Imaginary Realities Imaginary Realities About Search Glossary
What's new? Index :)
Select issue:
Join a discussion Resources
     

Setting the mind to work
- David Mallard
Advancement Revisited
- Scatter ///\oo/\\\
Why Rent?
- Ilya
So You Want to Start a mud
- Robert M. Zigweid
Tasks: a new variety of quest
- Hugo Jonker

Letters to the editor

Enter your email to be informed when this site is updated.


Comment on articles

Letter 1
Contact editors

   

Advancement Revisited

by Scatter ///\oo/\\\

Last month, the magazine featured an article called "Advancement the Old Way" on the topic of skills and levels. That article picked up on a number of problems with the more traditional treatment of skill levels and pointed out a number of good and desirable things that can be gained by moving away from it. However, the suggested mechanism doesn't quite seem to achieve them all, so here are a couple more ideas for improvement.

It is certainly true that virtually all muds are implemented so that the higher a player's abilities become, the more difficult it is to improve them further. There are two main reasons for this.

Firstly, most mud implementors do not want players to become all-powerful. If abilities scale linearly with effort spent on them, players will inevitably become very powerful very quickly. If there is no upper limit to development then the result is that high level players become gods - nothing can touch them and they are able to slay everything the mud has to offer. In most muds such players are utterly out of context. In the traditional medieval fantasy settings, players are encouraged to become heroes and do great deeds. Allowing them the sort of powers more usually expected of gods can seriously break belief in the world - especially when the world provides its own pantheon of gods. These players also tend to wreak havoc on any kind of game balance, simply by being able to do anything they want.

Chief Sitting Bull, sitting

Old man take a look a my life, I'm a lot like you were.

Of course it could be the aim of the game to become god-like in this way, and built into the theme of the mud that it is a common thing to happen. If so, the game must provide worthwhile things for these gods to do.

This is the second problem. If abilities are open-ended and just keep increasing indefinitely, a race rapidly develops between the players and the world-builders. As player abilities increase and increase, the builders must keep building higher and higher level areas in order to challenge them. Without such challenges there is a risk that the goal of the game effectively becomes boredom. You play until your skills are good enough that nothing can beat you and you then get bored. Then you either restart, or go play somewhere else. Either way, boredom is hardly a good reward for 'winning' the game.

So there must be a way of slowing down the development of abilities and making sure that they never pass a carefully planned level. The most obvious way of doing so is to apply a law of diminishing returns. This is how learning and training works in real life and so it is easily understood and accepted in the mud world too. For example, you can gain conversational skills in a foreign language fairly quickly, but it takes a heck of a lot more work to be capable of good poetry or emotive speeches in that language. A law of diminishing returns also effectively provides an upper limit - at some point a plateau is reached where improving abilities further means that an impossible amount of effort is required. This prevents players becoming omnipotent, allowing the mud's designers to be sure that they can challenge even the best of heroes.

The point has been made that on many muds improving abilities quickly leads to an "always-or-never" situation, where a player's attacks always succeed and defenses never fail. This should only become true when the player is facing an opponent of radically lower abilities. If a human steps on an ant, their attack always succeeds - the ant can't defend itself. On the other hand, should an attack by the same human on a four-headed dragon get the same result? Only if the human's abilities are massively superior to the dragon's - or if the dragon has a weakness to exploit. If a mud reaches the point where a player's abilities are massively superior to everything else in the game, the developers have lost the race. There are no areas of high enough level to challenge this player's abilities. Or perhaps the mud has made some balancing mistakes and allowed player abilities to grow too high.

In addition, a method of comparing skills that leads to an "always-or-never" situation can be considered a flawed design in itself. A design that prevents this situation will be presented later in this article.

Last month's article suggested that rather than making skill levels progressively more difficult to gain, some of these problems could be addressed by separating measurement of ability from the skill level. The example given was a technique like this:

Skill:    1  2  3  4  5  6  7  8  9  10 11 12 13 14 15 16...
Ability:  1  2  2  3  3  3  4  4  4  4  5  5  5  5  5  6 ...

Rather than making it progressively more difficult to increase your skill, the suggestion was to make it always cost the same to increase your skill but make that increase mean less and less in terms of ability.

Unfortunately all this system does is to hide ability measurement from the player. Fundamentally it is the same system - in practice it still takes longer and longer to increase an ability. However, it has an additional problem - when players look at their skills under this system they don't see their abilities. All they have is a measurement of how much time and effort they've put into the skills. Without knowing or working out the formula you have used, the skill number is no longer a real measurement of ability.

It wouldn't take long for players to see through the system, and deduce where the real ability milestones are. It may be that they can gain level after level for the same effort, but the levels themselves are no longer the important points - especially with the example formula given above. It doesn't mean anything at all to go from level 11 to level 12 - ability is still rated at 5. It is necessary to get to all the way to 16 before any possible improvement could be noticed. This removes the reward for the time put in gaining each level.

It was suggested that this system would change how people view advancement and this is true, but probably not in the intended way. The kind of player who wants to get as powerful as possible as soon as possible will still want to do so. The change to the system means skill levels themselves rapidly become so divorced from ability that the only way players can judge their ability is through comparing themselves to the best and worst around them. (This could be considered quite realistic, but if this is the desired result why bother with showing skill levels to the player at all?)

Also, under this system the skill level number becomes meaningless at high levels because it takes a large number of levels before the actual ability improves at all. Therefore comparisons between players of similar skill levels are pointless - their abilities will be exactly the same. Players must compare with skill levels further and further from their own before being able to detect differences in ability.

Indeed, rather than improving the advancement system, this concealment of real ability seems instead to lead to more problems.

A different comparison

What is the real point that the separation of ability from skill was trying to address? The main aspect of skills that it seemed to be addressing was the way the mud compares different skills to determine the outcome of an event. The typical example is a combat event where there is an offensive skill which must overcome a different defensive skill.

As pointed out in the previous article, skill comparisons are often carried out by finding the difference between two skills and using that to determine whether something succeeds or failed. This leads to the situation where a skill of 10 versus a skill of 8 is twice as effective as a skill of 9 versus a skill of 8 - because the difference is 2 rather than 1.

This makes the difference in ability between successive skill levels nonsensical. A skill of 20 versus that skill of 8 is twelve times more effective than the skill of 9, whereas a skill of 14 is six times more effective. Therefore a skill of 20 is twice as good as a skill of 14. Increase the defending skill to 10, and suddenly a skill of 20 is twice as good as 15 instead.

The change to separate ability from skill appears to help with these comparisons because it means that, for example, a skill of 16 is needed before the corresponding ability (6 in the example system above) becomes twice as good as a skill of 9 (ability 4). However this change is really just obscuring the problem. The same problems occur eventually but over a much larger skill range so that they don't become apparent for longer.

The problem doesn't stem from the way skills are being represented, rather it is inherent in the way skills are being compared. The problem stems directly from that initial comparison - the subtraction to find the difference between the levels. A better comparison needs to work without subtractions or additions, instead using a probabilistic method incorporating an element of random chance. For example, a quick and easy method is:

  • Take a random number between zero and the attacker's skill.

  • Take another random number between zero and the defender's skill.

  • If the former is higher, the attack succeeds, otherwise the attack fails.

Back to that example of the attacker with the skill of 10 attacking a defender of skill 8. The attacker has a possible result of 0 to 10, and the defender of 0 to 8.

This means the defender is capable of defending 82% of possible attacks from this attacker - and on average is likely to succeed in doing this 50% of the time. The other 18% of the attacks (when the attacker scores a 9 or 10) cannot be defended against- they are beyond the defender's skill. So the final toll is that the defender has a 41% chance of successfully defending any attack from this attacker.

Double the attacker's skill to 20, and the result is that the defender now has only a 20.5% percent chance of successfully defending. This makes the attack twice as likely to succeed - a logical result of having twice the skill. Compare this to the original system where doubling the attacking skill from 10 to 20 would make the attack six times more likely to succeed!

This kind of skill comparison has some very useful features. For example, as shown above, doubling someone's skill always makes them twice as likely to succeed and this is both logical and intuitive. This rule holds regardless of the skill levels involved, keeping the skill level a direct indicator of ability.

Another advantage is that as skills get higher and higher, the differences between them mean less and less. As shown above, the difference between an attacking skill of 10 and a defending skill of 8 means the attacker has a 59% chance of success. But with the same comparison system, a skill of 100 vs a skill of 98 (still a difference of 2) has only a 51% percent chance. (Identical skills would of course make it 50%).

To take a more dramatic example, attacking skill 10 versus defending skill 1 gives a 91% success rate whereas 200 versus 191 (still a difference of 9) reduces the success rate to 52%.

With this comparison mechanism it is assured that things never come to an 'always succeeds' situation - even if the attacker's skill is a thousand and the defender's is a mere two, there is still a chance that the attacker's random score will be zero or one, allowing the defender success with a score of two. An obvious way to extend this system is to add things like 'critical fumbles' when the attacker scores zero, and 'critical hits' when the attacker scores the maximum possible.

Wisdom and practice

Another aspect of skills and advancement that last month's article examined was getting logical results when comparing both wildly different skill levels and very similar skill levels. The example was given of two martial-arts masters. If one master was to learn a new trick it could significantly change the outcome of a fight between them, despite their very similar skill levels. On the other extreme, knowing the new trick would make no difference to the outcome of a fight between the master and a novice.

The second situation is already covered by the 'probabilistic' comparison suggested above. The large skill difference between the master and the novice means that so many of the master's attacks are already beyond the novice's skill to defend that an extra new trick makes virtually no difference.

The first situation, that of masters of equivalent ability where one learns a new technique, can be better addressed by separating knowledge from ability. Your skill level in, say, 'combat.martial-arts.karate' reflects your ability to perform the karate moves, how much you have practiced and trained and so forth. Separately from this skill level, you also have knowledge of tactics, strategies and moves. Under a system like this, you can get better and better at what you already know without learning new tricks - improving your ability skill. You can also learn new tricks, moves, etc., by improving your knowledge.

Knowledge could be implemented as an additional skill, for example 'knowledge.martial-arts.karate', or by having the moves as extra commands that the player can learn, or other ways. The key is that this extra knowledge can be learned without necessarily having the ability to do use it - just as studying a book on karate would give a lot of knowledge of moves and tactics. In order to perform them usefully, it would be necessary to train and physically practice and gain the ability to use them.

Another good example of this knowledge/ability split is the mage. She can improve her spellcasting abilities and so cast the spells she knows more and more successfully. She can also learn new spells, without improving her casting ability. She may not be able to cast them successfully with her current ability. Learning the new spells will not make her better at spellcasting, and practising spellcasting will not give her knowledge of new spells.

There is another very valuable advantage of this knowledge/ability separation - the enhanced potential for character development and roleplay. A thief may never be able to learn to cast spells, not having the gift for magic. But perhaps that thief could spend time learning about spells and magic, learn to recognise spells being cast and, for example, know whether an enemy's casting sequence means she should prepare to duck a fireball, hold her breath against poisonous gas, don the Helmet of True Sight - or run like hell.

Skills in general

Most muds place great emphasis on skills and their development. This makes it imperative that skill implementation is done well. Ideally the skill system should address these points:

  • The skill structure should be easy for players to understand.

    For this, it needs to be structured in a way that is logical and intuitive. It should be clear why each skill is arranged as it is, with no blatant exceptions to the structure. If at all possible it should be comprehensive - in order to avoid having to 'bolt on' extras at a later stage.

  • It should be easy for players to judge their skills and their improvement in them.

    If skill advancement is a goal, then the system must provide ways for players to judge how they are progressing towards this goal. This includes making sure any numerical representation of skills is a good way to judge practical ability. Note that this does not mean it must be easy for a player to compare their own skills with that of another player or npc.

  • The skills should provide a believable system of learning things, practicing things and knowing things.

    It is an important point that the system is believable. In general the closer it can model real-life situations the better, though there are dangers to beware in cluttering the model with unnecessary details just to make it more realistic.

Of course there are many other aspects that should also be addressed, but a skill system that addresses these three points is well on the way to being a good design. The suggestions given above on managing skill comparisons and knowledge are just a couple of possible ways these points can be addressed.