The author thinks that Joel is contradicting himself (since how can it be both easy and hard to tell if a developer is great?), but I don't think so. In fact,

The author thinks that Joel is contradicting himself (since how can it be both easy and hard to tell if a developer is great?), but I don't think so. In fact, it's hard to tell for sure how good a developer is before hiring, but easy after having worked with them for a while. This is similar to how it's hard (as in, expensive) to be sure if a car is a lemon before buying it but easy to tell after putting 10,000 miles on it. No contradiction. Unrelatedly I agree anecdotally that from a developer point of view the market for workplaces looks more like a market for lemons (it's hard to tell if a workplace is crazy before joining it, easy afterwards, and a developer will stay a lot longer at a nice workplace than a crazy one, so most job openings are probably for places that have a high turnover for a _reason_).