What if the economics of software have changed in such a profound and monumental way that nobody quite understands what’s happened yet?
I've been thinking a lot lately about the productive power of individual tech workers, and how this translates to the real world. It's pretty remarkable. And I think it's going to make good engineers a lot harder to hire.
Someone’s been telling engineers that they’re replaceable, but so far this hasn’t been the case, and I think that human decision-making is only going to become more valuable, not less, in the months and years to come.
The Flash contract reversal experiment has got me thinking about scale, and what can only be described as the radical asymmetric productive power of engineers in the modern era.
But at the same time, to realize how much power a well-organized developer has on a project like this, you need an anchor. Something to compare it to, otherwise none of this makes any sense. Macromedia’s a good one to think about.
In 1998, Macromedia employed 491 people worldwide. That included developers, product managers, marketers, administrative staff, technical writers, designers, sales teams, and one overworked customer support department.
Keep in mind that this was the company that brought us Flash, Dreamweaver, FreeHand, Director, and several other products that defined an era of creative software.
Part of what made these products remarkable was that even a small team of engineers would have struggled to recreate them. Certain thresholds of quality, sophistication, and ambition were simply inaccessible without substantial teams, financing, and organizational infrastructure.
And here I am, a month into this thing, having made major product and design decisions that might once have evolved organically over years, or decades, by myself, with three coding agents, a modest inference subscription budget, and a sense of nostalgia.
It's an interesting little experiment. And it’s been some of the most challenging AI work I’ve ever done. But the scale of the thing is so expansive and involved that it’s a project I simply wouldn’t have taken on as recently as late 2025.
Now, hypothetically, imagine that I can somehow finish the thing (actually feeling pretty good about that), and I decide to sell it as a commercial product.
What have I just accomplished in terms of scale?
In terms of production, actually quite a bit. But what about breakeven, and marketing?
Now, let’s further imagine that I wanted to be extremely selfish on the personal ask, and put an ideal salary somewhere in the area of $312,000 a year. Obviously, I don’t need that much, living in an affordable city in the Midwest, but it would be really nice.
My baseline annual operating expenses would be somewhere in the area of $900 for infrastructure, $100 in domain names, $1,000 for Apple development and distribution expenses, and $3,600 for inference costs.
That gives us an estimated annual operating overhead of $5,600, before salary, taxes, marketing, payment processing, and other incidental expenses.
Not awful.
If we include my extravagant salary requirement, we're looking at $317,600 a year before those additional expenses.
So now we know what our target and our baseline costs look like.
Cool. Allow me to continue.
Marketing is a challenge, but since I don’t need to make a billion dollars, I don’t need to spend a hundred million on conventional advertising.
In fact, I think that if I worked hard, used my social capital, got on a few fairly well-exposed podcasts, and did some strategic niche marketing to animators, creatives, and tech enthusiasts, I could find a market for somewhere between 11,000 and 15,000 copies of my product, sold annually mostly from my website, but also through distributors and affiliates.
That's a projection, of course, not a demonstrated market. But let's explore what those numbers would mean.
At $85 per seat per year, which is incredibly reasonable for an app like this, the base revenue is roughly $935,000 a year at 11,000 seats sold. At 15,000 seats, you’re looking at $1,275,000 in gross revenue.
Over three times my totally ridiculous salary requirement.
Fair to point out that historically, I’ve done better at higher price points for simpler software products, with less marketing effort. But let’s get cynical for a moment.
We also know that things don’t always work out. So the cynic needs to imagine that instead of selling 11,000 copies, the best I could possibly do was 6,660 copies.
We’re still at $566,100 in gross revenue.
And consider the other end of the equation. At $85 per license, I'd need to sell approximately 3,737 copies to cover my ambitious salary target and baseline operating expenses, before taxes, commissions, marketing, and other costs.
That's a very different kind of business problem.
Dismal, right? Even the best-case scenario is a drop in the bucket by the standards of billion-dollar company valuations, and selling only 60% of the original projection is bad. There’s no way I could support an entire operation of 491 people on that, right?
Exactly.
But for one person who didn’t even have to come up with an original idea?
If you can build sophisticated commercial software without the traditional organizational machinery, you no longer need the same number of customers, the same amount of financing, or the same definition of commercial success.
A product that would once have been considered a memorable achievement but an abject business failure could now support its creator quite comfortably.
All you have to do is build a product that's worth buying.
At some point, the hand of the market will tilt towards rational self-interest.
I spend a lot of time in agentic engineering circles. We're about a year ahead of everybody, and have been since the beginning of the 2020s. It's a good crowd.
And to most of us, this seems obvious. Either people are doing it and sharing their notes, or they're sitting there planning something along the same lines.
What's less obvious is what this is going to do to the job market in the near future, as the most qualified candidates start opting out.
And it's probably started already.
So be nice to your software engineers.
In strict hypothetical terms, you need them a lot more than they need you.