Analogy doesn't really hold up. Increasing conversions by 0.01% probably takes an hour at momandpop.com and maybe a year or more at Amazon. Not to mention low-hanging fruit already being picked at large corps.
A lot of things at big companies aren't ever going to be worth doing at little companies. In cloud companies they spend tens of millions of dollars going after 1-2% efficiency gains that really make a big difference when many of these little gains are all stacked and applied to billions of dollars of equipment. Maybe if these companies got broken up, several little companies could sell these little gains, but that seems much less likely than one of the broken up companies naturally becoming big again and doing it themselves.
This field has a ton of natural incentives to scale as big as possible. I feel like the main beneficiaries of breaking up big tech companies are business people and lawyers getting to insert themselves everywhere to interface between groups that used to be the same company.
I think you've proved my point. Working for momandpop.com produces value at $10 an hour. Working for Amazon and assuming 2000 working hours in a year produces value at $500 an hour. And that's if that 0.01% improvement is worth only one million dollars.
