Why You (Mostly) Hate Your Jobs
In an awesome but perhaps over-mathematized post by “Peter C”, the answer is simple: employees who learn too much become irreplaceable. A standard piece of career advice is to become irreplaceable at work for this reason: the harder you are to replace, the more of the enterprise’s economic surplus you can capture. Bosses prefer to be managing people in McJobs, jobs where everything can be learned very quickly – this means that everyone can be replaced. It’s very good and makes a lot of sense. However, I think it also does a poor job of explaining the behavior of employers with regards to the people with extremely slow learning curves.
Namely, that the best way to create McJobs for most of your employees is to have a small number of extremely highly-skilled workers with relatively little supervision. You absolutely need the people that automate the tasks, that design the systems of people that can deskill jobs effectively, and so on. Peter C’s model treats all jobs equivalently, and suggests huge rewards for getting rid of people that have jobs that will take a long time to master.
Working in a small software company, I can go ahead and tell you that this model doesn’t really accurately reflect the desired mix of staffing. It probably does reflect the desired staffing ratio for back-office functions like Finance, but there’s a huge demand for the actual irreplaceable people in the functions where success is very unevenly distributed, like Sales and Engineering. Managers don’t want replaceable talent in those areas, because replaceable talent in Sales and Engineering has a negative expected value. The key off-point assumption here is that all employees are actually replaceable for the right price. The market for skilled labor, especially in technological sectors, doesn’t just have “transition costs” – it is sharply discontinuous.
If you want an skilled data scientist with experience in predictive modeling, analytics engineering and pre-sales, fluent in Python/C++/R and Hadoop engineering, you cannot find that person for any price. Google could. A hot new Sequoia-backed startup could. But basically any company would love to have them, an insight backed up both by common experience and the well-known “10x rule” – a top programmer is more productive than 10 mediocre ones, and a bad programmer has negative productivity. So, in short – Peter C’s model is interesting and makes sense, but the more specialized the skillset involved, the more illiquidity becomes the defining factor of the employment market and the less relevant the model is.