More Genre Coffin Nails

Clan lords and intertubes (Interactive Storytelling, Part 3 Addendum)

In this series’ previous entry, there was another important element in the demise of the adventure genre I should have discussed: the advent of the internet.

When there was no internet, if you were playing Sam & Max Hit the Road, and got stuck, there weren’t a lot of options open to you. The main one was calling LucasArts’ Hint Line and paying 1.99 USD per minute for help.

The internet was opened to the public in 1991, and by 1993, it started actually being used as such. America Online’s carpet-bombing strategy of distributing free trial disks came the same year. Quake allowed multiplayer online matches via the internet in 1997, a year before Grim Fandango’s release.

The problems this posed for adventure games were manifold: walkthroughs appeared online nearly as soon as someone solved the game, meaning tip-line revenue dried up. But these were symptoms of a deeper vulnerability. The walkthrough problem only matters if your game has exactly one path through it—a single chain of solutions the player either finds or doesn’t.

Adventure games weren’t just susceptible to walkthroughs; they were structurally defined as walkthroughs. The entire genre was a sequence of gates with a single key each, and once someone published the key ring, the game collapsed into a to-do list.

Most importantly, the genre was feeling long-in-the-tooth compared to other offerings and needed not just an evolution, but a revolution.

There were attempts to address this—mine among them. By the time I was designing quests like The Cairn Stones in Diablo II, I was very aware of the walkthrough problem. My solution was to use the procedural generation system already used to generate dungeons and many other elements of the game. In the Cairn Stones quest, the order of both the scroll’s solution and the stones themselves are randomized. Your solution won’t necessarily be the same as another player’s, and a walkthrough will do you no good.

By the time I was designing quests like The Cairn Stones in Diablo II, I was very aware of the issue. It’s for this reason I used the procedural generation already used to generate dungeons, and many other elements. In the Cairn Stones quest, the order of both on the scroll and the stones are randomized. This means the solution to your stones won’t necessarily be the same as another player’s and a walkthrough will do you no good.

The much more elaborate puzzles in the adventure genre, however, did not lend themselves well to such design solutions. Escape from Monkey Island included the “Monkey Kombat” subgame, wherein “monkey words” each defeated two others in rock-paper-scissors fashion. Which defeated which was randomized, requiring exploration to solve. But it felt tedious rather than fun and was one of the most hated elements of the game.

Both were the same strategy with different execution: make the solution non-transferable. Cairn Stones works because the randomization sits atop simple physical action—touch the stones in order. You don’t need to understand why the sequence works, just that it works. Monkey Kombat fails because it asks the player to reconstruct the entire logic table through trial and error, which is tedious with or without a walkthrough.

But even the good version of this fix didn’t touch the deeper problem. I felt actual puzzle gameplay was outside what made sense for Diablo II, so Cairn Stones was really just lock-and-key—get the scroll, have it translated, and use it on the stones. Even still, a player could just keep clicking the stones and hit on the solution fairly easily. Randomizing the solution was a tactical fix for a structural vulnerability. The genre didn’t need unpredictable answers. It needed better questions.

I had to face that structural problem head-on when I joined the strike team for WarCraft Adventures: Lord of the Clans (WCA).

In Blizzard’s parlance of the time, a strike team was a group from one studio tapped to look at another’s first playable, give feedback, and make recommendations. I did this for StarCraft, then joined the team to continue in this role right up to its successful launch. WCA was less fortunate.

While Diablo revitalized a supposedly dead genre, we did so by making significant changes to it, removing the P&P RPG cruft in favor of a quick-to-the-fun experience that still had a ton of depth to explore. WCA brought nothing new to the table. One reviewer of a leaked section of gameplay hit the nail on the head, calling it:¹²

[A] conventional, borderline dull point-and-click adventure [… that] doesn’t quite feel like Warcraft.

That second clause is the sharper observation. It didn’t feel like WarCraft because the adventure genre’s core verbs—find the item, combine it with the other item, use it on the thing—are fundamentally wrong for the IP. WarCraft’s native verbs are faction navigation and political violence. WCA as a story is about Thrall moving through systems of power: slavery, gladiatorial combat, clan politics, the formation of identity under pressure. That material wants dialogue, alliance-building, and consequential choices—not “use rope with bucket”.

Every time the designers reached for a traditional adventure puzzle, they were translating a political story into a mechanical language that couldn’t carry it. The adventure verbs atomize the world into discrete inventory objects; Thrall’s story is about navigating webs of loyalty and coercion. These are not complementary impulses.

Despite Blizzard’s protests to the contrary, the playable really was that bad. For these reasons, I lobbied hard for the game’s cancellation. I was pleased to have eventually convinced Blizzard’s leadership a mercy killing was the best course of action.

But the mercy killing, correct as it was, raises a question worth sitting with: what would an internet-native adventure game actually have looked like in 1998?

Not randomized puzzles—that’s the tactical fix again, treating the symptom rather than the disease. The structural fix—like Diablo II’s skill system—would have been to make each player’s experience genuinely unique, so that walkthroughs merely describe someone else’s game. Not unpredictable answers, but a possibility space where the path itself varies.

This isn’t hypothetical. The same year WCA was cancelled, Planescape: Torment (PT) shipped. It wasn’t an adventure game in the LucasArts mold, but it demonstrated the shape the genre needed. It was dialogue-heavy, systemically branching, and walkthrough-resistant not because its puzzles were randomized, but because your Nameless One wasn’t mine. Your choices, your stats, your conversational paths through Sigil produced a version of the game that no FAQ could fully map. The genre’s evolution wasn’t about making the lock harder to pick. It was about replacing the lock with a conversation.

WCA had an IP that could have supported exactly this structure. Thrall navigating orc clan politics, earning trust or forcing compliance, building an identity from nothing—that’s a branching-dialogue game, not a puzzle-chain game. PT proved the format could work. The WarCraft IP had the characters, the factional complexity, and the emotional stakes to fill it. What was missing wasn’t the opportunity but the willingness to abandon the adventure genre’s assumptions about what a game in that space should look like.

I’d have enjoyed working in the genre and taking on the challenge of innovating it for the internet age. But by the time the strike team was done, it was clear that WCA wasn’t trying to innovate—it was trying to execute a form that had already expired. We made the right call. What died with it was the chance to show that the adventure genre could mutate rather than just go extinct.