The Game Developers Conference is different. It has already been two weeks since game developers swarmed the streets of San Francisco, but GDC is always an enlightening experience. The developers who attend GDC are of course eager to tell you all about their successes. But they’re also willing to open up about the pains of making a game, in an environment where they typically find themselves nestled among their own kind.

Consider “The Failure Workshop,” a panel where 6 different designers broke down their biggest missteps. The most amazing thing about these admissions of failure is also likely the main reason these conversations keep cropping up: when given a chance to respond, the audience is supportive, showing a heightened level of respect for developers willing to admit they’ve met their match.

This approach isn’t limited to the small, indie developers. Valve’s Chet Faliszek and Erik Wolpaw, two of the writers behind , discussed at length the many different failed versions of Portal 2 that never saw the light of day. The pair revealed that their first attempt at a sequel to Portal threw out almost everything that made the first game great, including the game’s heroine, it’s iconic villain Glados, and even the physics-defying portal mechanic itself.

Much time was spent chasing dead ends. One of their most difficult tasks, for example, was effectively scripting the ability to drop Wheatley, the metallic orb that serves as your AI sidekick for the first portion of the game. The team recorded enormous amounts of dialogue and scripted collision detection for different ways Wheatley could be dropped, but it was all ultimately scrapped. In the end all of this work was abandoned: playtesters simply weren’t all that interested in the nuances of dropping an animated metal ball.

The writers went on to show four early builds of the game, including a pair of lengthy opening sequences. Both were hilarious, well animated re-introductions to the world of Portal. And both ended up on the cutting room floor, when they didn’t mesh with the overall game. Once again, the audience listened intently. Instead of jumping on the team for their missteps, the conversation evolved into figuring out how the team managed to learn from their mistakes so effectively.