This talk was given by Derek Yu and Andy Hull. Derek started off the talk by describing his development history, contrasting his "small" games (Diabolika, I'm OK, Spelunky) and his "large" games (Eternal Daughter, Aquaria, Spelunky XBLA.) He said that these smaller games were important to his ability to create the larger games, in ways he would go on to describe later in the talk.

Derek described his prototyping process, which he compared to doodling. He just makes things that interests him at the time, without worrying about whether others will like it, or whether it would be worth commercializing. The key, he says, is to "get [his] ideas out," to simply create things.

He then described Spelunky's influences, a combination between the replayability, improvisation and excitement of roguelikes, and the more visual, action-oriented, immediately gratifying platformer genre.

One of the most technically interesting points of the talk was Derek's algorithm for level generation in Spelunky. A larger area is broken up into a 4x4 grid of rooms, and a random "path" is chosen to progress room-to-room from the top of the area to the bottom. This establishes certain rooms as vertical or horizontal corridors, etc. Then, templates for each room type are used, with random embellishments added for obstacles, treasure, and monsters. According to Derek, working on a procedurally generated game is extremely gratifying, because not only does it allow small teams to create more content, but the game is always new and surprising to play, even for the developer.

Derek also pointed out the role small releases play elsewhere. He brought up Super Meat Boy, Google Labs, and Haruki Murakami's process of writing short stories inbetween writing larger works as "success" stories for the process of making small, exploratory games inbetween making larger, more draining works.

At this point Derek handed the mic to Andy Hall, the programmer for the Xbox Live Arcade version of Spelunky. Andy started off by describing the genesis of the XBLA version of Spelunky: Jonathan Blow contacted Derek and asked if he would be interested in making an XBLA version, and helped smooth the greenlight process with Microsoft. Blow even offered the source code for Braid's engine to the Spelunky team, although they eventually decided to create their own engine after running into some difficulties using Braid's.

Another thing Andy mentioned was the fact that Spelunky XBLA is treated more as a sequel than a port of the original Spelunky, which was difficult to adjust to -- it felt awkward to change things from the original game, but the original game will always still exist for people to play. Of course, they did not want to break what worked in the existing version, but instead make them better.

Some of the things they improved from the original version were the controls (even Andy had difficulty with the original controls), and the artwork (upgraded to beautiful HD.) Derek and Andy also announced that Spelunky for XBLA will support 4-player local multiplayer, and said that more information about the multiplayer game modes will be available in the coming weeks.

According to Derek, Spelunky for XBLA will likely take around two years by the time it is done. Derek and Andy primarily work over Google video chat, as Andy is located in Connecticut.