People

Hiring for both early career engineers and executives is prone to a kind of myopia that I’m calling “capitalist ignorance.” This is a form of blindness that causes leaders to consider people filling these roles as fungible parts in a giant machine, rather than the designers of it.

There is a lot of anxiety in tech right now because it looks for all the world like sophisticated LLMs are going to wholesale replace junior (early career, recent graduate) engineers.

If an LLM can write mostly working code in response to a prompt no more specific than you would hand to an early career engineer, the economics would point to that conclusion. But if that happens, where do tomorrow’s senior and staff engineers come from?

Now, let me flip this on its head. It is common practice for a company at scale X to go out and hire leadership from companies at scale X+1, with the assumption that: 1) reaching scale X+1 is repeatable, and 2) a person who has been on that journey can teach the company how to do it. But does company X want to become company X+1?

Both of these scenarios have happened and will happen. What they have in common is that they are based on capitalist ignorance. What do I mean by that? I mean that on paper, in the near-term, those strategies may pay off (hiring fewer junior engineers saves money; hiring an exec from a big firm makes stockholders go bonkers). But in both cases the clear near-term gains mask the long-term failure modes.

Story time!

In 2000, I took an hourly contract gig at a giant insurance software company. My little team of four would spend all day, every day, looking at terminal screens of this insurance management software system and converting them into analogous GUI windows using Java Spring. We designed the UI visually in IBM VisualAge for Java, a now-defunct IDE.

To pipe the data in and out of the form from the backend, we had to write boilerplate code to connect the form’s input controls to a sort of middleware that was already written, and the rest was handled for us. Everything was named by convention so it was very rote and boring.

So boring that I spent almost a whole day figuring out how to write an extension for VisualAge for Java that would look at the names of the GUI controls and just spit out the boilerplate initialization code. This saved many hours of pointless labor per day, allowing humans to focus on the part of the job that only a human could do.

That extension was sent over to their UK office where another team was doing the same work, until finally the project was canned and everyone was let go. Except for me. They offered me a salaried role as a junior engineer and I joined another team doing the same sort of project with webpages.

Fast-forward to 2018 and I’m working at Wayfair, leading our “Development Platforms” team, a group of about 40 scrappy engineers working to make life better for the other 1,200 engineers there. As an aside, that role was easily the most challenging one I’ve ever had.

One day I’m in a one-on-one with my manager and he tells me that they’ve hired someone to come in as Director of Development Platforms. At the time, my title was Associate Director. This was one example of what had become a broad-ranging strategy of pushing our internal team of executive recruiters to pull in leaders from other, larger companies. This new guy was coming to us straight from Intuit.

Getting “folded under” some new jackass was a kick in the balls, so I made a big show of quitting and took a role at HubSpot for a couple of years. While I was doing that, Wayfair was steadily hiring people out of PayPal, American Express, WalMart, and so forth. These people were installed in senior leadership positions in the engineering department and started running things the way they had at those other places.

Which is to say, you know, just in my opinion, uh, poorly. You can fight me on that but Wayfair’s stock value right now, today, is about the same as it was on the day I quit in 2018, so I don’t think they’ve done a spectacular job. At least they did manage to take a 5x jump in equity value during COVID-19 and totally squander it. No, that’s unfair, most of them cashed out a lot of that value for themselves while continuing to manage the company like a Vietnamese sweatshop. But I digress.

The guy who was hired to replace me was let go, by the way.

“So Aaron,” you’re thinking, “what is the point?” The point is that computers can’t replace juniors because no LLM anywhere, ever, is going to be able to squint at the task it’s been given and suddenly automate half of it away without being asked. Sure, it’ll do the original task faster, maybe better in some ways, but it’ll never innovate, it’ll never give feedback, it’ll never really surprise you.

An LLM has no fucking ambition because it’s not a fucking human. So go ahead and use LLMs for stuff, and by all means soak up the value of the billions of dollars of VC funding that’s keeping these systems artificially cheap right now, but don’t stop hiring junior engineers.

Likewise, just because WalMart is a juggernaut retail company and the largest private employer on planet Earth (with over 2 million employees) doesn’t mean their leaders are the anointed saints of engineering management. In fact, the opposite is likely true: strategies that worked at WalMart may only have worked there because they are WalMart.

This isn’t a broad condemnation of hiring people with experience (which would be a foolish thesis), but it is a condemnation of the capitalist ignorance that leads a CEO to believe that “what worked there will work here.” Get back to me when you have an 800-billion-dollar market cap. But you won’t.

People are the primary driver of value for any organization. It is people’s ability to be unexpected, to think outside the box, to have vision and purpose, and to apply their creativity that produces outsized results.

Don’t hire people because they walked somewhere you want to walk; that path can only be walked once. Hire people because they understand how to clear a new path.

Don’t stop hiring people because it’s cheaper for a computer to do 90% of their jobs. Guess what, all the real value comes from the other 10%.

Comments