Branching out: How limiting skill choices made Diablo 2 more fun

On Diablo 2’s 18th anniversary, Blizzard North’s designers recount the triumphs and pitfalls of growing the game’s skill trees.

In the mid-1990s, computer role-playing games were not easy to play. Their beginnings were mired in rolling virtual dice to set dozens of character attributes, their interfaces cluttered screens with buttons and icons and text. Worse, gameplay plodded along, making even best-sellers like publisher SSI’s vaunted Gold Box series of Dungeons & Dragons-licensed titles appealing only to players who spent their free time agonizing over every roll of a d20.

This an excerpt from Stay Awhile and Listen: Book II, now on Kickstarter, which chronicles the making of StarCraft and Diablo 2, and explores the culture of Blizzard Entertainment and Blizzard North from 1997 through 2003. The first book was published in 2013 and is available in paperback and on Kindle.

Throughout the development of the original Diablo, Blizzard North’s team followed one rule. Their game had to be simple. So simple a new player could sit down and learn how to play in minutes, if not seconds. Although the game went through numerous changes—from turn-based to real time, from single-player to support for up to four adventurers over Battle.net — its singular focus remained unchanged from the project’s origins in 1995.

“Why is it so hard for developers to understand that the base core gameplay mechanic isn’t the leveling up or the item collection? It’s the clicking,” said Michio Okamura, the lead character artist on Diablo.

“Every once in a while, Dave Brevik would shut his door and evaluate the game, and one of the ways he’d do that is play the game one-handed,” agreed Rick Seis, a programmer on Diablo, “and we’d make changes based on that. That was part of our goal: To make things easy, so easy your mom could play it.”

Dave had cofounded Blizzard North as Condor, Inc., with Max and Erich Schaefer. The concept for Diablo had come from Dave’s fixation with roguelikes during college. After Dave, Max, and Erich had met Blizzard Entertainment cofounder Allen Adham at the Summer 1994 Consumer Electronics Show, Adham had taken an interest in Diablo and urged Blizzard’s parent company, Davidson & Associates, to acquire Condor, which it rechristened Blizzard North in 1996.

Diablo became a best-seller thanks in large part to its click-centric gameplay. Following that golden rule facilitated other advancements over CRPGs of yore, namely a cleaner interface relative to classic RPGs, and a blazingly fast pace. The sound of frantic clicking punctuated by the shrieks of demons and the merry clink of gold coins raining onto the ground was as much a backdrop to millions of players’ descent through Diablo’s gothic dungeons as Blizzard North composer Matt Uelmen’s moody soundtrack.

Above: A boss fight in Diablo 2

As Blizzard North staffed up for Diablo 2, many components of the original game came under scrutiny. One of the team’s largest concerns was the original game’s spell systems. Players learned spells by reading spell books found in libraries nestled away in dungeons, dropped by monsters, or sold by Adria the Witch in the town of Tristram.

Spell books presented several problems. For one thing, the books were procedurally generated. Warriors could make do with any club, axe, sword, or other blunt instrument the game’s algorithms spat at them, but players who rolled a Sorcerer had to cross their fingers and hope those same processes blessed them with tomes of Firebolt, Chain Lightning, or some other damage-dealing sorcery.

Another issue was the interface, a book with several pages, used to cast spells. “The spell book was kind of a disaster that didn’t really work how we wanted it to work,” said Dave Brevik. “Blizzard South wanted it to work one way; we wanted it to work another. They wanted all spellcasting to come out of the book. So you had to have the book interface up, you clicked on the spell you wanted, then clicked on the monster in the play window. We wanted it to be tied to the mouse so you could just click and [assign] spells to the mouse buttons. It ended up being a big fight between us, and both systems went in. But nobody ever used the spellbook system.”

The third problem was that any of the three hero classes could use almost any weapon or magic spell. “The character classes were the main thing that I knew was broken from a design standpoint. The fact that all of the classes had access to the same pool of spells meant that everyone was essentially playing a Fighter-Mage,” said Stieg Hedlund, designer on Diablo 2.

Overlap between classes played to Diablo’s accessibility. No matter which hero players chose at the outset, anyone could wield a sword, fire an arrow, or cast a spell by investing experience points in the appropriate attributes. At the same time, that versatility detracted from the pride RPG players sought by building unique avatars.

Dave Brevik thought of an alternative late one morning during his shower, a ritual that guided many facets of development at Blizzard North. Dave worked late and let his mind wander as he cooked under a torrent of hot water. More than once, his meandering guided him to a solution for a problem the team was struggling with, or a new feature. The team learned to recognize Dave’s “I just had a shower idea” expression when he marched through the front doors, and braced for features that would translate to loads of extra work. This time, that extra work would pay off.

“I came into work and told it to Erich,” Dave said of his idea for skill trees. “I loved Master of Orion, games of the 4X space-conquering genre. Traditionally in those things, they have tech trees where players have to make research choices. It just dawned on me one day: What if player classes could choose their skills, just choose a path through a tree.”

Above: The first Civilization’s tech tree.

Dave’s early visualizations for skill trees resembled those of Master of Orion and Sid Meier’s Civilization — the game credited with introducing research and tech trees to video games — almost verbatim. 4X and other turn-based strategy titles such as Civilization tended to offer dozens if not hundreds of upgrade possibilities. Stieg Hedlund was vital in paring down Diablo 2’s rendition.

“While I was taking a break from Diablo 2 and working on polishing off StarCraft, we [both Blizzards] made a tech-tree poster,” Stieg explained.

The tech-tree poster shipped in every StarCraft box when the game launched in the early spring of 1998. One side of the poster showed an easy-to-follow flowchart of each building players needed to construct first in order to build more advanced structures later. The other side showed a similar flowchart for units such as the Terran Marine, Zerg Hydralisk, and Protoss Templar.

“While no one had done something like this in-game — in fact in game these games tended to be very digital, only showing you the things you could build next — this struck me as doable, and a great potential way both to afford and message class progression,” Stieg continued. “I started working with this idea building in the RPG-relevant concerns of levels and skill ranks, arriving eventually at skill trees.”

Although Stieg was the only developer officially credited with the title of designer on Diablo 2, Blizzard North’s egalitarian culture meant that anyone, from any team — character artists, music and sound effects, environment art, engineering, even office managers such as Karin Colenzo and Kenny Williams — were welcome and encouraged to chime in with ideas. For that reason, the instruction manuals for Diablo and Diablo 2 credit game design to the entire studio before listing credits for specific teams and roles.

“It was very collaborative. So collaborative that I can’t even remember where certain things came from,” said Phil Shenk, Diablo 2’s lead character artist. “I think I came up with corpse explosion, but I might be wrong. If I didn’t come up with that, I pushed for it.”

“Designing skill trees was a lot of work but a lot of fun,” added Erich Schaefer. “It was maybe the hardest thing to do, design-wise, both creating them and balancing. We’d have meetings and decide what kinds of skills we wanted to do. I’d do rough layouts and we’d try some out, and half of them wouldn’t be very good. We redid those trees over and over again. It was just brute force.”

“We’d have a list of skills that either we came up with them or we were given them, but either way they were really vague,” remembered Tyler Thompson, a programmer on Diablo 2. “Like, ‘Amplify Damage: it increases how much damage you’re going to do.’ So we’d say, ‘Well, what is that? Is it a percent? Is it for normal damage? All damage? How about damage over time? How does all that work?’ None of those questions were filled in by a designer. Those were all programmers going, ‘Well, this is what seems reasonable to me, so I’m going to do this.'”

Above: Diablo 2’s Barbarian skill tree

Diablo 2’s five heroes — Amazon, Paladin, Necromancer, Sorceress, and Barbarian — ended up with three skill trees apiece. Each tree, organized in a separate tab, contained 10 skills for a total of 30 per class, all unique to that hero. Every time players leveled up, they earned a single skill point. Trees were arranged from top to bottom, with early-game skills closer to the top and the most advanced skills at the bottom, introducing a form of linear progression that complemented the seemingly arbitrary nature of item drops from monsters.

Most advanced skills called for at least one point in prerequisite abilities. Investing additional points into a skill increased its efficacy: greater damage, further range, longer duration, or some other effect, keeping early-game skills viable even as players learned better skills later on.

“One of the most awesome things about skill trees is that they show the player how he can grow his avatar — what his avatar is capable of becoming at the highest levels — right from the beginning of the game,” Stieg said.

Additional skill points could be earned by completing certain quests found in Diablo 2’s four Acts, and players could grow their characters to a maximum level of 99. That might seem high, but even factoring in quest rewards and leveling up 98 times — players start at level 1 — no player would ever earn enough skill points to fully upgrade all 30 of their skills.

On the surface, limiting players to the skills of their class, and forcing them to choose between maximizing a few skills or spreading their points between a dozen or more, could make Diablo 2 seem more limiting than the first game, where players could boost stats to wield any weapon or cast any spell regardless of their chosen class. In point of fact, the first game’s freedom often led to disaster.

“They’d pick the Warrior and go in and start casting spells,” remembered Dave Brevik, “then say, ‘This game’s just too hard.’ And it’s like, well, that’s because you’re not doing the thing you should be doing. So we figured, in Diablo 2, let’s take the ability for them to screw themselves out of the game by giving them something [more defined]. They would say, ‘Oh, I know what a Paladin is supposed to be’ or ‘I know what a Sorceress is.’ They could kind of understand what those classes represented to begin with. Then what they can actually do, and the choices they can make, are interesting and fun.”

Despite the linear structure of skill trees, the system was deceptively open-ended in a way that the first game’s freeform, one-class-fits-all spells were not. One summer vacation, I hauled my computer over to a friend’s house so we could spend a weekend playing Diablo 2. We each rolled a Barbarian, but our characters were dramatically different. He specialized in the character’s War Cries, buffs that did things like increase our skill efficacy and dig potions out of monster corpses. I chose to concentrate on combat disciplines. I was the heavy hitter, while he boosted our efficiency in battle using various cries.

Players and critics embraced skill trees when Diablo 2 launched on June 29, 2000, favoring the diversity they provided. Furthermore, the fact that players could not reallocate skill points — respeccing was added 10 years later in patch 1.13 — gave players more reason to create multiple heroes of the same type. A player may decide to create an Amazon who specialized in bows and crossbows, and then roll another Amazon to explore her javelin- and spear-oriented powers.

Over time, Diablo 2’s skill trees received some criticism, not just from the community, but from the developers who had crafted it. “Inevitably,” Dave Brevik said, “you’re going to have the munchkin people that run the numbers and say, ‘This is the official best build, it’s all over the Internet, and unless you’re playing this build, you’re a noob.'”

Dave saw min-maxing as an acceptable style of play. Just because a player had taken the time to do the math and determine the most optimal build for any character didn’t mean other, more off-the-wall builds were not viable. There’s a tendency among RPG players to fixate on optimization. A subsection of Diablo fans believe that any character unable to survive Diablo 2’s three difficulty levels (Normal, Nightmare, and Hell) holds no redeeming value.

Anecdotally, however, Blizzard North’s developers noted a tendency for most players to complete the game once, maybe twice with the same character, and then start over. They had no data-gathering or statistics tools, but they spent time on forums talking with others, and confessed to the tendency themselves. Just because a Sorceress who specialized in Enchant, a skill that added fire damage to any melee weapon, wasn’t equipped to survive Hell difficulty, where many monsters tended to boast immunity to fire, didn’t mean the player who ran the build couldn’t have fun playing through the game once on Normal difficulty.

Hoarding skill points, saving them up for hours until more advanced skills became available, was another proclivity. Stieg Hedlund saw no harm in it. “The whole point of skill trees done correctly is that a player can develop a strategy and pursue it,” he said, arguing that studying up to build an advanced type of character can be fun, too.

One such character is the tri-elementalist, a Sorceress who invests carefully in her Fire, Cold, and Lightning trees in order to use very specific skills in each tab. Creating it requires players to save up skill points until powerhouse skills such as Meteor and Frozen Orb become available. As they level up, players will have excess skill points to spend in each of their favored spells until they’ve burned through their supply.

Some players may find that tedious. For others studying up on a character class ahead of time, planning how to construct it, looking forward to unlocking specific abilities and searching for items that augment those powers, remains a hallmark of role-playing games.

“I think all of the skills were designed to be appealing, and were at some level, but pursuing a strategy sits on a higher level” Stieg continued. “If a first-time player comes to the game can’t find skills that sound interesting, then that would be an issue.”

There was another reason players hoarded points, one that Dave Brevik found more troubling. “It turned out a lot of people were paralyzed by choice. You wouldn’t think that would be the case. It was like, ‘Oh, the smarter thing to do is to save up points, spend the minimum number of points possible on required skills, then blow all the others on high-level skills.’ So there were some bad things about the system that needed to be fixed, but in general it seemed like a good idea.”

Above: Skills in Titan Quest

Diablo 2’s skill system, though flawed, introduced a revolution in action-RPG design and paved the way for different interpretation, many of which smoothed out its wrinkles. Diablo 3 unlocked skills at specific junctures and allowed players to swap skills in and out at will, removing the need for a re-spec option. Titan Quest, Iron Lore Entertainment’s well-received contribution to the action-RPGs in 2006, let players mash up any two skill trees from a total of eight, making for unique blends of abilities and nearly limitless replayability. Other games such as Path of Exile were designed in the vein of Dave Brevik’s original epiphany, offering dozens upon dozens of skills for players who enjoyed rolling up their sleeves and digging into character customization.

Diablo 2’s developers remain proud of the influence they’ve had on skill trees. Proud, and a little amused, when they reflect on how haphazard the development process could be at times. “There was nothing preplanned. We wanted to put cooler stuff toward the bottom, so we’d say, ‘Oh, yeah, that uh, that seems like a level-15 spell. Yeah,'” Dave Brevik recalled, laughing. “Eventually we came up with the theme of the different tabs and stuff, and that all kind of solidified. But it was a very iterative process.”

David L. Craddock is the author of Stay Awhile and Listen: Book II – Heaven, Hell, and Secret Cow Levels, now funding on Kickstarter in paperback and Kindle formats.