Unannounced Action MMORPG

Third-person shooter MMORPG built from the ground up on new tech and tools

Description

Since this project was "shelved indefinitely" and never officially announced, the details I can reveal are pretty limited. I cannot even confirm rumors or leaks that have been widely reported. What I can say is it was an ambitious project with a huge team of veteran developers building an action MMORPG in an entirely new engine, tools, and science fiction IP. Anyway, here goes:

[GAME] is a third-person, action massively-multiplayer online roleplaying game (MMORPG) based on the planet of [PLANET]. On [PLANET], in the sprawling city of [CITY], humans and aliens live in relative harmony, or at least that's how it might look on the surface. Underneath this facade, [CITY] is a city filled with various alien syndicates struggling in an endless battle for power. Inequality is widespread with the most powerful and influential enjoying the best the city has to offer while the lower rungs of society struggle to survive. Deputized by [NAME] the city's [ROLE], humans act as peacekeepers and investigators called [PLAYERS] charged with maintaining the balance of power between the syndicates in order to ensure the ongoing stability of the city. Like other Zenimax titles, [GAME] features deep storytelling and a compelling narrative driven by player choice with mysteries to unravel and secrets to uncover at every turn.

Responsibilities

I was responsible for coordinating development of combatant AI systems including decision making, action execution, awareness, weapons, and abilities, driving best-in-class AI combatants.

Notable Contributions

• Architected AI behavior system (utility-value model), enhancing decision-making depth and system scalability.

• Built AI awareness and combat systems, enabling responsive pre-combat and in-combat state transitions.

• Created flexible AI weapon systems using hierarchical state machines, empowering designers to quickly author AI weapons.

• Prototyped AI using Unreal Engine Behavior Trees & Blueprint scripting, supporting early exploration of potential game mechanics

• Mentored associate and mid-level designers with structured onboarding, improving retention and ramp-up speed.

Key Takeaway

Only simulate things when it is clearly discernable from randomness.

When I first arrived on the project, the AI systems had been designed and implemented by engineers without any input from design. The result was very intricate systems that allowed for a lot of nuance in decision-making but were extremely hard to work with. Those engineers were also no longer on the project, so we couldn't ask them what their original thinking was.

As such, much of my time was spent re-architecting the AI behavior system more or less from the ground up to reduce complex simulation and replace it with simpler alternatives wherever it made sense while reusing as much as possible of what had already been created. For example, the initial behavior selection design had the AI considering a variety of different factors and then selecting the best behavior at that time. This made the AI wildly unpredictable, changing behaviors frequently without clear reasons as to why.

Instead, I moved us to use random rolls based on fixed priorities set for the AI at tool time. The AI would make a random roll against all valid behaviors with extra weight applied toward the current behavior as long as it was still valid. The result was AI that still seemed intelligent, but were more predictable, easy to debug, and took a fraction of the time to create.

Videos

Since the game was never officially announced, there is no publicly available information or media related to it confirmed by Microsoft.

Images

Since the game was never officially announced, there is no publicly available information or media related to it confirmed by Microsoft.

Links

Since the game was never officially announced, there is no publicly available information or media related to it confirmed by Microsoft.