Maybe because v1 and v2 did not even have working lasers on the hardware level...?
The idea is coming from "reality", Starlinks own reporting, industry talks, tech press etc.
Anyway, to shorten this we can agree that we have different definitions of what one expects from having a dedicate backbone. I would expect seamless handover amongst other things, which I have never ever seen, and unless you show me a video recording of a 24h Starlink session with MTR running I simply will trust the data I have over a random claim.
As said elsewhere in this thread: It is extremely hard to find detailed benchmarks from happy Starlink users. Next to all positive content is paid content. And a quick look at trustpilot & co clearly hint that there a huge chunk of Starlink customers might be unhappy. And even if it's just because their online gaming sessions getting interrupted on every Sat hand-over, which exists in reality, but not in your mind :)
Seriously, if you have access to any benchmark data sources, please gimme. I'm not here for "winning" an argument. Data, Data, Data.