Rare that a day goes by without my worrying that algorithms are increasingly taking over. Not in the Skynet (Terminator) sense, or even the NSA’s SKYNET. In the much broader sense, the business world is encouraging algorithms even where it doesn’t seem appropriate, or (at least) a complex algorithm is required.
Recruitment, certainly in the rank and file of software development, is very big on buzzwords and acronyms. An employer says what it’s looking for and, usually, a consultant (internal or external) goes hunting for candidates. Automated searches of CVs and online profiles present the consultant with options that meet certain criteria based on keywords. There’s only a superficial understanding of the terms in the criteria and if they can’t be reconciled with a person that is a little unusual, that’s that. If you want to find work easily, you need to bend to suit the algorithm.
The algorithm isn’t just the one that does the automatic searching. The consultant, too, runs on an algorithm, which is usually simpler than it could (I’d say should) be. Recruiters and their clients often have misaligned goals created by perverse financial incentives – their interests are the short vs the long term, respectively. The consultant isn’t interested in the unusual candidate that can stir things up, as they may be hard to place. The client might need exactly that, to create the next stand-out product or otherwise create positive change.
A candidate who is re-training, who isn’t in the expected place in a hierarchy, perhaps has experienced issues and bounced back – these don’t look like the best hope of placement and so are easily overlooked. The recruiter wants as many wins as possible, so their algorithm is optimised for that, because that’s how they’re measured. This makes things cheap for the client and more profitable for the consultancy, in the short term.
Across many industries, we’re moving in this direction of wanting cheap convenience from providers who wish to maximise profit. This sets boundaries on human agency and how much of humanity we can embrace. Humans, of course, are massively diverse and complex.
Simple algorithms are great, unless you need a very complex one.