I've been complaining about companies wasting candidates’ (and their own) time with coding assessments for many years. I firmly believe that the cost to an applicant of hours of time for each and every company they're interviewing with is grossly unfair, especially if the candidate is between jobs or, as I was, contracting at an hourly wage. Additionally, for companies to be spending their hours suboptimally on hiring just doesn't make financial sense!

So far, the best compromise I've come across is using real open-source projects as a hiring tool (I wrote about it here: https://therightstuff.medium.com/effective-technical-interviewing-with-github-773509af3501), and in a recent conversation with a coworker I revised my strategy a little: explain the role requirements to the candidate and let them pick an issue for a project that they feel is relevant and reasonable to accomplish within a given number of hours.

This method would give the interviewer quite a few data points: the candidate's understanding of the requirements, their own estimation of their abilities, their ability to get the job done, the quality of their code, testing and documentation, and the quality of their interactions with the maintainers of the project. Additionally, the candidate ends up with recent, real work for their "portfolio", a legitimate sense of self-direction, and actually contributes something real to the open source community.