second method I can see, and one I wouldn’t have even thought of until I tried Eve Online, is time based. You select a skill to ‘practice’ and after a fixed amount of time it improves. There’s no XP to earn, just time to spend. It’s an interesting approach, but I’m not 100% sold on it since a player who is highly active progresses at the same rate as someone who only logs in when they need to choose the next skill to train (an oversimplification, but you get the idea). Still, since the skill gain occurs in real time whether you are online or not it definitely reduces any feeling that you are obligated to grind in order to increase skills.

The last method I can see is, of course, the use-based method. Skills are increased as you use them. This, unfortunately, tends to lead to grinding, grinding leads to macroing, and macroing leads to the dark side.

Enter the law of diminishing returns. This isn’t something I’ve talked about yet but it’s something that I’ve started to consider as possibly being a very important item in design of an MMO. Yeah, ‘started to consider as possibly being’ is a pretty weak statement, but I’m certain of the idea yet, so I’m not about to trumpet it as a golden rule. Given that it has the ability to prolong enjoyment of a game by removing the importance of grinding, as well as levelling the field a bit between casual and hardcore gamers I can see a lot of potential in it, but I still need some time before I convince myself.

The law of diminishing returns is something that occurs in real life all the time and what it basically means is that as more energy is put into something the lower the return on investment is. This doesn’t mean that the return is less than it was before, simply that the return doesn’t scale linearly with the energy expended.

This is pretty much a universal constant, at least once certain minimums of expenditure are met, and the clearest example I can think of is a car. In order for a car to travel twice as fast it has to spend more than twice as much energy. This is because as speed doubles wind resistance squares. This is just one example of course. The world is full of others.

Learning, likewise, suffers from the law of diminishing returns. The longer a person spends trying to learn something without a break the more fatigued they become and the slower they will learn. A person who spends fourteen hours in a single day trying to perfect a skill will learn less than a person who spends two hours a day for a week trying to learn the same skill.

How do we translate that into a game system? Well, for starters I would consider breaking skill levels into ‘long term’ and ‘short term’ levels, much as human being have long term and short term memory. As a skill is used any gains from it go immediately into the ‘short term’ pool. Points drain out from this short term pool and into the long term pool at a rate based on how high the short term pool is. A short term pool that is higher will transfer faster than a short term pool that is lower

The catch is that the transfer rate isn’t linear but is an inverse geometric function. A pool that is holding a hundred points might transfer points twice as fast as one holding twenty five points, three times faster than a pool holding eleven points, etc. In the mean time the short term pool will additionally ‘bleed’ points faster depending upon how large it is. The points are simply lost and not converted to the long term pool. This rate however is linear, so a pool with a hundred points would drain away at a rate four times faster than a pool with twenty five points and nine times faster than a pool with eleven points. Thus, to continue progressing twice as fast as another player a person would have to spend four times as much energy.

Most of this would be kept hidden from the players for the sake of simplicity, but of course as Raph Koster points out, any system you create will eventually be analyzed and decoded by the player base. To help prevent or at least mitigate ‘min-maxing’ on the part of players all individual pools could be combined together for the purposes of transference and loss. A character with two pools at twenty five and one pool at fifty would transfer just as many points as a character with a single pool at one hundred, though the points would be split with a quarter of them going to one skill, a quarter to another, and the remaining half going to the third. Likewise the character would ‘bleed’ points at the same rate, again with the loss being divided between the three pools.

Obviously such a system would take some balancing to get the math right. Four times the work for twice the gain might be too much work or too little work, but the concept remains the same. As more energy is spent the character progresses faster, but the rate of increase is slowed so that the law of diminishing returns applies.


Excellent use of the “law of diminishing returns-” one of the principles I’ve been harping about on many a game forum for years, it seems.

The law can be applied elsewhere in the skill ceiling: like developing developing a soft skill cap.” In many games, a skill reaches a hard cap, after which you just can’t get any better. Some have a cap in the mechaincs where, even IF you can go higher than a value, you get no bonus for doing so. CoH allowed the “flight” power to take up to 6 enhancements, but after the 4th (I think) speed boost, you’d reached a cap… even as the game let you continue adding more boosts without any notice. For a while, SWG calculated defense based on a “cap” of something like 125 points, so a template stacker hoping to reach 180 points saw no benefit from the extra 55 (it’s been a while, so these are guesstimates)

In a straight percentage advancement, a person with 75 points invested in a skill would be 25% less effectie than a person with 100 points in a skill(the theoretical ‘cap’ in this example). With diminishing returns, the “curve” might make the 100 points closer to 10-15% as effective. It might take 150 points to get to the same strength as the 100-pointer in the more linear approach.

If the curve is too steep (like (to some) CoH’s effort at “enhancement diversification) then people will simply treat the arc of the curve as a lower “hard cap” but a proper grade would create a min/max’ers nightmare- a system where there’s open debate on where the best cutoff point should be!

This can be further blurred by assigning penalties in different ways:
Let’s assume a curve based on this (made up right here, not optimal)

Success Rate: Point Cost
50% : 50 points
75% : 75 points
80% : 85 points
85% : 100 points
90% : 120 points
95% : 145 points

Now, in this system, a person with 120 points and 145 points invested only have a 5% difference in success rate. If I apply a penalty of -40% accuracy for darkness, they’re equally affected and still only have a 5% difference between them. If I instead offer it as -40 points, the percentages drop to about 77.5% to a little over 85%. The bigger investment starts getting a bit of an advantage for the added investment.

Ideally, some negative modifiers affect percentages, some affect points, some affect both. The result becomes that the “ceiling” will be different depending on many factors. As a lower-skilled player, I may still be your near-equal on a clear day with optimal conditions. Add rain, add distraction from pain, and add darkness, and your mastery will allow you to perform at a much higher level than me.

Comment by Chas — July 20, 2006

I think one of the problems that is faced by utilizing the law of diminishing returns is in making sure that it is correctly applied. As an example a lot of level based games use the law of diminishing returns in their experience charts. However they then go ahead and increase the experience values for creatures at a commensurate level so that while the amount of experience required to gain a level increases the energy expended does not.

A diminishing returns curve could be used to calculate the XP awards for creatures depending upon the difference in level between character and mob. This would give a good bonus for players attacking creatures a few levels above them but would slow the award as the character became more and more out of their depth (reflecting the fact that there’s probably some jiggery-pokery going on either with twink loot or some kind of trick involving higher level characters helping that allows the character to defeat creatures so far above him) and would ramp down the XP on creatures below the characters level fairly rapidly. With that in place the XP requirement chart could be constructed to generate a proper progression over time.

Comment by Evan — July 25, 2006

Good comments, Evan.

I think many of us older-timers might need to rethink some of our assumptions regarding the leveling process, as the broader gaming market seem less and less likely to accept the vastly longer intervals between level advancement. When CoH came out, it was considered to be rapid-leveling devoid of “grind.” Now, that word’s used regularly as the WoWer’s have invaded and seem to expect a much faster pace of advancement.

Perhaps the problem has become that we’ve made advancement to be one of the few tangible rewards, so when the “reward” slows, we lose some of the game ‘fun.’ By making diminishing-return affect the scale of the advancement (less of a boost each skill level increase) rather than increasing the time spent waiting for that ding, we might give them that much needed reward.

Sure, they might notice that the reward is nearly negligible, but to date, when they “get better” their foes ‘get tougher” to largely the same relative magnitude as before the leveling, and few seem to have really noticed THAT.

Comment by Chas — July 27, 2006

Leave a comment