Action Points?

[Anvilfolk]

Well, you'd surely have a scale, detailing which adjectives are better than the others.

My point was that the focus changes. The idea is precisely that you don't have an exact idea about where you are in terms of skill, just a general, relative idea. This is different than a progress bar, which is still a lot more exact than 6 adjectives to cover 100 different values. Then the focus isn't about getting sword X above sword Y, because X is 10-16 damange and Y is only 8-12... it's whether, for example, which fits your character the most.

Since these changes and slight improvements aren't as obvious, more than likely the player won't notice them... he can't give it as much importance, then. He'll focus his attention elsewhere, changing the way the game is played.

I still think that for "classic" die-a-lot roguelikes, numbers are fine. All the problems you set for "adjectifying" (does this word even exist?) can be resolved by, say, reading a help page.

[Z]

Then, why show these 6 adjectives at all? The player should have a general idea about their skills without them. For example, the player should know (from their knowledge, or manual, or experiments) that elven archers are very good with bows. No use for writing "bow skill: very good".

[purestrain]

I think a general classification should always exist. I planned that the player should be given no real numbers; and only estiminations about what he or others can. It would help and doesn't require me e.g. to count hits and make an average of them to know my skill in swordfighting. I'm against number; since numbers can force a good game into a calculation sheet.

[stu]

the fudge system of using "very good" for skills instead of "75/100" its pretty good and works well.

underneath you still need to turn it into numbers.

[Gamer_2k4]

the fudge system of using "very good" for skills instead of "75/100" its pretty good and works well.

It's too general for me.  If I'm "very good" with both swords and hammers, but my actual stats (unseen) are 75 for swords and 80 for hammers, I'd like to know that.  And if the difference isn't significant enough, then why is it on a scale of 0-100 instead of, say, 1-10?

[Anvilfolk]

Again, it comes down to what kind of gameplay you want to emphasize.

You're looking at it simply from a roguelike's perspective. In that case, I do like knowing what I'm more proficient with so that I can use that weapon and have better chances of ... not dying:)

Imagine a MMORPG, where RolePlaying is actually meaningful in the context of the game (it is not simply an excuse for a hack'n'slash environment)... it makes sense to me that certain characters will want to re-inforce their "burliness" by using a warhammer instead of a sword (of realistic proportions). If both their stats are "very good", the player will, generally speaking, see no technical advantage in using either. He'll probably choose the warhammer for his stocky character - even though in your case, the sword would be deadlier.

Anyone seeing him wearing blacks, bearskin cloak, heavy (blackened) plate mail, chainlinks for "adornment" and such, along with his nice'n'heavy warhammer will think him a lot "cooler" than if he had a regular run-of-the-mill longsword at his belt.

This is just one of the limitless number of examples one could find where players start caring more about the actual roleplaying than about maximizing numbers. The underlying system can still be as technically complicated as you want, even equal to the roguelike's, but you've effectively changed the way the game is played - which was my original point.

[Xan]

A natural way is to have each being have a T value, which means, when will it be able to perform the next action (in global time). After each action, we find the creature which has the smallest value of T, set the global clock to T, let it do what it wants, and increase its T, which can depend on its speed and action. The system proposed by Gamer_2k4, and also used in ADOM at least, works similarly but is more complicated than my natural system, I guess in order to make it more effective, although it sometimes adds some weird behavior. (However, my system can be implemented effectively, in O(log n) per action, where n is the number of creatures.)

I had thought about implementing action scheduling this way in my roguelike, but I ran into a little problem with it, and I'd like to know what you think about it.  The problem I had was in determining the T value for the next action of a creature whose speed was being modified, and that speed modification depended on the elapsed time from the beginning of the effect.  E.g., the speed modification decreases over time, so it wouldn't be accurate to calculate the T value from the creature's current modified speed.  It could be done using calculus, but that assumes that the scheduler knows the formula for the modification, which in my case it cannot, since modifications are all self-contained units.

The same problem is found in action point systems as well, since there time is incremented in discrete ticks, but there is a good approximation there if the ticks are generally small enough in relation to the time normal actions take.

So I ended up going with an energy system similar to ADOM's, and although I'm not completely satisfied with it, it seems to work better than the alternatives that I've tried.

[Z]

I just base the T-value on the current speed, ignoring the fact that the speed can be changed in meantime. Can you show me a situation where this loss of accuracy is really a problem?

[Xan]

I just base the T-value on the current speed, ignoring the fact that the speed can be changed in meantime. Can you show me a situation where this loss of accuracy is really a problem?

Well, one example of an action that takes a long time is equipping some sorts of armor, which can take up to ten turns at normal speed to complete (so, somewhat similar to ADOM).  There is also a potion which dramatically increases your speed, but has a sharp dropoff of effect over time, so while you start out moving e.g., 10 times normal speed, before even the time of a normal turn has passed, you will be moving less than 3 times normal speed.  So if the time to equip the armor was calculated based on the current speed right after using the potion, it would only take the time of one normal turn to do it, rather than ~3 that it would using an approximation of the decreasing effect.  That's certainly a noticeable difference;  I'm not sure everyone would think that it's a "problem".

[Z]

There are also other reasons why you should not use this system directly for long actions (say, complexity 1000 where a normal action has complexity 100). You should be able to interrupt the action if something unexpected happens. The effects of wearing armor should not be immediate. Note that using a different time system does not solve these problems. You have to explicitly make long actions interruptible and have effect after completion only (or partial effect when in progress).

Replacing such action by 10 normal actions of complexity 100 (and interrupt checking after each of them) should make the inaccuracy very small again.