Bashiok Explains Higher Difficulty and AI ImprovementsPosted 24 January 2011 by Flux
After I posted news about the Holiday Card plagiarism last week, I IM’ed Bashiok to let him know about it. He notified the Bliz web guys and the offending image went away, and while we were
laughing discussing that tragic incident in very serious tones, I took the opportunity to relay a reader question to him. Bashiok answered, and we talked about the issue for a bit, getting into some new and interesting info on the subject of monster AI at higher difficulty levels.
Afterwards I asked him if I could report on the info, and he said sure, but asked me to do as I’d done when reporting on some of our previous conversations, and not to quote him directly since he’d been speaking casually and with lots of sarcastic humor, rather than in his official capacity as the Diablo III Community Manager. So here’s a paraphrased version of our chat, with some added content and context.
The question that got us started was the first comment to this news post, from Unknown Prophet:
Hey Flux could you or someone ask @Diablo if monster AI will scale with difficulty? Or has this question already been addressed? I think it would be very awesome if the AI got more intelligent in Nightmare and Hell difficulty rather than there just being more monsters.
I quoted that whole comment to Bashiok, and he said that he didn’t know of any AI changes between difficulty levels. He doubted there would be any, since each monster was built with a simple and pointed purpose, and pointed out that the complexity and strategy of the game comes from how the different types of monsters and mixed and matched. I said that things might actually get *easier* if they tweaked the AI of just one type, and he didn’t disagree. Bashiok then stressed that the way they make Diablo III harder is by mixing the monsters in different, more complex ways.
All that said, he admitted that scaling dup to the higher difficulties was still largely a conceptual exercise. (I marked that issue to bring up later, since the D2 devs hadn’t gotten around to balancing the higher difficulties before release either, which resulted in a lot of imbalances and problems on Nightmare and especially on Hell shortly after release.)
Click through for the non-transcript of the rest of our discussion, with a special concluding tease about the imminent male Demon Hunter.
I replied to Bashiok’s opening remarks by redefining the concept of “improved AI.” What if that meant not a change in movement style or footspeed, but changes to the timing of events. For instance, what if monsters had a shorter delay between the time they saw the player and the time that they started to attack. Or if the cool down between their spell casts was shorter, or if they swung their weapons or fired their bows more rapidly, or if shamans were quicker to resurrect or buff their minions.
(Watch the end of Part 2 and the start of Part 3 of the Crafting Sanctuary panel for a great demonstration of how this sort of thing works in Diablo III.)
Bashiok thought that over, and said he didn’t think that more complex mechanics would be added to the monsters on higher difficulty level, but that something like my suggestions might possible work. Assuming the animations could keep up with the changes in speed.
I looped the conversation back around to my earlier thought at this point, returning to the issue of balancing the higher difficulty levels. I often force Bashiok to listen to stories about the D2 days, and at this point I talked about how the D2 Team never had time to really balance the higher difficulty levels. They did great work on normal difficulty, in terms of character progression and the experience curve, especially in acts 1 and 2, since they play tested through that part of the game countless times during development. But they didn’t have time for thorough testing past normal, as soon became obvious to players who reached Hell difficulty in D2C. (Most of it was way too easy, but occasional bosses with nasty combos of mods were unkillable or incredibly deadly.)
Bashiok commiserated, and said that testing higher difficulties was tough. He said the theory was to build the base normal experience, and tweak the numbers up from there.
I hoped he’d say a bit more about that, but he didn’t. So I segued to the difficulty of difficulties, and how tough it was to balance a game like D2 or D3 when there was inevitably so much variety in character strength at the later stages of the game. The Diablo games aren’t like WoW, where all the best items, the end game loot, are designed in advance, and the deepest dungeons are custom-tailored for raiding, where the required assortment of characters can be accurately planned for.
The guys designing Diablo III have no idea if your Clvl 50 Barbarian will have godly twinked uniques and crafted items, or if you’ll be scratching along with only the stuff you’ve found with your own bad luck. And they have to make the monsters beatable for the low end characters, while keeping them challenging for the l33t chars. And they have to make the monsters playable for all the different char classes, rather than knowing that an appropriately-mixed party will be there to cover for each other’s weaknesses, as in most MMORPGs.
While thinking such thoughts (but not typing them), I said that the various comments by the D3 devs about their plans for the end game were encouraging, and that I hoped they could at least do better than “endless cow runs.” Bashiok said that they sure hoped so, and I quickly chimed in, “So do we fans!”
He then elaborated, saying that the hardest part was adapting to or balancing character builds that they could never hope to conceptualize during testing. He pointed out that Diablo II had that, and that Diablo III would probably be one hundred times worse, “or better, if you’re a player.”
I agreed, and pointed out that fans would always find builds and exploits the developers hadn’t thought of. I then added that the biggest problem in the early days of Diablo II was that the developers hadn’t realized just how big a factor the experience multiplier would be once people were playing nothing but eight player games. It didn’t come up during the Beta, since there were only 1000 CDs sent out to fans, plus others to media, Blizzard people, and some of their friends and family. We beta testers played a lot of multiplayer games, but seldom with more than three or four people, and with all of Act One to explore and all five characters to play, there wasn’t much grinding or repeating of areas. We tended to get through Andariel with a character, then abandon them to try something new. (Often after one crazy ending PvP session in Andariel’s throne room.) As a result there were only a few characters on the beta ladder over level twenty, very few beta testers played the same character repeatedly, and eight players games were almost unheard of.
After a couple of months of that, when the beta test ended and the open beta/stress test began, we beta testers were shocked to see numerous characters in the high teens after just a few days. The open beta was basically a demo, with just the Barbarian to select, only his first two tiers of skills to use, and only the Blood Moor, Den of Evil, Cold Plains, and Cemetery to play in. Knowing that, I expected open beta players to level their Barbs up to about 10, and then get bored. Oh was I wrong! The open beta was fantastically popular, B.net hosted non-stop 8 player games, and with just one class and a few skills to chosoe from, players played their same character over and over again, item hunted Bishibosh and Blood Raven, PvPed like mad, and quickly achieved levels as high as characters had in months of the full beta test.
That sort of behavior continued after Diablo II’s release, even with B.net totally overwhelmed for the first couple of months after the game broke sales records and smashed Blizzard’s usage time predictions, and characters very quickly leveled well out of the planned experience curve, breaking the game difficulty in the early going.
I didn’t give that whole explanation to Bashiok, but I did point out some of the early bugs and exploits in Diablo II, including the amusing fact that with high enough Magic Find, it was possible to make every single dropped item set or unique (or rare, if/when there was no set or unique of a given item type). That was before there were diminishing returns on MF, thanks to players who immediately began to specialize entirely in MF gear; something the devs hadn’t tested out to its logical and inevitable extent during development.
Bashiok laughed at that, and when I gave a quick explanation of the problems with eight player games and the experience curve in D2, he said that D3 would fix that with a three-player limit per game. He was joking, but not by much, since unless something’s changed, they’re still targeting four players as the maximum in a game. At least on that front they won’t repeat D2’s eight-player problem, since they can, and are, testing all possible permutations of a four player game during development.
I then sneaked in a couple of questions about the Demon Hunter, since at that point last week I was like you guys, and anticipating, if not expecting, the male Demon Hunter to be revealed at any moment. I sent Bashiok the URL to the concept art you see to the right, a picture taken from the Gameplay Panel at Blizzcon 2010, and asked if he’d be the male DH if you changed his robes to black leather.
Bashiok would have none of it, and said that was old artwork, and that as we’d seen in the Gameplay Panel at Blizzcon, the original DH concepts had been a sort of desert/Arabian themed hunter. That was long in the past though, and according to Bashiok, the male Demon Hunter now is a good old gothic arrow shooter, like the female. Bashiok also said that the concept they’re releasing for his reveal is “bad ass.”
I don’t know if that’s a hint that there might be something more to it than just the concept art? Might the male DH be seen in action, or pictured doing something more than standing around and looking cool? And sexy for the ladies, as was promised during the Blizzcon panel. Hopefully we’ll find out about that this week, since last week didn’t work out quite as hoped.
Thanks to Bashiok for the info and clarifications, and this shows that it’s not futile to to post your questions for @Diablo in the comments here. You just might get them answered, and with a lot more than 140 characters!