Why I have no interest in Coils more specifically, T9.

[Mercutial]

Moogle EX was fun... but not entirely unique... Shape phase of Kaliya is similar.

That is an unbelievably astute observation. Right down to the random AOE's and debuffs. Did I mention I hate the add phase in T11 as well?

[Kosmos992k]

Omega,

I'm not sure that the majority of posters on this forum are welcoming of anything that disrupts their existing mastery of team jump rope in Coil and faceroll easy speed running in everything else. Just mentioning organized memorization - aka team jump role, aka team dancing, aka boring and requiring memorization skills more than anything else is usually enough to earn outrage from that very noisy minority of players. Personally I would prefer fights that used your skills in terms of mastering your class and being able to formulate strategy both before a fight and during in response to the intelligent actions of the Boss you are fighting.

Simple RNG driven Boss fights are annoying, because the level of difficulty depends almost entirely on RNG. Scripted fights become the organized memorization team dance challenge we have. But, fights with a Boss that has some decent AI and a skill palette similar in scope to players is far, more engaging and challenging. Difficulty can be varied simply by extending status effects, increasing their potency or increasing the damage done and/or Boss HP/MP. The actual decisions the Boss would make would be AI based, and not predetermined by script, so regardless of the difficulty level the fight would still be engaging, but the fight would scale better to player & gear levels.

It would still be practical to script particular phases by changing the strategy used by the AI for each phase such as; using status attacks, or the Boss becomes enraged and gains certain bonus status effects for a time, use of a one time ability (let's call it limit break) to drastically boost it's defense, or attack, or make an ultra wide range AoE with major damage. Those things can be scripted in to the different phases along different target priorities, or even whether it's playing defensively or offensively. There are many ways to adjust the strategy, and with multiple Bosses on the stage at once, players would be challenged doubly to handle them both, especially if their skills differ, but they can cooperate dynamically.

This is much more difficult to code than a simple script or RNG, but so much more rewarding for the players. Once that kind of system is in place though, the difficulty can be scaled by several means, nerfing stats, nerfing defense/attack/status potency or even dumbing down the AI for an easy mode. Conversely allowing the full AI, boosting stats and defense/attack/status potency or even opening up an additional 'ultimate' attack could increase the difficulty. you can vary the difficulty dynamically based on player gear and level too, so an ilvl 130 party would have a harder time than a party with an average ilvl of 90.

Frankly I don't understand why they have relied so heavily on organized memorization, except that it is better than simple RNG driven battles and far easier to code and balance than an AI based fight. Still, once the AI is in place, it can be built upon and tweaked for each individual boss. Personally I think this would be a very worthwhile exercise for them, and it would prevent content from becoming what some would term faceroll easy when players memorize and moderately overgear the fight.

[KarstenS]

Personally I would prefer fights that used your skills in terms of mastering your class

Its alway interesting, how people, that never have done the content they complain, think, its just about mechanics and not personal skill.

Thats not the case. You cannot beat that content just by caring about mechanics. You also have to play you class at its best or you won't meet the requirements to beat that content. You have to plan ahead as a team or you won't beat that content.

[Whocareswhatmynameis]

Both randomness and scripted have pros and cons. I can think of a few times where Imdugud did nothing but heat lightning-charge (no tether-charges) and that made the fight easier. So even with randomness there is a real possibility of getting the really lucky set of mechanics and a really unlucky set of mechanics. Then ppl would be complaining the fights arent fair cuz team A got "lucky" or team A just wiped until they got the "lucky" string of mechanics.

Even randomness has it's drawbacks.

[Ayrie]

Kosmos:

Simplest explanation I can offer is it is far harder to manipulate/bug an encounter when the boss' skill sets are on timers.

Script and AI are the same thing.

AI is an illusion created by stacking many conditionals. If the game sees these three triggers do that. The problem with that is it can be figured out and manipulated. You could eliminate or avoid critical moves by never triggering their conditions.

A real quick example is Matt samurai genkai in FFXI. He would not start his combo until hit 3 times. Want a easy way around the fight? Know that you get 3 easy hits before he is fixing to rail you. It didn't change the outcome, but it shows how conditionals can be exploited to lower the intended difficulty.

[Gillionaire]

Kosmos:

Simplest explanation I can offer is it is far harder to manipulate/bug an encounter when the boss' skill sets are on timers.

Script and AI are the same thing.

AI is an illusion created by stacking many conditionals. If the game sees these three triggers do that. The problem with that is it can be figured out and manipulated. You could eliminate or avoid critical moves by never triggering their conditions.

A real quick example is Matt samurai genkai in FFXI. He would not start his combo until hit 3 times. Want a easy way around the fight? Know that you get 3 easy hits before he is fixing to rail you. It didn't change the outcome, but it shows how conditionals can be exploited to lower the intended difficulty.

and that's why we need test servers to work out the kinks. It may be harder to make, but the polished product will be superior. SE's in-house testing team doesn't quite seem capable of catching all the bugs evident by the hotfixes every single time they roll out a patch. I get it hotfixes are inevitable from time to time, but having to do it after every single patch seems wrong.

[Kosmos992k]

Kosmos:

Simplest explanation I can offer is it is far harder to manipulate/bug an encounter when the boss' skill sets are on timers.

Script and AI are the same thing.

AI is an illusion created by stacking many conditionals. If the game sees these three triggers do that. The problem with that is it can be figured out and manipulated. You could eliminate or avoid critical moves by never triggering their conditions.

A real quick example is Matt samurai genkai in FFXI. He would not start his combo until hit 3 times. Want a easy way around the fight? Know that you get 3 easy hits before he is fixing to rail you. It didn't change the outcome, but it shows how conditionals can be exploited to lower the intended difficulty.

I'm not sure I follow your reasoning that AI and scripting are the same. Yes AI has to function based on a set of rules - just like we do, but if it's nothing more than a fixed set of rules that trigger a response, without evaluating differing courses of action and their outcome, it's not really AI.

AI needs to be capable of responding not only to player actions and it's own state, but also needs to be able to anticipate possible responses to actions and determine which is better/worse. It also has to have a forward looking capability to predict responses and determine which course of action has the best potential outcome. I'm not thinking of Deep Blue chess playing AI here, just an ability to look at the potential responses to actions and choose the best one.

In your example, why did Matt start his combo after 3 hits? Because it was scripted, not because AI decided it was the best course of action. Of course players can exploit that kind of situation, just like we can exploit the enrage timer in Turn 2. Relying on such fixed triggers is not really artificial intelligence, though it's often described as monster AI in game descriptions. IMHO, it's only artificial intelligence if it adjusts based on player actions.

With regard to your example, the combo is triggered by a single definable condition. That is easily exploited and lacks intelligent direction. On the other hand, if the AI evaluated the situation and determined that despite the trigger making it available, using the combo would not be the best outcome, the player would be unable to exploit it so easily. That's what makes it more complex to implement and test - obviously.

[Ayrie]

A small dev team would take years to fully debug a script from exploits. It takes a player base a fraction of the time to discover them.

That deck is woefully stacked against the devs. You like new content, like I do, yah?

[Ayrie]

Another example of scripted AI decision making is tail swipes.

Most of these moves (t1, t10, ogres doing the rear facing elbow drops) do have specific triggers: boss state is not engaged in a higher priority attack and someone is in position to be hit. Triggers are met and the boss takes a swing. How do we exploit it? Some bard floats in and out of it to force it and force a damage lull for the healers.

All AI is: nested conditionals that gives the illusion of decision making. Once the conditions are discovered you can easily exploit it. League of Legends bots come to mind as another example. So either move in a defined pattern on a timer... Or move in a prescribed pattern to force, or eliminate, mechanics. Devs call.

[Kosmos992k]

Another example of scripted AI decision making is tail swipes.

Most of these moves (t1, t10, ogres doing the rear facing elbow drops) do have specific triggers: boss state is not engaged in a higher priority attack and someone is in position to be hit. Triggers are met and the boss takes a swing. How do we exploit it? Some bard floats in and out of it to force it and force a damage lull for the healers.

All AI is: nested conditionals that gives the illusion of decision making. Once the conditions are discovered you can easily exploit it. League of Legends bots come to mind as another example. So either move in a defined pattern on a timer... Or move in a prescribed pattern to force, or eliminate, mechanics. Devs call.

Is that an exploit or a strategy? The three card force is used quite effectively against living breathing humans, so why should we suppose that other living breathing humans cann not come up with a strategy to force certain moves by an enemy in a game. It's not whether it can be done, it's whether that strategy (which exploits player knowledge of the enemy) breaks the fight or simply mitigates a certain move by the Boss.

Developers do not and should not be looking to block every possible strategy players can use, instead they need to ensure that the AI cannot be totally circumvented effectively disabling it.