Is Tech Hiring Broken?

"You gotta give me one thing. I'm a scary judge of talent"  -  Al Pacino as Walter Burke

When I was hiring multiple tech roles before 2020, I often felt like Al Pacino. Most of my hires were gut-feel hires, and many of them went on to exceed my expectations. I believed (and maybe some part of me still does), that human instincts played a role far more important than any other tests did. That body language, facial expressions and speech were better giveaways then a smart problem that obviously only ingenious people could solve because you, the interviewer, also did not get it on the first attempt.

I looked for people who were smart. People who did not necessarily possess existing knowledge or experience, but had solved other problems before. People who believed in themselves.

Around 2021, when the world was still recovering from covid, the tech industry was going crazy. Organisations were wooing software engineers with gifts and promises. Talent acquisition teams were begging hiring managers to make decisions based on no more than 2 rounds of interviews. I remember talent acquisition telling me once that the candidate I had rejected already had a salary that we could not afford. I started supressing my gut-feels, and making hope-based hires.

Once the layoffs hit, two things were clear - 1) Everyone had made hires that they were not 100% onboard with and 2) The new-hire freebies were now just a notepad, pen and a flimsy water bottle. However, hiring started getting back to what it was before. Or at least it felt like it did for a while. Suddenly there was this emergence of coding puzzles everywhere. What even is “competitive coding”? The only competitive coding I know is proving that my language of choice is better than someone yours, and I do that by abandoning all high priority tasks to go build a POC that compares the benchmarks of my code over yours.

Anyway. The whole 5–6 rounds of interviews started looking like a good idea to everyone. Even companies that did not even have 5–6 people to take that many rounds of interviews decided that this was the best way to hire people. It wasn’t so much about the multiple rounds of interviews, than it was about the breadth of expectations and requiring the ability to talk, think, code and sing a song, all at the same time.

So here’s my theory - when multi-year initiatives that were kicked off in 2021 started failing, it was obvious to all the decision makers that these failures were due to bad execution, which was obviously due to bad hires. Furthermore, with AI code generators getting increasingly better at demoralising developers, the decision makers realised that a standard 1x developer should be able to product 5x the output when armed with AI weapons.

Interestingly, even today, most of the interviews still gauge people based on things that AI can do really well. For example, most junior developers would make obvious mistakes if they attempt to build an uber-like system on their own. But if they know enough, AI could fill in all the missing experience, provide clarity on why it recommended specific actions and even write some of the implementation. Why is system design so much more important suddenly? How many people in your organisation are going to design large-scale systems from scratch independently, and without any support from humans or non-humans?

I don’t know what’s next. We’re probably going to throw away our interviewing guides soon, and hopefully replace them with a more human-centric way of hiring. Or maybe we just have AI make hiring decisions for us.

“You gotta give me one thing. I’m a scary judge of talent” — AI Machina