Use- and study-based learning

= How to improve without experience points = There are two ways to improve your character, learn-by-doing, and learn-by-study. It is critical (especially for learn-by-study) that the player "spend" some resource to improve his character. Often this resource is "experience points" but this is not mandatory. Other resources might be Money or Time, or even "health" (in the context of "risking" ones' health in a fight to test ones' skills)

Learn-by-doing
This is also known as "use-based level ups" in the cRPG industry, however, it is typically implemented in a miserable grinding way. These rules attempt to divorce "skilling up" from general quests while still preventing "skill grinding". Generally, the idea is that as a _skill_ (not a stat, trait or other property) is used in a "mission-critical" setting, there is some chance that the character learns something new. What is required is that the skill-based task have both a significant chance of success and a significant chance of failure. This is true independent of the skill level of the character. It implies the following:
 * The character must have at least a "default" level of the skill being used (for example "nuclear physics" has no default, if you did not study the skill or create your character with it, you cannot learn-by-doing it).  However, you must have enough skill such that the task is possible for you to succeed at.
 * You must not have so high a level of skill (relative to the task) such that it is trivial for you. This should make it de facto harder to increase a skill by doing it the higher your skill level if harder tasks are more rare in the game universe.
 * It should be possible, with clever default definitions, to learn-by-doing based on a related skill.

Now, there are a lot of constants here left undefined for tuning purposes:
 * How to define a "mission critical" task. As an example, combat skill improvement chances could be restricted to "battles" (initiations of "combat mode"?) that the PC and friends actually have some chance of "losing" (not necessarily being killed, but also seriously wounded, and/or forced to flee).
 * What constitutes a "success" (Read: Zenbitz:Thoughts on Task Resolution for a discussion of partial successes and failures)
 * What are the minimal and maximum chances of success.
 * Is "trying" the task enough - or does the character actually have to succeed?  It is often said that we (humans) learn much more from failure than from success.  If success is necessary to improve the skill  - then the higher the skill level relative to the task, the 'easier' it will be to improve, which seems counter intuitive.  So we would have to correct for that.
 * How often does a skill improved when "used" -- and by how much. Both of these should depend on at least the relative difficulty of the task, and probably the absolute character skill level as well.

The above, of course, must be turned into a detailed formula for the poor programmers.

Lastly, no matter how the detailed formula is crafted - it will probably still be exploitable by a determined player. To prevent this, we need to put a "cap" on the number of times a skill can be "used" (for the purposes of learning from it). A simple way to do this is that no skill can be "improved" (including failing a possible "improve" check!) by learn-by-doing more than once/day - or even once/week. This gives designers two ways to "control" player advancement by doing. Both are necessary, but I am not sure if they are sufficient.
 * 1) By restricting the difficulty of tasks available to the player
 * 2) By forcing the player to spend "time".

The idea behind learn-by-doing is that the character 'naturally improves' his skills by going about his business rescuing princesses or razing towns or finding waterchips. It is not meant to be a way for a Player to grind his skill up in a conscious manner. For that, see "Learn-by-study". Rules and restrictions must be added to prevent this system from being "gamed".

Learn-by-study
This is a type of learning initiated by the player to increase a skill, stat, or ability. It (at current draft) is the only way to increase a "primary stat" (which at this stage aren't even defined yet, but work with me here). It is something of a given that some stats and/or characteristics cannot' be improved at all (for example, height). Learning-by-study represents devoting a period of time to improving a skill (including some stats, etc. ) 'or' learning a new one. It is the only way a character can learn a new skill area.

The critical difference between Learn-by-study and Learn-by-doing is that there is no "task" involved in studying. The player can decide to study a skill at his leisure. Generally, a few things might be required:
 * A block of time (minimum 1 week)
 * A teacher - who has equal or greater skill than the student (and also has the time)
 * Learning materials, non consumable (i.e, books)
 * Learning materials, consumable (if the task requires them - classic example is live ammunition to learn how to shoot)

There might also be hindrances to studying - for example, if the character is wounded, sick or starving - or has no books. As in learning-by-doing, it is easier to improve a "low" level of skill than a "high" one. This is implemented by setting the rate of learning based on the students skill level.

Teachers
We may wish to allow characters to "self study" i.e., without a teacher. This opens up numerous obvious loopholes for abuse, which then must be closed. However, it may be ultimately too restrictive. We also may wish to restrict teaching to NPCs not in the "party" (i.e, not under the Player's partial control) for similar reasons. In any case, studying without a teacher will result in a much reduced learning rate. Also important are the relative skill of the teacher and student. The student cannot end up with a higher skill level than the teacher. This is a simplification of reality- of course, many students turn out better than their teachers, and many teachers learn from their students, but this is not really worth modeling in the game (in my opinion). Perhaps it would make sense that a PC can study a new skill without a teacher (as long as he has "books", but could only get a "basic level" of the skill. He could never "improve" past an initial score without a teacher (again - this is for game balance not simulation value).  NPC teachers will probably need to get paid.

Time
I feel it is critical that the Player spend some game time to study a skill. And that this time must have a cost to the player, in terms of his ability to do other things. This is why I say that we "need" Time to be a critical resource to have non experience-point based character improvement. Note that spending a week studying does not mean spending 24x7 (168 hours straight studying). We would assume 8 hours of sleep and 8 hours of "living". Luckily that last 8 hours allows us to worry about food and heat and other 'housekeeping' functions. We may want to restrict a place of study to somewhere food is readily available. (i.e, not out on the ice flow), or else have the player consume his food stocks.