Pixels are free now
Well, not quite now, and not quite free, but soon and almost.
What does this nonsense mean? Before the internet happened, there were lots of businesses that were somehow predicated on the idea that certain kinds of distribution had high cost and friction. But when the internet appeared, distribution became "free" and any business that depended on that - they managed it (like retailers), amortized a bundled package of businesses with it (like newspapers), or took advantage of it to be gatekeepers (like the media industry) - was disrupted and reinvented. Many resisted, many failed to adapt and are gone, but the basic economics fundamentally changed, and the world changed with it.
Pixels still have high friction, just like distribution used to, but this is changing rapidly, and the end state will likely look a lot like the change that happened to distribution - they will be mostly "free".
What does it mean that pixels have "friction"? It means that anything that puts a pixel in front of you has fairly high costs now and lots of layers of a stack behind it - friction. It's easiest to see with digital images. A year or two ago, a pixel in a digital image had thousands of hours of, say, Photoshop engineering that went into the tool that the user then spent their own hours on - learning and then drawing. But now, almost anyone can get a generative model to generate those pixels (or "good enough" versions of them) much more cheaply. Those pixels are becoming "free" in the sense of low friction (and yes, I'm leaving out artist compensation here - it's a real and important issue that is separate from what we are discussing here. Even for artists though, pixels will become "free" as the tools get better with AI - and just like other digital media, the market will get bigger, and flatter. It's super easy to produce music or video now, much harder to get attention, etc).
This isn't just true for digital art though. Imagine an HR lead who gets asked to do a salary analysis. This used to be high friction and slow - get an engineer maybe or a data analyst, build connections to the data or sign up with a SaaS company that has built a bunch of expensive code to do that, and spend weeks on the task. But with AI, and things like code integrated to agents, even simple ones like ChatGPT, this task is much faster and easier - days or sometimes hours instead of weeks. The pixels of the final report just got much cheaper.
It's not terribly hard to imagine that "applications" or documents in a few years will be more like conversations - tell the assistant what you want, and it does the drawing, builds the UI for you. In fact, in the early days of GDocs, I would predict (easy prediction) that anything not connected to the internet would soon seem "broken" (this was pre smartphone, so it was surprisingly controversial). This is going to happen for something like "smart" - if your agent can't talk to an app or a business, if you can't ask a document to "show me a graph explaining this", if someone expects you to push pixels around a spreadsheet - it's all going to seem "broken". Very soon. Pixels will become free, and we will become irritated and then annoyed and then route around the places where they aren't.
(An interesting side question here - what are the primitives that this rests on stably? And how will applications "unbundle"? Albums broke away first from physical media and then moved into streaming, where we are more focused now on tracks as the primitive. Newspapers broke into multiple different businesses and markets, and many more small niches are now addressed by things like blogging (and other) platforms. What happens as applications unbundle from fixed UI, as operating systems become more like conversations, as databases can more easily handle ad hoc queries and combinations? What's the next settling point for each layer of the stack?)
There are many challenges here, and much to build. The point of this essay is not to cheerlead this change, but to call it out so we can think about it. It's not likely that we will, collectively, decide we want the friction to stay - that almost never happens in any economic activity, at least for long (sometimes gatekeepers hold it at bay for a while, in some domains, but eventually people figure out there is a better way and pressure builds). At best there will be niche markets like the current market in vinyl, but the mass market, the bulk of users, will always choose the easier and more convenient (which usually means cheapest) option.
Pixels will become cheaper over time. What are you building that might have a much easier way to be built, that's much better and more flexible for the user? What "toy use" or small niche market you serve has already started to figure this out, and build your disruptor?