For instance, I have a Warrior Mage. He has a charge attack, acid rain attack, and a projectile ice attack. If the system states that usage of a skill will increase its power and abiliities, then the player be rewarded only for usage of that skill? I would believe that even if all the player did is use ice attack for 5 hours straight that the acid rain and charge attacks would get a slight boost as well. Or even the attributes related to those spells (ex: Intellect, mana, stamina, casting speed, attack distance, etc). So in short, use of one spell boost the stats (albeit slightly) of other spells or should use the of one spell only boost that one spell?

In most games other skills rise too. A mage of level 50 will kill some level 1 monster with one sword attack, although he only used magic. Still, he won't use melee/weapon attacks on monsters of his level.

Now I don't know whether you're using a skill or level based system or a hybrid approach? I guess it all comes down to:
Why should a player want to use melee AND magic? Is there a disadvantage? For example some monster may be completely immune to magic and he will require to have trained his melee skills. If he can't get past that monster because these skills are still at level 1 although he is level 50 and he never needed them on a regular basis, that would be a design error, in my opinion.

I am of the opinion that all training should affect all statistics. In real life, it is fairly hard to train dexterity without training strength, and almost impossible to train a dexterity-based skill (i.e. the violin) without also gaining a benefit in dexterity.

Using a specific skill should raise the potency of other skills, but only if they are somewhat related.

You could create a dependency on skills for abilities. For example, you have the following general skills :
- Magery
- Evocation
- Conjuration
- Ice Spells
- Acid Spells
etc.

And power specific skills :
- Ice Bolt
- Acid Rain
- Blizzard

Then for each skill, assign their dependency on each skill. Ice Bolt could depend 10% on Magery, 15% on Evocation, 15% on Ice Spells and 60% on the Ice Bolt skill itself. When using Ice Bolt, it raises all these skills in equal proportion. Blizzard could use Magery, Conjuration and Ice Spells. Acid Rain would use Magery, Conjuration and Acid Spells. That way, casting Ice Bolt raises the potency of Blizzard because Magery and Ice Spells are tied skills. Similarly, casting Acid Rain raises the potency of Blizzard because they share Magery and Conjuration, but Ice Bolt gets raised less because it only has Magery in common. Of course, using the skill itself is still the best way to raise its potency, but you do not have to start at level 1 if you never used the skill before.

My vote is also that it should boost more than just the one spell.

In terms of real life, think of any skill and chances are, performing that activity affects related skills, as well as underlying abilities. For a simplified example, programming in C++ also improves your ability to program in other languages, and in addition increases your problem solving ability, typing skills, reading comprehension, etc.

So you might have a couple layers of attributes affected by casting that projectile ice attack. Using it might increase the character's ability to use that skill specifically, such as increasing casting speed for that spell. It might also increase your ability to perform any magic and any projectile attack, as well, by altering the character's basic attributes like intellect, mana, and accuracy/attack distance.

You could add a layer of complexity so it improves the character's ability to perform other ice-based magic more than it improves fire-based magic, depending how deep your magic system is. In my real life example, programming in C++ improves your ability to program in other imperative languages like Java and Python, but only slightly improves your ability to use declarative languages like OCaml and Erlang. Note to pedants: I know, just go with it.

An action should boost a bunch of things: that specific action, more general actions, and every action overall.

Example:
a mage casts fireball. Fireball is a fire evocation spell
The mage gets bonuses as follows: +10 fireball, +5 fire magic, +5 evocation magic, +2 magic in general, and +1 worldliness

worldliness is best described as just that: a bonus that factors into everything, representing your general experiences. So as you use abilities, you get better mostly at those, but also in general.

I would say that in general, related skills should get a boost, to the degree that they are related. In your original example, using the ice attack (or indeed any magical ability) for five hours straight is a strenuous endurance exercise, and all magical skills will surely benefit from it; this would probably be expressed by increasing the players mana store or something. But the more specific parts of the other skills, say the exact skill of turning magical energy into acid rain or other weather phenomena, won't be affected. So yeah, boost the underlying attributes.

I really prefer the traditional levelling systems where you just have one kind of experience points to improve your character. All the skill based levelling systems have trouble rewarding you for completing missions or other challenges rather than just using the skill a lot and I can never seem to get anywhere without dedicated grinding sessions to chop up some rats.

The trick is to decide how the skills should be related. Nobody ever got better at mathematics by doing a lot of pull-ups, but beefing up your German by reading German newspapers will inevitably provide a boost to your knowledge of current events.

In your example, I'd link it to attribute gain. The ice projectile wouldn't teach you anything about acid rain as a spell, but it might help you boost your arcana attribute or increase your mental focus, just like playing the trombone will train your breathing and musical abilities to better perform on a flute. You can make it as specific or general as you like, from the strength/agility/toughness that factor into everything from threshing to forging to swordfighting in Dwarf Fortress to a highly specialized hand strength/leg strength/visual acuity/cardiovascular endurance/equipment familiarity/pain tolerance set that might impact a rock climbing activity.

As a rule, the computer can do the math easily, but the burden is on the designer to balance the system and make it worthwhile.

I do hate it when I kill 500 slime monsters with a sword and that teaches me diplomacy. It's as if I wanted to lift my car over my head, so I practiced peeing my name in snow until I got strong enough.

Quote:

Original post by swiftcoder I am of the opinion that all training should affect all statistics. In real life, it is fairly hard to train dexterity without training strength, and almost impossible to train a dexterity-based skill (i.e. the violin) without also gaining a benefit in dexterity.

I think there is validity in this approach but I would prefer to keep the base statistics distinct, and to have them as unrelated aspects by definition.

You could build another layer of attributes on top of them, eg. Violin Playing might be 1/5 Strength and 4/5 Dexterity. Consider these the practical abilities that always require a mixture of the more abstract abilities.

If you do have layered attributes or skills I think you need to be careful to keep the 'tree' quite shallow and predictable, otherwise balance issues are likely to arise. If you can train up a super killing spell by virtue of having the benefit from many simple spells trickle up to your super spells, it is easy to imagine the gameplay being very hard to pace.

Even if you just have it so any one skill can affect a handful of others with no further propagation across the skill graph, you'll still find that some skills are better connected than others and therefore advance more quickly.