Conformance with the original game was always the highest priority when modifying the engine. Aleph One, the open-source version of the Marathon 2 engine, was invaluable to the project as a reference implementation that could be examined in a debugger (the original Marathon 2 had been written with CodeWarrior, but so long ago that the project file was no longer readable by any current version).

This was invaluable for ensuring the correct behavior of the monster AI, which picked up a number of bizarre quirks during the port. Monsters would rapidly twitch back and forth in places where the correct response would be to simply reverse direction. By comparisons with Aleph in the debugger, this was eventually traced to an error in renaming conflicting variables.

A pre-existing familiarity with the game’s levels came in handy when it was time to test. Knowing the location of all the secret areas, which tend to have tighter tolerances for movement and timing, made for much more efficient testing, as many problems could be caught by a coder during a cursory examination. Marathon allows for "grenade hopping", a sort of primitive form of rocket jumping, and many secrets require its precise use to clear jumps by inches; all of these had to still be reachable on the Xbox version.

Another benefit was that a large number of reports from testers could be dismissed immediately as conforming to the original game, even if they were unusual by modern standards, such as the bizarre behavior of monsters pelted with a steady stream of bullets- they'd smash into the wall, slide up until they hit the ceiling, and then hang their for several seconds before falling to the floor.

Our relationship with Microsoft. The Xbox Live Arcade team was an invaluable part of the project. In addition to being a ready conduit for technical questions, they provided us with many resources during the project. Implementing text flow in a way that supported every locale we wanted to target is extremely nontrivial, but we were given an internal library that provided it. When optimizing the save game screen, I was able to send off a profiler log for expert analysis. This direct assistance made development on the Xbox 360 a much smoother experience than it could have been.

If that wasn't enough, our account manager was available on MSN Messenger for most of the project. Questions about the certification requirements or some aspect of the 360 platform could be answered very quickly, and being able to simply converse with veteran Microsoft personnel gave us insight into the platform's nature and intended functioning. Uncertainty stemming from any of the 360's requirements never held up the project for long, and with the Arcade team's guidance we were able to bring the project in on time and without having to crunch for more than a handful of weeks. And only one person ended up sleeping in the office.