> Why pay $230k a year for a "scalability/reliability engineer", when I could pay exponentially less to Google or AWS and not have to worry about anything?

I think you're massively undervaluing the complexity of systems at scale. Let me give you a more concrete example. In Facebook and LinkedIn there is a feature that says "You may know this person". It's a small box in the upper right corner and looks like the smallest feature ever. But the underlying complexity of building such a feature is easily millions of dollars worth of man hours. Just that small feature has probably single-handedly grown those companies by > 1x factor. You don't get that simply by "throwing it at AWS".

I appreciate your enthusiasm for this topic, but I think the lens at which you see the startup world is being blurred by media and survivor bias. There is so much under the tip of the iceberg that 99% of the people here will never see.