Jonathan Blow mówi o Braid, The Witness & Order of the Sinking Star - Grysław 438

Summary

Jonathan Blow discusses his path from childhood programming to creating Braid, The Witness, and The Order of the Sinking Star. The interview focuses on how those games were developed, how their puzzles are designed, and why they encourage players to explore and think for themselves.

Early programming and game development

• Blow first programmed in a school class on a Commodore VIC-20 with 4 KB of RAM. His first remembered game was a simple timing challenge: stop a rapidly changing number when it reached a randomly chosen target.
• He later learned assembly language on a TRS-80 and moved to a Commodore 64, whose sprites and sound made it better suited to games. He recalls technical tricks such as reusing sprites during screen scanning and changing scroll offsets to create effects the hardware was not designed to support.
• Before the internet, learning how to make games depended on chance: books and useful advice were difficult to find. Blow encountered programming techniques through bulletin-board communities and game-cracking groups, but did not consider himself part of the formal demoscene.
• After studying computer science, he and a friend formed a small team and created Wulfram, an online science-fiction tank game. It used 3D graphics without dedicated GPUs and had to work over very slow modems. Blow describes the project as a major learning experience, though the team lacked business experience and he was still learning game design.
• He later worked as a contractor on larger games, including Deus Ex: Invisible War and Thief: Deadly Shadows, and contributed to Phase, a rhythm game for the iPod.

How Braid, The Witness, and The Order of the Sinking Star came together

• The basic idea for Braid emerged while Blow was on vacation in Thailand in late 2004. He wanted to make a platform game with time-rewind, then explored what would happen if some objects behaved differently when time was rewound. This quickly led to puzzle possibilities, including objects immune to rewinding and time affected by the player’s position.
• Prince of Persia: The Sands of Time was an influence, but Blow wanted rewind to be more than a way to avoid death. In Braid, time manipulation becomes a system players can use to solve puzzles.
• The Witness took about six and a half years to develop. Blow says he was exhausted by the end, but that experience helped him endure the even longer development of The Order of the Sinking Star.
• The Order of the Sinking Star took roughly ten years in part because it became a very large game, required a new engine, and served as a test of Blow’s programming language, J. Its premise centers on different worlds coming together and their mechanics interacting to create puzzles. Blow describes the overworld as part of the puzzle adventure, rather than merely a level-selection map.
• The game’s literary, sometimes archaic dialogue draws on works including Tennyson’s “Ulysses,” Milton’s Paradise Lost, and the Bible. Polish subtitles were not confirmed; Blow said localization decisions were being handled with the publisher. Physical editions were also being considered, but none had been announced.

Puzzle design and player guidance

Blow’s approach is to trust players to solve challenging puzzles without constant instructions, while making sure the game feels fair and deliberate. The opening needs to establish that players are in good hands. After that, an unfamiliar puzzle should encourage observation and experimentation rather than make players think the game is broken or expects them to read the designer’s mind.

Key points about the design process:

• Puzzles are built manually in a simple editor, using basic shapes for layout and collision geometry.
• There is no tool that automatically generates the intended puzzle structure. Designers have to find the version of an idea that is both workable and satisfying.
• Difficulty alone does not make a puzzle better. The team spent time reducing frustration and aimed for difficult puzzles to reward players with an interesting insight, a satisfying interaction, or a clearer understanding of the game’s mechanics.
• Larger “star” puzzles often mark the culmination of a sequence of ideas, rather than simply being the hardest puzzles.
• The game avoids level names that might spoil a puzzle or be weakly descriptive. Region names, by contrast, help communicate the mechanics and the theme of worlds combining.
• Blow did not use AI to design puzzles, judging that current AI was not useful for the creative work of finding a puzzle’s satisfying form.

The game’s size also gives players room to return later and rediscover puzzles they have forgotten. Its first ending is intended to feel complete: players need to substantially explore four of six possible world-combination areas, but can leave some of the map unexplored. More remains for players who want to continue.

The interviewer, Ryslaw, reports spending over 80 hours with the game and completing 3,310 puzzles at the time of the interview. He praises its difficulty and describes solving some puzzles after taking breaks, including one he eventually cracked after trying an AI-assisted search without success.

People and works featured or mentioned

• Featured: Jonathan Blow, interviewed by Ryslaw (Ryszard Chojnowski).
• Also mentioned: John Carmack.
• Games and works discussed: Braid, The Witness, The Order of the Sinking Star, Wulfram, Deus Ex: Invisible War, Thief: Deadly Shadows, Phase, Prince of Persia: The Sands of Time, Blinx: The Time Sweeper, Steven’s Sausage Roll, Heroes of Sokoban, Skipping Stones to Lonely Homes, Mirror Isles, Promised, and Infocom text adventures.