Relevant to this thread, the third link includes the Scottish Government's evaluation report saying passenger numbers on the borders link were within 2% of the predicted number in Year 1.
Indeed they were, we did our homework, and we did it well.
Passenger usage cannot be predicted
It can. Very accurately even. Again, this is getting off topic, however I'd posit very simply; if passengers numbers couldn't be predicted, then why would open access operators exist.
We can model new demand incredibly well, what we struggle with is modeling changes to demand if say a new park and ride station opens on an existing line. Working out how many passengers would be new versus how many would be shifting their travel patterns from neighbouring stations is almost purely psychological, and therefore very difficult to predict.
While a press release for the Cowley approval quotes that:
Having seen the numbers for how much it actually costs to run short railway lines with only two stops, an hourly service, and no chance of extension, in a city area where you're competing with buses, I can firmly say I think that one million passengers a year will not be financially viable. It's a publicity stunt and a white elephant, trains aren't really very efficient at serving a three mile branch line.
Now, I humbly request we move on from Portishead and Cowley and back to updates with the Borders wiring.