That framing doesn't change my logic at all: replace the direct earning of $N/year with the probability of earning $N/year and the same logic still applies. An employee doesn't contribute the probability, an employee working within the Google system contributes the probability. The system is doing the bulk of the work, and the per-employee probabilities of earning all that money are dramatically smaller in any other system.
The very well-compensated early employees, leaders, and executives, working together as a team to create something that was greater than the sum of the parts. The average Googler in 2025 wasn't even on the team that created the initial system, they're a cog in a machine that someone else designed years ago. A very well paid cog.
It's not lol. It's more akin to construction. If you need 100 people to do a job in a given amount of time, you might be able to get it done with 99 people without issue. But that one person might have had an insight that would save you a lot of money overall. And if you kept getting rid of people, at some point the others will not be able to pick up the slack and might even ruin the project unwittingly.
Firing an employee most likely won't reduce the bottom line.
But every employee is a chance to increase it by creating a new project/product etc.
So firing would reduce the probability of a new idea generating an increase to the bottom line.