Goods distributed as software and, to a lesser degree, online services have phenomenal scaling leverage arising from very low marginal costs. I can write some

Goods distributed as software and, to a lesser degree, online services have phenomenal scaling leverage arising from very low marginal costs. I can write some software publish it and in a matter of months have it improving the lives of hundreds of millions of people. Engineering aptitude is rare, or at least we haven't learned how to mass produce high end engineering aptitude. It's hard to argue that any other domain of engineering has anywhere near the kind of leverage that software does. Automation has also radically improved the effectiveness of top software performers. Even a one man 'team' with a moderate level of skill can have pretty comprehensive set of automated tests. There is design automation in other engineering fields, of course, but it's especially potent for software because the automation can capture basically everything and because software engineering is also the skill needed to improve the automation itself. Some of the bimodality may be due to binning what really is functionally separate fields under one name. I've known people who were great at wiring up prefabricated functions to build UIs and simple processes, but wouldn't have constructed a novel datastructure if their lives depended on it. It's hard to draw bright lines because there is a bit of continuum and people grow and change... but there seem to be some pretty distinct categories out there.