Sunday Service: John Carmack on Antifragile Ideation
Discount great ideas heavily. Put a premium on execution. Expose critical flaws through implementation.
š About Sunday Service: This is a break from regular postings. I use it to share something on a Sunday that I learnt from other inspirational figures in the week.
The most popular video game category today is the first-person shooter genre. It was invented by John Carmack, arguably one of the most talented software engineers of our time. His early work on Wolfenstein and Doom paved the way for games like Call of Duty and Apex Legends.
Carmack is a rare breed of talent. Heās recognised alongside influential programmers like Linus Torvalds (inventor of the Linux operating system), Sir Timothy John Berners-Lee (inventor of the World Wide Web) and Ada Lovelace (author of the first computer program.)
Whenever I find a talk or presentation by Carmack, I pay attention and try to learn what I can, even if most of his technical ideas are beyond me. This talk he gave at Facebook in their āDeep Thoughts Engineering Speaker Seriesā is one such example.Ā
In the presentation, Carmack runs through various computer graphics problems to illustrate how he works with ideas. But if you abstract away the domain-specific examples, itās clear that his overarching philosophy about ideas is relevant to venture capitalists and tech entrepreneurship. Here are some of the key concepts I took from the talk.
Donāt invest in ideas. Invest in the process.
āThe ideas really arenāt the important thing,ā Carmack says. āItās the execution where a good project is a result of hundreds and hundreds or even thousands of good decisions that are made along the way.ā In some ways, if you relate this back to the broad idea of investing in founders and companies, you find that the art of venture isnāt necessarily about investing in ideas. Itās about investing in a process ā a process of discovery.
Carmack still believes in the power of great ideas. However, sitting in an armchair and pontificating your way to great ideas rarely works. In Carmackās experience, the more reliable path to interesting ideas is to hack together prototypes, iterate, and learn as you go along.
This mentality is crucial in pre-seed and seed-stage investing. Early-stage founders often have great enthusiasm for a business āideaā, but the ones that get ahead have a bias for building and testing things and learning through direct contact with potential customers.Ā
Carmack puts it this way: āYou get the ideas by being down in the mud working on the problems ā itās the hard work that leads you to the insights that are hopefully the great ideas.ā In other words, founders who hack together rapid prototypes to test with customers as early as possible do better than entrepreneurs who spend months or years ideating things that people might not want.
Make contact with reality to expose critical flaws.
As a junior investor, I used to fall in love with startup ideas that sounded great in theory but were yet to be tested. Experience has taught me to apply a significant discount to these opportunities. Without meaningful contact with the real world, the brilliance of an idea can mask critical flaws.
Carmack uses a psychological hack to protect himself from getting too attached to a promising idea. From the outset he ascribes a low value to ideas so that he can consider them more critically. āIāve been through the idea train enough times now,ā he says. āI get the idea high at the beginning, but Iāve done it enough times now that I can say, well, it probably wonāt [work], and it can almost be a puzzle game then to find out how to bust my own idea.ā
This is the approach Carmack took when he considered a clever idea to improve the performance of Minecraft in virtual reality. Usually, the iconic blocks of the game are drawn on the screen dynamically, depending on where a player is looking. Carmack wondered if he could boost performance by replacing this system with a fixed order for drawing the blocks. He thought the gameās cubic structure might not need the dynamic ordering typically required in 3D scenes. However, when he started to implement the idea, he found that even simple cubes could still obstruct each other in ways that a fixed ordering system couldnāt resolve. An exciting idea unravelled the moment it met execution.Ā
There are many examples of this in tech that serve as epic cautionary tales. Quibi raised $2 billion before launch but shut down within a year of market exposure since no one really wanted another short-form video app. The Segway scooter cost $100m to develop but failed to revolutionise urban mobility because it was impractical. Juicero raised a similar amount in VC funding but crumbled when users realised they could squeeze the juice packs by hand. These companies could have avoided failure at scale had they made contact with reality sooner rather than later.1
Cultivate enthusiasm for trial and error.
VCs must be able to identify founders who are deeply committed to building and learning quickly while maintaining enthusiasm through what can be a protracted journey of ideation.Ā
John Carmack personifies this philosophy. āI can pat myself on the back when I break my own ideas. Iām very happy with showing clearly that this doesnāt work,ā he states. This was the same attitude that Thomas Edison had on his way to inventing the first practical light bulb, as his associate Walter S. Mallory recalls in the passage below:
I found him at a bench about three feet wide and twelve to fifteen feet long, on which there were hundreds of little test cells that had been made up by his corps of chemists and experimenters. He was seated at this bench testing, figuring, and planning. I then learned that he had thus made over nine thousand experiments in trying to devise this new type of storage battery, but had not produced a single thing that promised to solve the question. In view of this immense amount of thought and labor, my sympathy got the better of my judgment, and I said: āIsnāt it a shame that with the tremendous amount of work you have done you havenāt been able to get any results?ā Edison turned on me like a flash, and with a smile replied: āResults! Why, man, I have gotten a lot of results! I know several thousand things that wonāt work.ā
If youāre curious about John Carmack and his views about ideas, check out the full talk here. I also enjoyed listening to his interview with Lex Friedman, which was more accessible from a technical perspective. However, itās a five-hour podcast!
Grocery delivery apps suffered a similar fate after they found that few people are willing to pay a premium for delivery of small items like milk or bananas you can get from a short walk to a convenience store.