A couple weeks ago I joined a conversation about John Romero and prototypes. Tyler posted some thoughts about Romero’s autobiography, Matthias shared a quote from a Tim Ferris podcast where Romero chided prototypes, and Matthias looped me in because I love prototypes.

No prototypes. Just make the game. Polish as you go. Don’t depend on polish happening later. Always maintain constantly shippable code.
– John Romero, Strange Loop 2002

A quick internet conversation changed the course of my week a bit. After that short exchange I dove in and listened to podcasts, watched a handful of talks, and read an entire autobiography… and I’m still shocked that John Romero doesn’t believe in prototypes.

id Software is synonymous with speed

To understand Romero’s perspective you need to understand the ethos of id Software. id Software was –to borrow a gaming term– a “Press W” company, meaning they tend to plow through all obstacles to keep forward momentum. As Romero recounts in his biography, they prioritized speed above all other design principles.

  • Speed in gameplay
  • Speed in development
  • Speed in using the latest tech
  • Speed in heavy metal MIDI soundtracks

Romero and Carmack were also prolific developers. Romero has over 100 titles to his name and in 1991 they released 15 games! And they did this while simultaneously pioneering 3D video game graphics for Wolfenstein 3D (1992) and DOOM (1993). How did they maintain such velocity?

  • Good division of labor. Carmack worked on the game engine, Romero on the level design, and Tom Hall did the art.
  • They lived together. I’m not advocating for workplace cohabitation, but I think you need to acknowledge the contextual environment.
  • Knew the hardware. Carmack and Romero were intimately familiar with the limitations of the hardware they were developing against which made porting games to other platforms (Commodore, Amiga, etc) a lot more predictable.
  • Always rewriting. They were always rewriting using the latest tech. Take an existing title and use the latest graphic capabilities, take an existing title and make it with the latest engine, rinse and repeat.
  • Built internal tools. Romero built internal tools to speed up level design: TEd (a text-based level editor), DoomEd, and QuakeEd. Romero frequently talks about making DOOM levels in a single night.
  • Predictable scope. While the underlying graphics or physics engine might change dramatically, the formula for the scope of the game (number episodes, levels, weapons, and types of enemies) stayed nearly the same so you could repeat processes, reduce code churn, and avoid surprise hardware limitations.

That formula gave the team a lot of shots on goal. To use another sports metaphor, that’s a lot of reps. id Software was crunch time all the time. Abiding in that breakneck thrill ride enabled them to create groundbreaking, industry-defining titles like DOOM and Quake.

Back to prototypes and the lack thereof…

Okay. Back to the main plot. John Romero doesn’t like prototypes…

“No prototypes. Just make the game. Polish as you go.”

Speed is great, but process effects outcomes. In Romero’s own words “Don’t depend on polish happening later” which is something those of us in tech know all too well. You’re probably going to run up against “the Iron Triangle” at some point.

I’m not trying to throw shade – I will never be as successful as id Software – but here’s a side-by-side of two similar side-scroller games.

One is a game from a company that prioritizes prototyping and using withered technology (Super Mario Bros, 1985), the other is a clone of that game from a company that prioritizes shipping quick on the latest hardware (Commander Keen, 1990). This isn’t an attempt to strip the game of its technical achievements, bringing Nintendo-like games to the PC, but there’s a palpable difference between the two; one is timeless and one is timely. While a myriad of skill, staffing, and technology factors contribute to the difference1, I believe the biggest delta comes down to pressing the publish button too early.

There’s a famously misattributed quote that seems apropos here…

“A delayed game is eventually good. A bad game is bad forever.”
– Siobahn Beeman, often misattributed to Shigeru Miyamoto

Who cares if some “not spectacular” graphics slipped out the door? That game, Commander Keen, was successful. id Software was successful. Over and over. From my perspective that comes down to the sheer volume of ideas they shipped, that continuous practice, and their novel shareware distribution model coinciding with the explosion of the internet and the PC. You need to be responsive and reactive to that wave and if there’s a commitment to release a new product every month, then I wouldn’t recommend a thinking tool like prototypes under those limitations either.

Do I –without evidence– believe they actually did prototype a bit? Yes. I think in the sense of experimenting with level design or evaluating community mods, they could cheaply explore new ideas with low effort thanks to their level editor tools. All those throwaway ideas amount to some level of what I’d call prototyping. But I take Romero’s point that they didn’t believe in prototyping, they were working on the product, not an idea, and could have hit publish at any point in the process.

So I’ll concede that maybe with a pair of once-in-a-generation talent, a perfect mixture of group workaholism, a breakneck cadence, all living in the same house, a convergence of new technologies, quality internal tools, the explosion of the home computer industry, an original and extremely violent IP that appealed to the Metallica generation, and over a hundred shots on goal… you too might be able to capture lightning in a bottle and create a smash hit like “the entire concept of 3D video games”…

… but for the rest of us there’s prototypes.

  1. A bit more on development timelines. Super Mario Bros was 7 people and took 5 months building on systems built in previous games and Commander Keen was 4 people and took 3 months.