A few (incomplete) thoughts: 1. We haven't yet figured out a way to get enough signal from two hours or so of work. 2. There are multiple ways to be a good pr

A few (incomplete) thoughts: 1. We haven't yet figured out a way to get enough signal from two hours or so of work. 2. There are multiple ways to be a good programmer, but companies need a standardized way to assess skill. There's an inherent mismatch. 3. Most teams don't even know how to gauge their own employees' skill/value. 4. Teams rarely have their own employees retake their hiring tests to calibrate effectively. 5. Unpopular opinion: they aren't. Skills like algorithms are vital, and people who argue that they don't matter simply don't want to do the work.