We're Going to Make Out Like Bandits
We're Going to Make Out Like Bandits
April 12, 2026
tech
ai
Here’s the plan. I reckon it’ll take about five years in all, and I<br>think we’re about 1.5 to 2 years in already.
Our starting point is that the AI models are now good enough at<br>coding. Not necessarily “perfect”, just “good enough”. Often, they’re<br>as good (better!) than a junior developer. I don’t know about you,<br>but I’m finding the new models pretty impressive. So are other people.
Next, let’s “eat the seed corn”. Junior developers can now be<br>considered (by some) to be superfluous to our needs; for the same<br>cost, we could get soooo many tokens, and get so much more done! Job<br>openings for junior<br>devs will dry<br>up. This is already<br>happening.
One of the interesting things about LLMs is that they love to generate<br>new code. Given their vast corpus of training data, they also know how<br>to write those supporting functions and methods we commonly get from<br>third party libraries. The result? We’ll start writing more and more<br>code. The size of our repos will balloon. This is already<br>happening.
Now, we also know that the tendency of the AIs isn’t to go and clean<br>up code, reduce duplication, or focus on maintainability. Context<br>windows (even the large ones!) can’t hold entire modern repos. They<br>miss things. Most AI written code is additive, and is frequently<br>duplicative. This is already<br>happening.
Machines have a higher “tolerance” for complexity than people, which<br>means we can now bear higher complexity budgets and tech debt. That<br>is, modern AIs tend to be very good at reading code and following the<br>flow of control, often faster and better than<br>people. However,<br>“debugging is twice as hard as writing a program in the first<br>place. So if you’re as clever as you can be when you write it, how<br>will you ever debug it?” still holds true. At some point the tech<br>debt and complexity will be so high even the AIs won’t be able to deal<br>with it. We’ve already frequently blown past human levels of code<br>complexity. This is already<br>happening.
And code is never bug free. Defect rates in AI generated<br>code are<br>typically above that of human-written code, but even if it drops below<br>that, because so much code is being written, the overall number of<br>defects will climb. Because code is seldom properly factored, fixing a<br>bug in one place won’t fix it everywhere. We’ll play whack-a-mole with<br>bugs. This is already<br>happening.
So, we’re going to end up with (we already have!) complex, poorly<br>structured codebases, rife with bugs and duplicated code. Who has the<br>judgment to decide what to delete, which abstraction is wrong, or<br>whether the whole approach needs rethinking? Who do we call in to fix<br>this kind of mess?
Senior developers.
However, there’s attrition in the industry. Senior developers leave,<br>not only because of the regular attrition that occurs over time, but<br>also driven by a 22% spike in critical<br>burnout<br>as they are forced to manage the massive influx of AI-generated<br>complexity. Because we’ve stopped hiring juniors, there are few new<br>seniors coming up to replace them. The ones who do make it through<br>haven’t experienced life before AI; they’ll be good, but they won’t<br>necessarily be great at writing maintainable code. In a market where<br>senior, production-ready<br>engineers<br>are already the primary bottleneck, I think we’ll need seasoned<br>developers with good taste more than ever.
In any market where there’s increased demand and reduced supply,<br>prices go up. This happened for the COBOL programmers of yore during<br>and shortly after Y2K. It’ll happen for senior developers soon, and<br>just like our COBOL-wielding brethren before us, there will a shining<br>window of opportunity. We just need to hunker down and survive the<br>storm of AI-driven layoffs.
Then, my friends, we’re going to make out like bandits.
The Dorodango Technique: Agents in Agents Driving Scripts Less recently
Related
The Dorodango Technique: Agents in Agents Driving Scripts
January 29, 2026
tech
ai
Principles of software development
May 20, 2025
tech
rules_jvm_external's New Maven-based Resolver
April 26, 2024
bazel
tech