When it comes to skills and pay, it all comes down to that most basic of laws: supply and demand. With some older skills, the
number of proficient technologists is relatively low, so employers end up paying a lot more to secure their services (this is a
key reason why people who know their way around an ancient mainframe, for example, can earn very significant amounts of
money; there’s only a small number of active professionals with a grasp of decades-old hardware and operating systems).
With “hot” skills such as deep learning (a key method for machine learning), the same theme applies: A small but growing
number of technologists have mastered these new technologies, and can extract high salaries from hungry employers. And
sometimes, there are situations where a large number of technologists have mastered a particular skill, but the market for that
skill is still so-white hot that salaries have kept rising; just look at how compensation for Swift, which is used to build apps for
Apple’s software ecosystem, has continued to rise despite a healthy pipeline of technologists who’ve mastered it.