I'm having trouble reconciling this point. How can you continuously increase your effectiveness exponentially in a particular role? Take being a developer for example. Seems like the progression of skill is usually linear (with some variability). Is Sam saying that investing in becoming a great engineer ins't effective if your ultimate goal is to be extremely successful? I feel like this is overly reductionist. Maybe being an engineer for your entire career isn't going to lead to that massive success, but spending part of your career cultivating that skill does provide indirect benefits that can help you become super successful. For example, being able to rapidly prototype an MVP when you are testing product ideas. Building a business is the high-leverage action that could lead to massive success, but the years you spent cultivating your technical ability could be what enabled you to be able to build the MVP in the first place. Or maybe it allowed you to recognize good technical talent if your first engineering hires. I think exclusively focusing on tasks that provide orders of magnitude improvement is too direct and superficial of an answer. I'd be interested in hearing people's thoughts.
For example, 10 years into your career as an entrepreneur, you'll have a network (and possibly financial assets, and possibly an existing successful business, plus vague intangibles like 'wisdom' and industry insight) which all lets you be vastly more effective than someone who is equally skilled but still in their first couple years.
I don't think Sam is telling you not to build any particular skill, but rather to put yourself on a path where, 10 years down the line, you'll be able to use those skills in a way that other folks can't compete with, thanks to the other assets/resources/multipliers you've built up for yourself, and the fact that you're playing in an area where there is a) the potential for scale and b) you own the upside if it works.
Caveat: I have no idea, just my interpretation of the article.
For me, I've put a lot of effort into both breadth and depth, and that's where, for me, a (maybe not exponential but) super-linear growth comes from. When you dive deep on a broad set of topics, you start seeing relationships between them that others don't see. But that takes a lot of dedication to diving into a diverse set of things, and go deep on it.
Not to mention some very keen, obsessed younger developer could easily replicate what you're doing.