Not that I disagree with you at all about cost overruns and schedule delays being an issue, or poor incentive structures all around being an even bigger issue, but going by your numbers:
>The terminal is pretty. But it was 10 years behind schedule. It costs $3.75 BILLION dollars to build, making it the world's most expensive transportation hub. And it runs about 46,000 passengers a day. (Compare and note that Grand Central does five and a half times that many). It was not worth it.
That... doesn't actually sound bad? According to what you wrote there it's serving around 16-17 million passengers a year, which would mean the entire $3.75 billion only works out to ~$220 a passenger for a single year, so it could be entirely recovered by a surcharge or temporary rate bump or whatever of around 50-60 cents a day in just a single year. If built properly so the capex can be amortized across five years or a decade or more, all that drops proportionally. That doesn't seem like a huge ask in NYC.
I mean yeah, it certainly apparently is far from the most efficient design. But even so it's worth considering that even a public infrastructure project here with huge scheduling delays and cost overruns that produces a big upfront bill isn't necessarily that inefficient compared to other forms of competing transport (like cars). It's easy to get too silo'd and forget the fundamental efficiencies and reliability vs the wider world, and in many cases it might be worth "reinvesting" some of those gains in the form of making something more aesthetically pleasing (and thus attractive to the public users) or politically greased, because even if it's worse relative to an internal ideal it might still be better for society. And humans are humans, sometimes some "inefficiency" is necessary from a practical standpoint.
The real question I'd have isn't the one time capex, it's ongoing maintenance. If all that money went into building something that'll be extremely reliable and have either equal or lower maintenance structures vs the rest of the stations, amortization can take care of the rest all right. If however delays and other issues result in higher running costs, that's a much much bigger deal and would be more interesting to know about then the $3.75 billion upfront.