One of the more depressing parts of recent history is the centralization of the internet. When I grew up in the 90s and early 00s, the experience of the internet felt very different. You’d go to random people’s webpages, which were often no more than HTML and basic CSS. They’d post whatever they wanted. One guy would post stories where you read a page, and then clicked a link, like a Goosebumps choose your own adventure. You had genuine, long form forums - discussions with other community members that would span a year. Despite the technological infancy of these websites, the internet felt special.
Even email was somewhat unique. In the early days there were what felt like limitless numbers of email hosts. Hotmail, Yahoo, AOL, and Gmail were the big ones, but there were thousands of small ones. You could even self-host. Nowadays it’s pretty much… Gmail for personal, and Exchange for enterprise. So why did Gmail win? They provided the best UX. It was far better than the convoluted, ad-filled Web 1.0 experience of Yahoo and Hotmail. And so the internet aggregated to Gmail, centralizing email. In some sense email is not even really a messaging protocol anymore. It’s two or three SaaS platforms with integrations to each other. The underlying technology is almost irrelevant.
Despite the internet being fundamentally P2P, P2P loses a lot on the internet. In real life there are a lot of practical reasons why peer to peer can win. Locally sourced, organic fruit has genuine advantages over GMO foods shipped thousands of miles. It is riper, lacks delivery overhead, doesn’t have to be sprayed with chemicals to be suitable to transportation, comes from pollen and other sources in the local environment, and so on. But with the internet it’s the opposite. With the internet, the best product takes all.
I’ll say it again: digital technology, which has democratized building more than any other technology, ended up with the most centralized power structures anywhere you can find.
This is actually the central tenant of Zero to One. Products should strive to “go meta”: you don’t want to be the best email provider, you want your platform to replace email as a technological medium all together. YouTube doesn’t want to host videos, they want to become the video streaming platform. Adobe doesn’t want Photoshop and Illustrator to be tools for digital art, they want “editing photos” to become “using Photoshop”.
Another example closer to home - Valve. In the early 2000s Value was a forefront video game development studio. Half Life 2 was very prolific game, even now. But Valve’s games weren’t considered major commercial successes. So how did Valve become more profitable per-employee than Apple? They went meta. Instead of building the video game, they build a video game distribution platform. Now, instead of having to make games, they take 30%(!) of every game’s revenue.
Similarly, why do most video games look and feel the same today? Why is, I daresay, their ideation so similar? The answer is quite simple: because the tools that underly them are centralized. When every game is built on Unreal Engine using assets from Blender, the whole endeavour of game development becomes constrained to those tools. You’re not “animating a character”, you’re “using Blender”. You’re not “developing a video game”, you’re “using Unreal Engine”. When a product becomes a verb, e.g. “let me Photoshop that” - watch out.
And it’s a slippery slope. Because these tools are immensely powerful and make the impossible possible. There’s less of a reason to build your own game engine than ever before. Such development is costly. If you’re a developer who has tried to replace any sort of “essential” library in code with your own, you’ve probably learned this lesson. Most of the time you just build a buggy, undocumented version of the original tool. And game engines are the most extreme example of that.
But with every generation of tools, the users become a little less creative. A little more dependent. And the cost of making a software product like Unreal Engine your entire means to production isn’t simply vendor lock. In the age of AI, now you’re also the product of Unreal Engine - the training data for them to build an autonomous game maker.
I see two primary ways to navigate the world. Let’s call them qualitative and quantitative. Quantitative thinking has so far been dominant. People will work their life away for wealth despite it having diminishing returns on life satisfaction. People want high status, at cost to their personality and integrity. People want to become the “best” in the sense that has already been defined. This is quantitative thinking: solving the problem without asking ‘why’.
AI is really good at quantitative thinking. Arguably it’s the only type of thinking AI knows how to do. Businesses, too, are great at quantitative thinking. The board at Meta doesn’t stop to ask if Instagram is causing psychic damage to children and young adults. They don’t care. They want to know what Instagram’s market position is, and how it can be leveraged to reach even greater positions and ultimately make them more money. Developers like quantitative thinking because it allows them to focus on being the best at their role without considering a bigger picture. “I optimized {N} by {X}% and decreased costs by {Y}.” Specific, concrete. Engineers love this. Most developers don’t even consider what they’re building to be part of their job. They simply strive to be the best - the fastest developer, with the least bugs, the soundest architecture, the most succinct way to solve a problem - and so on. The best minds of our generation are creating new ad optimization algorithms and UX dark patterns.
Quantitative thinking is useful. There’s a reason it’s dominant. In everything we do, ensuring quantitative success is important. You can’t become a good developer - or a good anything - without quantitative thinking. But it can beome poisonous. When no one asks “why?”, we often optimize for the wrong problem to begin with.
Qualitative thinking is asking “why?”. And I think asking why is more important than ever. When you ask why, you begin to see that everything has hidden costs. You’re not simply a bystander using a product, you’re a soldier fighting for what the future of your creative segment looks like.
AI will increasingly have a part in our creative endeavours. That’s okay. But consider the steelman argument of AI - at the extreme, you’ll simply tell the AI you want a certain type of game in a certain style, and the AI will build it. Your involvement will be little more than a prompt. And while that would undoubtedly be cool, it denotes a point in history where you are no longer a creator. Look, I’m not saying everyone ought to build a game engine from scratch. Or hand-draw every frame in an animation. But at a certain point, especially with AI, the act of creation is lifted out of your hands. Soon, perhaps very soon, the question will genuinely be “how much part in this creation did I have?”. Is the purpose of a painter to produce paintings? Or something else?
And I’m certainly not propositing Luddite or Neo-Amish movements. Maybe nothing can be done, and the future of the digital technology is void of human creators. AI talking to AI, facilitating all of the aspects of life that we brought to the internet. Photorealistic movies generated by AI using data from our personality types and interests. Proceedurally generated infinitely expansive video games where the player feels genuine interactions with the AI bots that curate their experience. But mark my words - when this happens, certain genres, ideas, and game experiences will be lost to time. And at a certain point this begs the question: why do we create?