There are potential spoilers ahead for anyone who hasn’t played through Portal 2 yet and would still like to. During a Portal 2 post-mortem at GDC, Valve’s Eric Wolpaw and Chet Faliszek explained the importance of testing new ideas, even stupid ones, throughout development. Eurogamer has the scoop on what ideas didn’t make it into the game and how others were created through trial and error.
Wheatley, the personality sphere who transforms over the course of the game from being Chell’s companion to antagonist, was originally supposed to be destroyed very early in the game. The original plan called for the player to meet upwards of seven different personality spheres, with Wheatley just being the first.
One of these spheres was lovingly referred to as the Morgan Freeman sphere, and would have been incredibly wise regarding the 20 by 20 space that the player finds him in. Once moved outside the safety of his room, “his mind was blown and he was pretty much useless. Although as the game progressed, he eventually got his feet under him and started delivering some homespun wisdom that all related back to this 20 by 20 space.”
There were other spheres as well, but testing found that players actually missed Wheatley, and none of the other spheres were around long enough to bond with. The idea of multiple personality spheres was scrapped and they eventually decided to stick with Wheatley, “Our simple answer was, let’s just kill Wheatley, and bring him back” Faliszek said. It’s a good thing they did, without bringing him back, Wheatley would never been able to grown as the game’s main antagonist.
Player testing also helped the Valve refine their single and multiplayer experiences, the two said. It specifically helped them to identify the difference how players act within a single player and then a multiplayer experience. They found players operating in single-player tend to “watch intently and pay genuine attention to the crafted experience.” The Co-op experience, however, is very different. Faliszek says “Co-op players, on the other hand, will interrupt your best material to ask each other what they had for dinner.” This data helped them to craft single player and multiplayer experience that would complement the level at which players experience the game.
They also tested a group of smaller along-the-way endings, such as the fire pit from Portal 1, which to some was the end of the game itself, “A small percentage of playtesters were just fine with riding into the fire pit, that was a good ending for them,” says Wolpaw. In Portal 2, developers experimented with a few other ways to end the game early, “Chell would die and that would be the end and we’d play a song, and if you wanted to you could just quit there,” says Wolpaw. They experimented early with some of these moments; one has you dying within the first 2 minutes of the game. Eventually, through testing, they decided to cut the endings because they felt it detracted from the game’s true ending, making it to the moon.
Finally, the two ended their talk by outlining their biggest lesson from Portal 2, “”One of the biggest lessons we learned is try to give yourself enough time for the obvious to become obvious, because at a certain point, most of the answers to what you need to do are buried in what you’ve already done,” Wolpaw said.