Mapping thought of the Moment: Chance.

As part of my endeavors to maintain interesting discussion in the General Development forum (it became a bit too basic Q&A focused for my liking) i'm posting the first of what i'd hope would be many discussions regarding the theory behind 'good' map making.

Chance.

In this new age of trigger enhanced EVERYTHING i think it's time we discussed the implications of random chance in a map.

Is totally Random safe?

As most experienced Wc3 mappers will be aware blizzard uses an artificially manipulated random number system for all of its % based skills (IE critical strike). This is designed to reduce the effect of random chance on the outcome of a competitive melee game - it would be unfair for that blade master to crit 3 times in a row, denying you any chance of survival.

Now that triggered spell effects are so common we come across a problem: now that many effects are triggered, using 'true' random number generation, do we lose a valuable safety net that ultimately affects the competitiveness of our maps?

Take the example of a "Critical Strike", using an example of a purely triggered 10% chance to do 250 bonus damage (maybe we want to detect it to add SFX or some such such tomfoolery). In a purely random scenario we have 1 in 10^3 chance of this spell firing 3 times in a row (pretty much an instant, unavoidable kill).

This may seem quite low but we have to remember that in the average game the hero might attack, say 1000 times (possibly a great deal more). At which point this 1 in a thousand event looks increasingly likely.

At which point we come to the crux of the issue, do we as mappers, need to address this issue? Was blizzards move towards floating chances completely necessary, can we do without it in our maps? DO maps with large numbers of % based skills lose some of their depth and strategical value?

Yes but you dont have one unit attackin one units for that long without commanding it to do other things, e.g. use items or other skills which affect the random loop.

All my chance based skills use cooldown so there is no way something will happen too continuously

I believe this should only be an issue for chance spells with low percentages. For chance spells with high percentages (say 30% or more), they should be meant to make no difference whether they fire 3 times in a row or 3 times over a period on a certain target since over time the chance should even out.

I had no idea blizzard used a manipulated random number system o.o
Vexorian's idea works, but it also makes the passive ability work like an active one, say a passive has a chance of firing every attack, by putting an absolute with time, the hero has less incentive to increase attack speed. For melee games, heroes really can't change much of their own attributes, like attack speed etc, and melee games are high stake games, the loss of a hero to 3 crits in a row might mean the difference between $15000 and $10000.

IMO, unless it's not much trouble, there shouldn't need to be any controlling of random triggered abilities. A low percentage ability can be controlled by giving it a cooldown (assuming it's percentage is freakishly low, thus meaning an increase in attack speed won't really be discouraged since the chance of the ability firing within the short cooldown is really really low no matter what attack speed). Anything higher should be balanced to allow itself to happen many more times than normal in a battle.

One option is to avoid chance alltogether. In my arena, I don't think I have a single spell with triggered randomness. I have a skill based on drunken brawler, but instead of increasing the damage multiplier with level, I increase the chance to hit, to disperse the effect of the skill over time.

Most custom games require players to manage fewer units than in a normal melee game, so replacing passive abilities with active seems a good way to compensate. I see less reasons to use random elements in active spells than in passives, so there is really no need to use chance skills in custom maps.

However, ditching chance skills doesn't eliminate the luck factor. Ever had the enemy dissapear from your sight range just when your nuke spell cooled off and you were about to kill them? Or, were you ever killed by a nuke spell the enemy managed to fire just before you escaped from their sight range? The difference between life and death can be a matter of a few fractions of a second - the usual random fluctuation of server-host delay.

Here I come to my main point, Prevention of Game Determination by Chance (TM). PGDC (still looking for a name with a coler acronym) is a game design approach which seeks to minimise the butterfly effect. A small event that could be caused by chance, rather than by the difference of skill between players/teams, should not be allowed to accumulate in effect and break the balance for the remainder of the game. Instead, the game mechanics should be designed so that a teams/players sucess has a minimal effect on the balance status of the overall game, and where it has, there should be additional mechanisms to counter that.

Ideally (according to PGDC), all hereos in a hero map should gain experience equally, no matter how good the player is doing, how many times his hero has died or how many times he killed another hero. Whenever a hero dies, he should respawn with as little delay as possible and no penalty in gold or items. If one team is better than the other, then it will win in the end even if the other team constantly has equally powerful heroes. It will win even if the other team is given an advantage for loosing. However, the game will be more challenging for them, and more fun for the team that's loosing, and more interesting for both. It's a win-win situation.

However, this is drastically different from any maps we're used to playing. Usualy, when your hero dies in an AoS type map, the enemy gains gold and experience for it, getting better skills, stats and items for their hero. They also get more time to gain experience and collect gold while you're busy respawning. Once you do, you even loose some gold in some maps. By then, the enemy might have destroyed a tower or a unit production building, gaining a troop advantage in addition to the hero advantage.

But here we get to a problem: even games like this can last a long time if the teams are more or less even in skill. In such a case, a PGDC ideal map would have an endless stalemate. Of course, there's no such thing as two perfectly balanced teams, but we still don't normaly have days of time on our hands to find out which of the two almost equal teams really is better. If we want to have normal lasting games, chance has to decide at some point, like in the form of a timelimit, where the team that happens to have a better score at that moment wins... but it's still better if it's decided this way, rather than by some random event at the start of the game that destines one team to win no matter the luck the other team gets later, it makes the game more interesting.