Braid

Blow saw himself as a curator presenting answers the engine produced. He describes Braid's puzzles as showcasing the unexpected consequences of the time-travel engine, with his role being to clean up those answers and present them to players.

Quote: "I had a curator role, cleaning up the answers and presenting them in such a way that they could be enjoyed by the game's players."

SpyParty made player-discovered exploits official parts of the game. Chris Hecker found players discovering exploits and unintended tactics, so he adopted them and pushed the game toward mind-games, a parallel to Blow curating what Braid's engine produced.

Apply a rule to every object in the game, including the hero. Once immunity to rewind existed, Braid applied it to enemies, keys, doors, clouds, platforms and the player character, each producing new puzzles.

In The Witness Blow asked the same question of the grid, cells, line, environment and panel. Arvi Teikari of Baba Is You cites this exhaustive approach from Braid as the philosophy he follows.

Baba Is You turned its rules into objects to explore every interaction. Arvi Teikari made Text a word so rules can manipulate other rules, and Level lets players move onto the map screen, following Braid's explore-everything approach.

Other words include Empty, Make, More and Not, each pushing the core idea further to surprise veteran players.

Playtest each new rule and turn surprising side effects into puzzles. After adding a rule, Blow played the game looking for consequences he had not foreseen, and each puzzle illustrates one of them. A time-immune moving platform, for example, lets the hero rewind to a point where it is no longer under his feet, so he falls when time resumes.

The aim is that the player stumbles on the same fact the designer found. Quote: "each puzzle became an illustration of one of those phenomena."

A solution available at only one narrow moment feels like a trick. One Braid puzzle needs an enemy to bounce off your clone's head so you can bounce off the enemy. It follows from the rules, but it can only happen at one moment, so players cannot experiment toward it.

Mark Brown uses it as a cautionary example: "But for many, it felt more like a trick than a revelation."

Braid's rules grew out of coding one mechanic: unlimited rewind. Jonathan Blow started Braid with the ability to rewind time by almost any amount. While implementing it, he realized he could exempt certain objects from the rewind, and that became a core rule.

Blow says more ideas came out of development and ended up in the final game than he put in as a designer. He applied the same process to The Witness, where play-testing line drawing showed he kept partitioning grid cells, which became a rule. Quote (GDC 2011): "The process of designing the gameplay for this game was more like discovering things that already exist than it was like creating something new and arbitrary."

Playtest each new rule and turn surprising side effects into puzzles. After adding a rule, Blow played the game looking for consequences he had not foreseen, and each puzzle illustrates one of them. A time-immune moving platform, for example, lets the hero rewind to a point where it is no longer under his feet, so he falls when time resumes.

The aim is that the player stumbles on the same fact the designer found. Quote: "each puzzle became an illustration of one of those phenomena."

Apply a rule to every object in the game, including the hero. Once immunity to rewind existed, Braid applied it to enemies, keys, doors, clouds, platforms and the player character, each producing new puzzles.

In The Witness Blow asked the same question of the grid, cells, line, environment and panel. Arvi Teikari of Baba Is You cites this exhaustive approach from Braid as the philosophy he follows.

Misdirection stops players brute-forcing past the interesting fact. In the puzzle Hunt you must kill all the monsters, but killing them in the most obvious order makes it unsolvable.

The failed attempt shows the player why it fails, which is often the truth the puzzle illustrates. Without it, trial and error could solve the puzzle without the idea landing.

Reprise a layout under a new rule to show what changed. A simple puzzle about unlocking two doors with one key is paired with a bigger version nearby, and the level appears in both Worlds 2 and 4, where each world's time rule gives it a different solution.

Quote: "by using familiar layouts in different worlds, with different rules, you can see how the consequences have changed."

Break the habits players have built, with traps for careless thinking. In the level Irreversible, players must realize they cannot use their rewind powers, and time works so that only one of the gates can be opened.

Cut rules that lack surprise, overlap, or say nothing. Blow dropped Braid's turn-based world because its consequences were not surprising or felt contrived. Braid and The Witness were themselves spin-offs of shelved games whose core mechanics did not offer enough to explore.

Mechanics, rules and puzzles that overlap with each other are also removed.

Keep a consequence that is interesting even if it isn't fun. Blow kept a very odd puzzle where a key bumbles along on its own, because it is a real, surprising consequence of the game's universe.

He also leaves in things whose removal would make the game feel incomplete, which sets him apart from many designers.

Treat each puzzle as an idea passed from designer to player. Solving a Braid puzzle is meant to feel like noticing something that was there all along, unlike a Rubik's cube or a riddle with a guessed answer. That makes the puzzles feel fair.

Blow told Gamasutra that the more a puzzle is about something real and specific, the more meaningful the epiphany. Quote: "a puzzle is never just a puzzle. It's a communication of an idea from the designer to the player."

Teach every mechanic upfront, then only recombine them. Braid introduces all its elements early with quick introductory puzzles. Harder puzzles change set-ups, combinations and layouts rather than adding new tools.

Most puzzles explore one idea in stages small enough to hold every moving part in mind, with few red herrings or arbitrary steps, so executing the solution is relatively effortless once seen.