Darkfall "Released"

[tmp]

You're failing to understand that the routines in geek_the_mage() could be simplistic or complex.

I'm just saying when your target selection is based on indiscriminate if() like that, it'll be mercilessly gamed. What makes you believe that mage is *always* going to be the optimal target? For this matter, which mage? Remember, multiplayer game.

[UnSub]

Here is my uber AI algorithm

1) Kill the healers
2) Kill the DPS
3) Laugh and run away from the tank.

Here is the player reaction to your AI:

Risk vs. reward! Mobs too hard! Going back to WoW!

I did think of one way to make things more interesting - remove the colour con system. It really is the cheapest way of letting players know which mobs they can steamroll and which ones they need to avoid. I know what it is there, but it really becomes a shortcut in terms of knowing what to expect from a mob.

[Lantyssa]

*sigh*

You're taking that example too literally.  I'm not saying always go for the mage first, it's just a phrase from Shadowrun where it is a truism.  Programmers can use conditionals to determine which targets have priorities.  Often mages are a priority though, so I used it as a snippet of code.  Once you determine priority, then the mob chooses its actions.

...

How does a player determine the best strategy?  How often do they change up the "Press 1, 4, 2, 3.  If dead pick new target and Press 1, 4, 3, 5 (because 2 is on cooldown).  If not dead then Press 4, 3, 5.  Oh, if the target is casting then Press 6" routine?  Take that, and make the mob do it.  Ta da!  Combat AI.  It's the easiest part of an AI.  It can be fine-tuned to think of every conceivable situation, it can be randomized to make it dumb, or it can be given weighted but still random priorities for a bit of realism, variety, and avoidice of gaming a rigid behavior.

What's tough is pack behavior (independent of game type), cover (independent), world interactions (independent), hunt and seek (independent), leading player(s) into an ambush (independent), etc.  MMOs are not harder to program this for.  They're not easier.

The fallacy is thinking single-player means single target.  Many single-player games have multiple units on your side.  Many MMOs people play solo, or groups try to pull single targets.  You cannot give a blanket statement that one is harder than the other.  You can only make such a claim by specifying exact conditions, and I can very easily give an easy MMO and a hard single-player as a counter-point.  Which means it is entirely dependent upon the game's design.

[slog]

Here is my uber AI algorithm

1) Kill the healers
2) Kill the DPS
3) Laugh and run away from the tank.

Here is the player reaction to your AI:

Risk vs. reward! Mobs too hard! Going back to WoW!

I did think of one way to make things more interesting - remove the colour con system. It really is the cheapest way of letting players know which mobs they can steamroll and which ones they need to avoid. I know what it is there, but it really becomes a shortcut in terms of knowing what to expect from a mob.

1)  that was my point.  (see my earlier post)
2)  removing con will also send players back to WoW

[Sheepherder]

I remember this one time I spawned something like 255 bots in a match of CTF-Face in Unreal Tournament.  There were so many it created a bunch of other teams with no flags that just ran around and shot people and stole their flags.  That shit was fragtastic.  Ran pretty well for a late model Pentium 4 HT with a gig of ram too.

Anyways Lantyssa, you just don't get it: there are so many possible targets that for an AI to handle them all is computationally infeasible.  Such a thing just would not run well on any server, because it naturally would have problems with scaling.  Any psuedo-code you can provide to the detriment of this arguement would be meaningless, because no example of bots operating on a massive scale exist.

Also, the arguement that people don't want smart bots is dumb.  They don't want to get their shit stomped by bots and have to run their shit back to their corpse for three minutes and eat a repair bill or experience drop that further pounds their cock into the ground.  Almost every evil in a MMO can be linked to or is a direct result of a heinous risk / reward structure that has been inbred into the genre by successive generations of cockpounding shithead rockstar developers.  I have yet to see a compelling reason why a person should be forced to run back to their corpse as opposed to waiting for a timer (anyone sadistic enough to say it's fun?), much less why they should pay for their failure in ways other than losing the fight and having to do it again.

[Venkman]

If WoW wasn't delivering half of Activision's profit right now with more people than the entire rest of the paying genre combined, this situation might actually change.

But they are, and all current big-budget MMOs under development will take cues from it.

And when they inevitably decline, so will the VC interest in big-budget MMOs in general, in favor of "Facebook for kids" type low-risk things.

Yea, so, I basically believe this problem will never ever be solved. It's about the only thing I'm pessimistic about.

[Merusk]

I have yet to see a compelling reason why a person should be forced to run back to their corpse as opposed to waiting for a timer (anyone sadistic enough to say it's fun?), much less why they should pay for their failure in ways other than losing the fight and having to do it again.

Since all things revolve around or come back to WOW discussions these days, I'll go with that game as an example.

Runback serves two purposes that I can see.  It provides a timer penalty and allows people to respawn in an area where perhaps not so many mobs are going to pound their face in.

If you only have the timer without the forced respawn radius you'll see players exploiting the fuck out of the system to leap around wherever. It becomes the new 'safe travel.' 

If you have the forced respawn location without the timer (you're always alive at the GY after death) then players bitch endlessly about how they've got to wade through a bunch of things to get to where they just were.  That sucks even more.

You could do a timer at the corpse with a respawn perimeter, but then you're fucked if you die in a bad spot.  Like falling off the edge of the world, or in the middle of a group of elite mobs, etc.   There could be an optional "release to graveyard" button, though. Not a bad solution.

I don't mind the death penalty of money like WoW's done it.  When leveling the worst repair bill I've ever had was only 10-11 gold and that was on a character that was more than able to pay that off in one or two quests.  Raiding is -or was- meant to be a more hardcore experience so its penalty was harsher. The better gear you get the more it costs to repair, after all.  As that system opens up to the average player, you do have to wonder if the costs are a little high.  Gold IS fairly easy to get, though, and fairly useless if you're raiding a lot and in any way decent at it.

 Xp debit/ loss systems can fuck themselves in the ass.  Gold you aquire while questing and mob killing in general, it's easy to get that back while still advancing your character.  XP, however, is a double set back. Lost time from the first time you failed and time needed to make up to where you'd be if you didn't have that debit.  Item loss systems?  I wouldn't even consider the designer of such a system in a modern DIKU to be sane.  That's a probability of an infinite loss of time there if your luck just plain sucks. You may as well just throw the dev costs out the window since you won't be making it back with the niche of a nice audience you'd attract.

[tmp]

You're taking that example too literally.  I'm not saying always go for the mage first, it's just a phrase from Shadowrun where it is a truism.  Programmers can use conditionals to determine which targets have priorities.  Often mages are a priority though, so I used it as a snippet of code.  Once you determine priority, then the mob chooses its actions.

Of course. But then you know, once you start introducing these conditionals and finetune it to encompass large enough range of situations... it stops being 'trivial'. So that statement simply isn't true. Not any more true at least than me saying coding the AI for single player game is trivial because it's just: kill_player( target ); This is of course just a snippet of code, too. And it doesn't tell anything about the possible complexity of the problem... it's just made to look simple to fit the argument.

[Lantyssa]

The code is trivial.  I've written it before.

[tmp]

The code is trivial.  I've written it before.

Your implementation was trivial. Out of curioisty how did it shape the resulting gameplay?

[Lantyssa]

Depends which one of several procs you want to talk about.  One was a combat monster that I would have been scared to fight as a player.  One cast spells believably, even using a decoy to pop out of combat and run away without the character noticing.  One just had neat effects.  Several I tweaked as they hadn't been touched in years and multiple updates.

Have you ever programmed an AI routine?  Seriously.  Because I get the impression no one saying how hard it is has ever done so and I'm just beating my head against a brick wall.

[Sheepherder]

You could do a timer at the corpse with a respawn perimeter, but then you're fucked if you die in a bad spot.  Like falling off the edge of the world, or in the middle of a group of elite mobs, etc.   There could be an optional "release to graveyard" button, though. Not a bad solution.

Of course, both the time limit and the resurrection radius are needed.  If you are really interested in a refutation of the current system in WoW go quest in Feralas.  I think they've recently added more spawn points in, but when I first leveled my warrior three minute corpse runs that required your rapt attention at the risk of falling into a crevasse and adding another two minutes to your resurrection time were common.

I don't mind the death penalty of money like WoW's done it.  When leveling the worst repair bill I've ever had was only 10-11 gold and that was on a character that was more than able to pay that off in one or two quests. ... As that system opens up to the average player, you do have to wonder if the costs are a little high.  Gold IS fairly easy to get, though, and fairly useless if you're raiding a lot and in any way decent at it.

The death penalty drives players to low risk / reward scenarios when possible and beats on the new guys who cannot afford their mounts (unless they are knowledgeable enough to farm trade materials on an established server).  This is a negative effect if your game is all about being a heroic type guy.  It's also detrimental to teaching players how to play their class well if they're always choosing easy fights.  Lastly, repeatable content is a shitty solution to the increasing gold sinks tied to the end-game.

[tmp]

Depends which one of several procs you want to talk about.  One was a combat monster that I would have been scared to fight as a player.  One cast spells believably, even using a decoy to pop out of combat and run away without the character noticing.  One just had neat effects.  Several I tweaked as they hadn't been touched in years and multiple updates.

Uhmm i thought we're talking about triviality (or lack thereof) of the mechanics responsible for target selection in multiplayer environment. This is what i thought you meant by your "it's trivial, i'd done that" comment. Your answer now is leading me to believe you are talking about something quite different (the 'moves' available to creature in combat or whatever)

Have you ever programmed an AI routine?  Seriously.  Because I get the impression no one saying how hard it is has ever done so and I'm just beating my head against a brick wall.

I do some programming rather regularly; not specifically game AI but i'm familiar with the concepts. So it's not really that i'm baffled by the technical side of it, but rather by the stance that such important to 'get right' part of gameplay which is generally present in mutliplayer games but not in single player games can be downplayed along lines of "oh it is so trivial the fact it has to be additionally implemented does not make the game harder to program at all". It simply doesn't make sense to me -- having to code A + B is always going to be more difficult than having to code just A. How much more difficult can be discussed, but to question the extra complexity itself... just mind boggling for me, sorry.

[UnSub]

Have you ever programmed an AI routine?  Seriously.  Because I get the impression no one saying how hard it is has ever done so and I'm just beating my head against a brick wall.

I've never seen a redname go, "Programming AI? Totally easy." In fact, I've seen quite the opposite.

The concepts of smart AI are easy. Implementing them appears harder.

[Lantyssa]

Have you ever programmed an AI routine?  Seriously.  Because I get the impression no one saying how hard it is has ever done so and I'm just beating my head against a brick wall.

I've never seen a redname go, "Programming AI? Totally easy." In fact, I've seen quite the opposite.

The concepts of smart AI are easy. Implementing them appears harder.

Last attempt, because I'm giving up:

1) There is combat.  Pretty easy, as most of your mob's abilities are going to be limited by the combat system.  You can cheat or add special mob-only abilities, but from a design perspective it's generally best to keep them similar to what the players get.

2) There is targeting.  Setting a priority list for your mob who to attack, who to switch to when.  Comparing game A to game B in difficulty only works if both have multiple targets.  They may be NPC targets, they may be all player targets, or they may be a mixture.  If you're going to compare a single-player, one target for the mob versus an MMO where you always have groups, duh, it's going to be harder to write an AI that fares as well if everything else is equal.  If you're going to compare a single-player game where NPCs can help, and a multi-player game with a similar number, then it's dependent upon the systems between games and not the number of targets.

3) Not all parts of AI is easy, which I've never said.  Making a mob behave realisticly is difficult, outside of combat moves.  Interacting with the world, pathing, cover, etc.  Some of these can play into combat, so it's not entirely a cut-and-dried thing, but I'm trying to talk in generalizations since no one has proposed parameters.

4) Being an MMO does not mean the AI is automatically harder to program for.  A sandboxy single-player game will have much more complicated AI than a poorly done EQ clone.  Had the efforts been reversed then the EQ clone would have had the better AI.

5) Target switching in MMOs tends to rely on aggro and not AI.  Mainly because design decisions have led to the holy trinity of Tank/DPS/Healer.  Having mobs always targeting your squishes would blow the entire Holy Trinity to pieces.  It's not harder to do, however no one wants to bother coming up with a non-Trinity profession system.  Now designing an alternative might be extremely difficult, but it's not a problem with the AI.

6) Design decisions regardling classes also make it very difficult to make an intelligent AI that doesn't rape one type of player while being much less effective against another.  It's a balance issue with class systems where you have a huge range of hit points, armor, and resistances.  Since MMOs have fallen into that trap, it's not surprising they use dumb AI.  Making it smarter isn't the problem.  Making it not drive players away is.  In that sense it is more difficult.  Again, for the fortieth time, it's a design issue and not because of the number of players.

Believe me or not, I don't care at this point.  I'm done trying to educate.

[slog]

Have you ever programmed an AI routine?  Seriously.  Because I get the impression no one saying how hard it is has ever done so and I'm just beating my head against a brick wall.

I've never seen a redname go, "Programming AI? Totally easy." In fact, I've seen quite the opposite.

The concepts of smart AI are easy. Implementing them appears harder.

Last attempt, because I'm giving up:

1) There is combat.  Pretty easy, as most of your mob's abilities are going to be limited by the combat system.  You can cheat or add special mob-only abilities, but from a design perspective it's generally best to keep them similar to what the players get.

2) There is targeting.  Setting a priority list for your mob who to attack, who to switch to when.  Comparing game A to game B in difficulty only works if both have multiple targets.  They may be NPC targets, they may be all player targets, or they may be a mixture.  If you're going to compare a single-player, one target for the mob versus an MMO where you always have groups, duh, it's going to be harder to write an AI that fares as well if everything else is equal.  If you're going to compare a single-player game where NPCs can help, and a multi-player game with a similar number, then it's dependent upon the systems between games and not the number of targets.

3) Not all parts of AI is easy, which I've never said.  Making a mob behave realisticly is difficult, outside of combat moves.  Interacting with the world, pathing, cover, etc.  Some of these can play into combat, so it's not entirely a cut-and-dried thing, but I'm trying to talk in generalizations since no one has proposed parameters.

4) Being an MMO does not mean the AI is automatically harder to program for.  A sandboxy single-player game will have much more complicated AI than a poorly done EQ clone.  Had the efforts been reversed then the EQ clone would have had the better AI.

5) Target switching in MMOs tends to rely on aggro and not AI.  Mainly because design decisions have led to the holy trinity of Tank/DPS/Healer.  Having mobs always targeting your squishes would blow the entire Holy Trinity to pieces.  It's not harder to do, however no one wants to bother coming up with a non-Trinity profession system.  Now designing an alternative might be extremely difficult, but it's not a problem with the AI.

6) Design decisions regardling classes also make it very difficult to make an intelligent AI that doesn't rape one type of player while being much less effective against another.  It's a balance issue with class systems where you have a huge range of hit points, armor, and resistances.  Since MMOs have fallen into that trap, it's not surprising they use dumb AI.  Making it smarter isn't the problem.  Making it not drive players away is.  In that sense it is more difficult.  Again, for the fortieth time, it's a design issue and not because of the number of players.

Believe me or not, I don't care at this point.  I'm done trying to educate.

so is the t;;dr version:  Combat AI isn't hard.  It's "Combat AI that players will like" when working with the Holy Trinity that's hard.

[Merusk]

There was a boss encounter in Magister's Terrace involving 5 mobs that behaved almost exactly like players.  They would use all their abilities to their fullest and had odd aggro mechanics/ CC immunites.  Players HATED it because it relied on more PVP-type skill than straight out CC & kill of most instance fights.   There's a similar encounter in BRD if you did the Tier 0.5 gear quests.

So yeah, devs can program mobs to act a lot tougher than they do. They don't because players hate it. 

[Venkman]

Making it not drive players away is.

Welcome to a page and a half ago.  What was this thread about again?

[Malakili]

There was a boss encounter in Magister's Terrace involving 5 mobs that behaved almost exactly like players.  They would use all their abilities to their fullest and had odd aggro mechanics/ CC immunites.  Players HATED it because it relied on more PVP-type skill than straight out CC & kill of most instance fights.   There's a similar encounter in BRD if you did the Tier 0.5 gear quests.

So yeah, devs can program mobs to act a lot tougher than they do. They don't because players hate it. 

I dunno, I don't know a lot of people who outright hated that encounter.  In fact, I think most people that did dislike it was because it DID often require a lot of CC.  Also recall the BRD fight you are talking about, and it was pretty trivial at the time, if I recall. (Though by the time .5 game out my guild was so overgeared for the .5 quests that we were only doing them for fun).

Anyway, the fights are definitely more difficult than the average PvE fight, but I don't think they are totally hated.  If however, there is a reason that it isn't liked, it is probably that it is less predictable.  Generally PvE in WoW is about planning, and executing a plan.  Its very proactive, while PvP is very reactive.  Neither playstyle is inherently better than the other in my opinion, its kind of like offense v. defense in a lot of sports.  The offense is trying to execute a pre-planned play, while the defense is reacting to the offense to stop it.


In any event, both can be fun, but probably both aren't fun for every single player. 

[Ratman_tf]

So yeah, devs can program mobs to act a lot tougher than they do. They don't because players hate it. 

I don't think they hate it so much as resent it when 90% of the game is aggro management and CC, and then drop in a boss fight that requires a change in their fundamental playstyle and expectations.

Kinda like having a 1-80 solo game and then making everyone go raid. 

[Modern Angel]

Kinda like having a 1-80 solo game and then making everyone go raid. 

Yeah, because people sure fled WoW when they found that out.

[Venkman]

If however, there is a reason that it isn't liked, it is probably that it is less predictable.

That is the point. People come for predictability and over the last decade, mobs have devolved to deliver that to them. Also this talk about making AI smarter runs right up against the expectation that they're not. Because anything else means slower XP, slower advancement to new abilities, slower moves to new content. The devs aren't stupid, rather, the game system requires the AI be.

Smarter AI is possible, but you need to change the purpose of the game to adapt.

The best thought experiment I can think of is replacing or complementing players in WoW BGs with bots. There it can make sense because while players want predictability everywhere, at least they're not expecting it in PvP (much, anyway).

[Ratman_tf]

Kinda like having a 1-80 solo game and then making everyone go raid. 

Yeah, because people sure fled WoW when they found that out.

Maybe they can go level some alts instead?

[Soln]

so is this game coming out or not?  and is it worth investigating or fence sitting until (like WAR) its real quality is revealed?

[Venkman]

Fence-sit unless you're absolutely desperate to have niave hope about something 

[Goreschach]

Fence-sit unless you're absolutely desperate to have niave hope about something 

You've got to be pretty fucking naive to even be near the fence on this one.

[Nebu]

You've got to be pretty fucking naive to even be near the fence on this one.

Are you suggesting that 10,000 screaming fanbois could be wrong? 

[sidereal]

Fences are for carebears.  If you were even thinking of just sitting on a fence, you are not worth of Darkfall.  The only good sitting is sitting on the face of the corpses of your victims and then only long enough to thrust your pelvis a handful of times before re-entering the fray.

[Senses]

I'm sitting on the fence because despite the assurity that everyone here has that it will fail, I am actually hoping for it to succeed.  All the years of development and bs that has come before is really unimportant to me, either its a fun working game when its available to play, or it isn't.  Why that deserves 38 pages of tangential discussion with interspersed apocolypse chatter I'll never understand.

*corrected the spelling error.

[UnSub]

We're bored and Adventurine is mean for not releasing more info to mock?

I hope Darkfall lives up to everyone's expectations and gives PvP the shot to the arm all its fanbois want. But it won't.

[Signe]

I'm sitting on the fence because despite the assurity that everyone here has that it will fail, I am actually hoping for it to succeed.  All the years of development and bs that has come before is really unimportant to me, either its a fun working game when its available to play, or it isn't.  Why that deserves 38 pages of tangental discussion with interspersed apocolypse chatter I'll never understand.

Dude.  It wasn't 38 pages of Darkfall talk.  Half of it was derail.  Even we're not deranged enough to talk about this game for 38 whole pages.  At least not in a row.  And, spelling aside, "tangental"?  Really?  Where did any of us have the tantrum? 

[Righ]

so is this game coming out or not? 

http://www.youtube.com/watch?v=WOdjCb4LwQY

[NiX]

What does any of this have to do with Hello Kitty Online?

[patience]

If however, there is a reason that it isn't liked, it is probably that it is less predictable.

That is the point. People come for predictability and over the last decade, mobs have devolved to deliver that to them. Also this talk about making AI smarter runs right up against the expectation that they're not. Because anything else means slower XP, slower advancement to new abilities, slower moves to new content. The devs aren't stupid, rather, the game system requires the AI be.

So you just implied XP rates can't be modified to compensate for the increased difficulty. Brilliant.

[ashrik]

I don't think that's what he was implying at all, specially since xp rates are already modified for difficulty within the current system we have- and in a clear way that people are used to. I'm no doctor, but I'm pretty sure you're not the only one in this thread who's played an RPG before.

I don't want increased XP for mobs whose sole ability is to be more annoying than others. Even more so when that annoya- ahem- advanced A.I. demonstrates itself like the goblin in that video.

So it's not that

So you just implied XP rates can't be modified to compensate for the increased difficulty

Or even that the difficulty cannot be increased in the way they're talking about here. It's that no one wants it.

Anyways- back to Darkfall. There's a new leak out, for those who are interested. I wonder how many people are in the beta atm. I really think I'm going to be trying it when I get the chance. Nice to see that not every Gobelin scout fights like the one we saw earlier.