I don't get your bus analogy. Surely the people who get on bus B which goes to destination Y do so because they want to get to destination Y? The act of getting on the bus doesn't cause them to want to go its destination.
And how does that choice seem to affect the outcome? When gender dysphoric children are met with a neutral model of care that primarily seeks to observe the child, about 4 in 5 desist by young adulthood. When they're affirmed, and especially if put on chemical treatment to suppress puberty, 98% or more persist with a trans identity. Even with a suppressed puberty, transgender people experience worse health outcomes than cis people across a variety of measure. To say that the former approach is a better healthcare outcome in aggregate is not a denigration of trans people but a recognition of the challenges they face. To justify affirmation, the improvement has to be demonstrated not only against an adult transition, but also against the population that desist and live life as cis people.
The studies presented above took a sample of the patients that visited a a clinic that voiced distress of their gender over the span of a period of time - the majority of them meeting the criteria for gender dysphoria in the DSM - and tracked which of them desisted or persisted in expressing a non-cis gender identity decades or more after the fact (average time from first visit to last follow up was 13 years). The lowest rate of desistance was 70%, 3 out of the 4 were above 80%.
To call the results of a study "unrealistic" indicates that one already knows a "realistic" result would be. This is essentially admitting to bias approach to the data: if it doesn't conform to your predetermined "real" result, and your criticism is solely based on that and not any methods in the study. By comparison, the studies that show extremely low rates of desistance are either studies with kids on blockers, or they are not cohort studies. E.g. studies recruiting respondents from the internet is vastly more susceptible to reporting bias than taking the group of patients visiting a gender clinic over the course of a year.
If you want to actually post and discuss a study finding high rates of persistence under a neutral model of care, I'd be very interested in reading. But my approach towards deciding what a realistic result is leans more heavily towards published research than anecdotal claims.
It's more like I have two different buses. When kids get on bus A, ~80% of them arrive at destination X and 20% at destination Y. When kids bet on bus B, 2% of them arrive at destination X and 98% of them arrive at destination Y. It sure looks like bus B isn't merely affording the kids "time to think" but is in fact altering their destination, does it not?
The patients in the study are diagnosed with the same criteria for gender dysphoria in the DSM. Heck, the author in the study I linked wrote the criteria for gender dysphoria in the latest iteration of DSM. I'm always puzzled by people who insist that the study was including patients that weren't actually experiencing gender dysphoria.