At this weekend's IndieCade, Braid developer Jonathan Blow expanded upon his "philosophy of game design" he originally presented at GDC Europe, discussing how it works in concrete terms for puzzle design in Braid and his upcoming puzzle title, The Witness.

Blow described how he knew that he wanted The Witness to be a "lush world" in which you could wander around at will, but "when you came upon a puzzle, you knew instantly that it was a puzzle."

His original concept was for players to register a completed puzzle by drawing a design on an in-world touchscreen. He reviewed the thought process behind that specific design process, asking himself, "Do I let people draw in a freeform way, or do I constrain them?" before realizing that if it was freeform, players on different platforms would find it more or less difficult to register their solution based on their input method, be it mouse/analog stick or so on.

He then decided to give players direction by having them draw in existing grooves. But one supposed solution just led to more questions. "Making it so you could only draw in pre-determined grooves was an early decision, but that still didn't pin down very much," he said. "One answer I decided on was that there would be a very clear place to start [drawing], but then where does it end?"

Even deciding upon placing a set end point, Blow found further questions remained: could the line cross itself? If it could, could it cross points multiple times or circle spots? Did that matter, and if it did, did he need a way to register the difference visually?

"That was a big mess and it wouldn't please me visually," he said, and so he decided that the line could never cross itself. "It was very fortunate I found that early, and it led to a lot of interesting mechanics before I even knew what any of the puzzles were; when the line can't cross itself, the puzzle becomes about correctly drawing a partition, and that's something that came from exploration, not from a top-down decision about what I wanted the puzzles to be."

Completeness

Blow explained how this kind of exploration of design required completeness of design -- to think design decisions all the way through, and go beyond "fun." He related this philosophy to his experience developing Braid's time-manipulation mechanic.

"Braid is a platform game and a game like that has a relatively small number of things in it," he said. "Keys, monsters, platforms, doors... for the purpose of completeness, if you are asking what happens if some things are immune to your manipulation of time, you will get an incomplete answer unless you test it with everything in your world: what happens if monsters are immune, what happens if doors are."

"I care about completeness a lot," he said. "I will take it past the point where some people think it's a good idea. I don't agree that games have to be the most fun they can be; I will get rid of fun if it means I can get at something deeper or more complete."

Blow gave a specific example in Braid. "Many people don't notice this, but in World 4 [a world where running right runs time forward, left backwards] every time you pick up a key it is exempt from being rewound when you run left. That's to follow the player's expectation: if you pick up a key, you want to run right with it, left with it."

Blow however ensured that this remained consistent and complete as part of the Braid design, making all keys that were exempt from being rewound glow green as all time manipulation-free objects did, and then placing a puzzle at the end of the world where the key was not exempt, a point which he said "made sense" but could be "very difficult to work out for the player."

This kind of moment: of surprise that still works within the game's systems, is something Blow felt was "very valuable."

Delivering The Unexpected

"Surprise is sort of the same thing as the information content of a message," he said. "If you send someone a message and it only says exactly what they expected to get from you, then there's no information in that message."

Blow gave an example of surprise via game mechanics and systems from Terry Cavanagh's VVVVVV: the legendarily hard section called "Veni Vidi Vici" which, by subverting the player's expectation of what would happen if they left via the top of the screen, calling it "a surprise, and it's wonderful."

Blow said that to design a good puzzle developers should create a situation, examine what the "truth" of the situation is and use that. "Even if I suck as a game developer, and the design is not good, then it will still be good in some way, because the truth is valuable."

[This post originally appeared on Gamasutra and was written by Mathew Kumar.]