Wednesday 29 June 2011

Interview: Zachtronics Industries on Dynamic Puzzle Solving & Trout

Zach Barth operates under the name Zachtronics Industries. He's responsible for the outstanding puzzle game with the so-so name, SpaceChem, which is probably the best puzzle game I have ever played. It's a molecular programming game whose solutions are dynamic, that never fails to respect your intelligence, and is impossibly challenging without ever turning you off. It tests logical problem solving in a far more immediate and mature fashion than my own attempt. On top of this he's infamous for developing Minecraft precursor, Infiniminer (which we're not going to talk about today because no doubt he's bored). Anyway, grab the demo and be in awe.

Hi Zach. I realised as I planned this interview that you're the first person I've talked to for Plot is Gameplay's Bitch who might not list 'writer' as their first job. On top of this, I can't claim puzzle games are regularly at the top of my play list. Why do you think SpaceChem has captured me - and other players - so thoroughly?

I think that the allure of SpaceChem for many players is the way that it approaches puzzle / problem solving. Instead of requiring you to reconstruct my contrived solution, every puzzle allows for a very large number of possible solutions, each acceptable but with different performance characteristics. This gives players a clear goal, but an unprecedented amount of room to be creative, innovative, and iterative in creating their solutions.

I usually found that I could come in above average in play cycles (time efficiency of the solution), but way over the top on symbols (the complexity of the solution). Maybe I'm trying to be too clever for my own good, or maybe I'm just not as smart as I think. I'm fascinated by the metrics you collect on players - any interesting insights? Are there traits certain types of players share?

One of my favourite experiments we’ve run is SpaceChem: An Average Solution, where we used some clever visual post processing to average out hundreds of solutions for the “research” puzzles in SpaceChem. It allows you to see how solutions for different kinds of levels (tutorials versus simple levels versus hard levels) converge and diverge based on the constraints of the puzzle, which I think is very neat.

With regard to how players respond to the graphs, I think that many take to them the same way that you do. The first time you beat a level you’re almost certainly not at the top, but the graphs invite you to find a dimension of optimization that interests you and improve it in that aspect. The fact that the dimensions are mutually exclusive helps to make players who choose to optimize feel awesome, as graphs are filled both with people who wanted to optimize that aspect and people who chose not to optimize that aspect to optimize another.

You've been working on this basic programming play through a series of games, but SpaceChem displayed an unprecedented level of polish in its visuals, soundtrack and narrative. Do you want to shout out to your contributors?

Without a doubt!

I had two programmers, Collin Arnold and Keith Holman, who handled a majority of the coding tasks and freed me up to focus on tasks related to design and production. The remaining tasks were outsourced to talented individuals: Ryan Sumo for the majority of the artwork, Ken Bowen for sound effects, Evan Le Ny for the soundtrack, and Hillary Field for the narrative. Without them, there would be no SpaceChem!

You've spoken previously about how the emergent nature of the solutions is a strong point of engagement, and I think that's what interests me here where point & clicks or LA Noire (for example) fail: in building a solution within the game logic, rather than just discovering the pre-authored one. Is that an approach you'd consider expanding to subjects other than molecules and engineering systems? Narrative linearity seems jarring by comparison.

Yes, definitely, although I think it’s far easier to apply the principle to games like SpaceChem that are highly mechanics-driven. If I had to explain why this was the case, I would guess that it’s because this kind of system fundamentally expects and requires the player to design something as a solution, which translates into a strong sense of agency, which in turn drives engagement. I’m not sure how it would work in an interactive story like LA Noire, but I certainly don’t think it’s the only route to take.

Finally, what's next for you? Can you spill any details that I can spin into a headline? "SpaceChem dev turns hand to Facebook trout farming sim" would be ideal.

I wish I could tell you something that exciting! We’re focused right now on the transition to being a full-time studio and wrapping up SpaceChem development. Trout farming does sound kind of interesting, though…

Thanks for your time.

You can check out all Zach's games at

1 comment: