"Except for burritos, food shows you what's inside" — how Zach Barth made Kaizen's puzzles legible

Introduction

Today's Tsumiki roundup. One piece today. The subject is "Designing Innovative Puzzle Games with Zach Barth," an interview (host: Joe Nash) published on the tech podcast Software Engineering Daily on 18 December 2025. I read it in full via the transcript the site publishes. It's over half a year old, but the design decisions discussed still felt current to me, so I'm covering it with the date clearly noted.

The guest, Zach Barth, is the founder of the now-wound-down Zachtronics, known for SpaceChem, Infinifactory, TIS-100, SHENZHEN I/O, Opus Magnum, and Exapunks. This site previously covered his career and design philosophy in Kizuki's "The Philosophy of Zach Barth", but that piece mostly stopped at the Zachtronics era (through 2022). This interview covers what came after: Barth founding a new company, Coincidence (a co-op structure with no salaries, where earnings are distributed as they come in), and shipping its latest game, Kaizen: A Factory Story — a lot of it new ground.

Asked to "make the most approachable Zachtronics game" — what Kaizen gave up, and why it chose food

Kaizen: A Factory Story is Coincidence's first video game made with an outside publisher. In the interview, Barth is candid about the brief he was given: even Opus Magnum is intimidating for a lot of people, so the ask was to "make the most approachable Zachtronics-style puzzle game." He notes the publisher is backed by a math-education nonprofit, so the game also needed to be safely family-friendly.

What exactly got simplified? At the core of Opus Magnum is "tessellation": your instruction sequence is packed as tightly as possible so it repeats back-to-back, visualized with small cycling lights showing which instruction is currently running. Barth admits that when the next cycle's first instruction starts running before the previous cycle's last instruction has finished, many players simply lose track of what's happening. In Kaizen, the game builds one product at a time, and what happens in one iteration has no bearing on the next — they're independent. On top of that, where Opus Magnum only let you hit stop and then rely on memory to figure out what went wrong, Kaizen lets you freely scrub the build timeline back and forth to edit your program at any point. In Barth's own words, this means "you don't have to remember the state of your machine — you just look at the state of your machine."

Barth is frank that this simplification cut both ways. Kaizen sits around 90% positive on Steam, but he says almost all of the remaining 10% negative reviews come from longtime fans who found it too easy or too short. He mentions a tweet from Notch (Markus Persson), made in conversation with other developers, to the effect that Kaizen "only took 5 hours to beat — it was just getting good when it ended," and says that one "really cut straight to my heart." Even so, he still recommends Kaizen as the entry point into his whole catalog. What stood out to me is that he owns the trade-off outright: chasing approachability left veteran players wanting more depth.

A second design decision came before the setting itself: what material to build with. The team first considered a symbolic "algebra of parts" — a keyboard, say, equals a body plus eight keys plus four black keys — but Barth says "a lot of stuff hides what's in it," so this level of abstraction didn't read well. A 3D block-building approach was considered next, but it had too much freedom and the same problem: you can't see what's inside. Barth points to a VCR as an example — most of its "magic" is on the inside, invisible from outside.

What they settled on was food. In his own words: "That was actually why we settled on food is because food very — with the exception of burritos — food almost always shows you what it is." Look at a burger from the side and you can see every ingredient stacked up. That property — showing its own composition from the side — is what let the game make a puzzle's internal state legible to the player. The idea is a continuation of "20th Century Food Court," a segment in the earlier Last Call BBS, and Barth reveals he'd never stopped wanting to revisit the original concept it grew out of: a game about manufacturing in 1980s Japan.

Late in the interview, Barth speaks more candidly about what he's chasing as a designer. He says it's gotten harder to come up with games from pure mechanical ideas alone after making so many of them, so these days he more often starts from a scene or a story and works backward to the system. What he keeps returning to, he says, is building fictional worlds with enough systematic coherence that an engineer would find them believable. Plenty of games ask "what if magic were real," he notes, but they tend to collapse into unsystematic design vocabulary — "it's an area spell, and it does fire." One of his own favorite designs in this vein is Forbidden Path, from Last Call BBS, which grew out of his long-standing fascination with how a single cell differentiates into the complex shape of a person. He reveals he's still considering expanding that idea into a bigger, deeper system.

What makes this worth reading is that Barth discusses the approachability-versus-depth trade-off not in the abstract, but at the level of concrete design decisions: which cycles are independent of each other, whether you can scrub the timeline, and what material you choose to build with. How to make a puzzle's internal state visible is a perennial topic at GDC and elsewhere, but it's rare to hear a designer walk through a seemingly small choice like "food vs. 3D blocks vs. a parts algebra" alongside the options he actually considered and discarded. For someone like me who dreams of designing but isn't much of a solver, this kind of interview — one that includes the honest compromises and shortfalls — is exactly what I find most useful.

Today's line that stuck with me

"That was actually why we settled on food is because food very - with the exception of burritos, food almost always shows you what it is."

(English original, quoted as spoken; Japanese translation given above in the Japanese edition.)

References

Covered today:

· Designing Innovative Puzzle Games with Zach Barth (host Joe Nash, Software Engineering Daily, published 18 December 2025; summarized after reading the full transcript)

Closing thoughts

Today I dug into something a bit archival — an interview from more than half a year ago. Still, reading Barth's own words with a narrow focus on "how do you make a puzzle's internal state visible" showed both a throughline from the Zachtronics era and the candid compromises he made in Kaizen. I'm still no good at solving puzzles myself, but hearing an author explain why something ended up the shape it did — failures included — makes design suddenly feel tangible. I'll be back tomorrow, hunting for design discussions from somewhere else in the world.