[Jihgfed Pumpkinhead]

Just out of curiosity, what do you all use for simulating random results? I mean, usually your "Super-Duper-Flying-Kick" won't do EXACTLY 1000 damage; it'll do somewhere around 1000, but randomizing adds a bit of spice.

Personally, I like simulating dice rolls; it's old-school, simple, and provides I think a nice distribution. For example, instead of "100", it would be something like 10 rolls of a 10-sided die. With "1000", it would be 10 rolls of a 100-sided die. Of course, the question is skewed considerably if efficiency is an issue, but I'm assuming it's not.

Essentially you're balancing fairness with novelty. So where, do you think, does the balance lie?

[ThousandKnives]

Personally, I don't like playing games where the results are too unreliable. The player can plan out more complicated strategies and you can EXPECT the player to plan more while still being fair if you make the results reasonably consistent.

In order to give some randomness to the numbers, I take the offensive value and make it +/- 1/8th of the value. It looks like this:
offense += (offense>>4) - rand()%(offense>>3);
Defensive power is constant, this it the only variation in damage. Also, if the total damage is 0 or less (and you're not supposed to be healing, heh), then I give a 50/50 chance of actually doing 1 damage (lucky dog), or doing 0 damage (like you should, you worthless swine).

I use simulated die, sort of, in calculating enemy hit points. It basically amounts to an enemy having a certain number of "hit dice", only the dice are loaded to only give results between 7-10. Once again, I don't like too much inconsistancy. I think it detracts from strategic planning. RPGs shouldn't be too much like gambling, by my reckoning.

Also, I can't imagine efficiency being an issue with most battle calculations (besides AI in a real-time game), as most of these calculations are only made once in a while, not every loop, and certainly not hundreds of times every loop.

[Modanung]

I like the idea of dice rolls, it very simple and adds a bit of randomness.
But Jihgfed, don't you think 10d100 is a bit too unreliable?
Rather do something like (10d20) + 895, then you'll end up with results between 905 and 1095, wich is somewhere around 1000. I don't like it when my special moves might do even less damage than my normal attacks.

Even in WarCraft 3 they use dice rolls for every unit, with the added value, as shown in the example above.

[Rainer Deyke]

In Feyna's Quest, damage is completely deterministic (although monster behavior is somewhat randomized). I feel that an action RPG doesn't benefit from randomized damage.

In Ghost of the Haunted Grove, I use the following formula:

[PoV]

I like picking random numbers and crashing the game when it's 74. ;)

[ThousandKnives]

[Rainer Deyke]

[Modanung]

[XMark]

I think dice rolls are a pretty good idea, because then you get an automatic bell-curve effect with the total damage inflicted from the rolls. Like when you roll 2d6, the range is from 2-12. However, you have a 1/36 chance of rolling a 2 or a 12, and a 1/6 chance of rolling a 7. It's realistic because in real life you would get occasional "flukes" or "fuck-ups" but you would gravitate around the same general effectiveness range with your sword swings/gun shots/etc.
_________________
Mark Hall
Abstract Productions
I PLAYS THE MUSIC THAT MAKES THE PEOPLES FALL DOWN!

[ThousandKnives]

[Jihgfed Pumpkinhead]

I've meant to get around to posting on this thread, given that I started it, after all. Sorry for being lax.

In Cairn Hill, I use a complex system for dice. Roll seven d100, and arrange them in order (so you could have 15, 31, 33, 48, 60, 78, 80). To find out degree of success, you take two dice and average them: your stats decide which two. If you have max stats, you take the top two dice, and would average them to get a score of 79; if you have average stats, you take the dice on either side of the middle die, 33 and 60, giving you a score of 47 (I round up because I'm nice).

Yes, it's fairly complex, and fairly random: but there are two reasons I like it. The first is that, in the game, success will usually be an either/or thing, which makes the degree of randomness less important, I think, and that, combat success and damage does not depend on this system, it uses something closer to a straight "subtract strength from opponent's health" system. The other thing, the one I really like, is that this system is explicit to the player. He can see the dice rolling around when he performs an action, can see which dice are used, etc. This way he can see the difference between levels not only as a final result, but by the process. I hate looking at a stat screen and seeing "Sword Skill 5", "Mace Skill 10", and wondering: "does this mean I'm only going to be twice as likely to hit with a mace, or not"? This way, it only takes one experiment to find out. Coolest in my opinion, though, is that there are spells to modify probability. You can set one die at 50, say, making all rolls tend more towards the mean. If you have the MP equivalent, you can set them all at 50, giving you an automatic 50 on all your rolls.

Anyway, I'm big on making things explicit to the player, and this seemed like a good way to do it. I did some testing, by the way, and the numbers came out as I wanted.

Usually, I think randomness is better the more similar encounters you will have, to provide variety. If all or most of your encounters are unique, you really don't have any need for randomness at all. Also, the more the enemies' statistics and the workings of the game are revealed to you, the more randomness is necessary, I think, so it doesn't just become number crunching.

Here's a thought: what about "forcing" deviation towards the mean? I don't know about you, but my big frustration isn't losing half the time when I should be winning twice as much, it's losing three times in a row. If one forced random results to be more fair, say, by halving the chance of an event re-occuring, would that make for a better game, do you think?

Thanks for all the answers, by the way.