How a Scrapped Diablo 3 Enemy Idea Became a World of Warcraft Raid Standout

Not every discarded game mechanic is a bad mechanic. Sometimes, it is simply an idea built for the wrong pace, the wrong camera angle, or the wrong player mindset. A concept can fail in one game because players have no reason to notice it, then become memorable in another where every decision gives it room to breathe.

That was the path taken by one early Diablo 3 enemy-system prototype, which was ultimately judged to be "not fun" during internal testing. Rather than disappearing forever, the central idea was later adapted into one of World of Warcraft's raid encounters: the Anubisath Sentinels in the Temple of Ahn'Qiraj.

Former Blizzard game director Wyatt Cheng recently revisited the story in an Instagram post, offering a useful look at how developers can salvage a promising design after its original implementation falls short. The account is also a reminder that a mechanic's success depends heavily on context--not merely on whether the underlying idea sounds good on paper.

Embedded Instagram post

A set-bonus concept for monsters

Cheng said the original experiment dates back to 2004, when he was among the developers working on Diablo 3. The team was exploring an enemy equivalent to player item sets. In Diablo games, wearing multiple pieces from the same equipment set can unlock extra bonuses. The proposed monster version aimed to apply a similar logic to a group of enemies.

For useful background on this topic, read World of Warcraft: Forever Sets Class Identity Above Perfect Balance.

The basic prototype was straightforward: a pack could consist of four creatures, with each monster possessing an individual ability. If one monster died, its ability would be passed to the three that remained alive. As the group shrank, surviving enemies would inherit more powers, potentially becoming increasingly dangerous.

It is an idea with immediate tactical potential. Players might need to identify which enemy should be eliminated first, weigh the danger of leaving a particular ability active, or recognize that a seemingly simple pack would become much more volatile as the fight progressed. In theory, the mechanic could create changing priorities within an ordinary combat encounter.

In practice, it did not work for Diablo 3. Cheng's assessment of the prototype was blunt: "It was not fun."

Why the mechanic did not fit Diablo's flow

The issue was not necessarily that transferring powers between enemies was impossible to understand. Instead, the mechanic clashed with the way players commonly approach Diablo combat. The series is built around fast-paced action, dense groups of enemies, immediate damage output, and the steady chase for stronger loot. During many battles, players are focused on defeating targets as efficiently as possible rather than carefully tracking every creature in a pack.

Under those conditions, an enemy inheriting the abilities of a dead ally could easily go unnoticed. A player might kill several monsters in rapid succession without registering which powers had moved where or why the final survivor had changed. The intended tactical layer could be lost in the visual noise and momentum of an action RPG fight.

That distinction matters. A mechanic that asks players to observe, interpret, and react has to be presented in an environment where observation is realistic. If the game's rhythms encourage players to clear screens quickly, a subtle escalation system may not create meaningful choices. It may simply feel invisible, confusing, or unfair when it finally becomes noticeable.

Cheng's explanation points to a broader game-design challenge: systems are not experienced in isolation. The same rule can be compelling or frustrating based on encounter length, player perspective, visual readability, number of simultaneous threats, and the stakes attached to a mistake. A good concept needs a game structure that supports the behavior it is meant to inspire.

From Blizzard North to World of Warcraft

After Diablo 3 was put on ice, Cheng was transferred from Blizzard North to work on the rapidly expanding World of Warcraft. He was assigned to boss encounters, a role that gave the abandoned enemy-set notion a very different setting in which to operate.

One of the encounters involved four large opponents: the Anubisath Sentinels found in the Temple of Ahn'Qiraj raid. In this version, the key ingredients that had made the Diablo prototype unsuccessful could be reconsidered around the needs of an MMO raid group.

Rather than facing disposable enemies amid the constant rush of loot-driven action, players were confronting a clearly defined group of powerful foes. The Sentinels were substantial targets, and raid groups had time--and a strong reason--to study what each enemy could do. That made the transfer-and-combination concept visible in a way it had not been during the earlier action RPG experiment.

Cheng said the Anubisath encounter solved the reasons the set-monster approach had failed in Diablo. The fight also introduced randomness through its ability selection. Each Sentinel could randomly choose from a pool of eight abilities, meaning players could not treat every pull as an identical scripted problem.

Random abilities made assessment part of the fight

The random selection element was especially important. Because the Sentinels could draw from eight possible abilities, groups needed to assess the particular combination they had received. The encounter was not only about executing a fixed strategy; it required players to look at the threats in front of them and determine how to respond.

That is a major contrast with the original prototype. In a raid encounter, players are already primed to inspect enemy abilities, coordinate with teammates, assign roles, and discuss kill order. Tanks, healers, and damage dealers each have distinct responsibilities, while the longer structure of a boss-style battle allows the group to recognize escalating danger. The mechanic becomes part of the encounter's language instead of being buried beneath rapid enemy clearing.

The four-Sentinel setup also gave the concept a clearer dramatic arc. Every defeat could alter the remaining group, and the consequences were more likely to be understood because the participants were few, prominent, and dangerous. Players could make informed choices about priority targets instead of simply deleting a crowd as quickly as possible.

In that context, the same broad idea--enemies gaining access to powers as allies die--could create the kind of tension it had originally sought. The player question was no longer just how quickly to kill everything. It was which threat needed to be handled first, what the current ability mix demanded, and how the group could survive the evolving encounter.

The importance of mechanical context

The story behind the Anubisath Sentinels is a useful example of iteration rather than simple rejection. Blizzard did not force the original Diablo concept into a game where testing suggested it was not enjoyable. But the team also did not assume a failed prototype had no value. Its core logic was retained, then rebuilt around a format better suited to deliberate reading and cooperative problem-solving.

That approach can be especially valuable for studios working across multiple series. Large developers often create prototypes that reveal both a mechanic's promise and its limitations. A system can produce poor results because of its original game, while still containing an appealing interaction worth revisiting elsewhere. Moving it to another genre or encounter type may expose the version players were meant to experience.

For Cheng, the lesson was that context can be the difference between a failed experiment and a standout encounter. The Diablo version asked a fast action game to support careful attention to shifting monster abilities. The World of Warcraft version placed that same kind of decision-making inside a raid, where deliberate analysis and communication were already central to the experience.

Temple of Ahn'Qiraj's Anubisath Sentinels therefore represent more than a memorable set of raid enemies. They are a small piece of Blizzard development history: an example of an early Diablo 3 concept being reconsidered, reshaped, and given a second chance in World of Warcraft. The underlying idea did not need to be abandoned. It needed the right game, the right encounter structure, and players with enough space to see why it mattered.