> I don't think it's self referential
It involves perception, and you believe it despite not having gone through what you claim is necessary to acquire the knowledge.
> the whole idea holds on that if you manage to do predictions that are confirmed, you still have to assume that your perceptions are consistent across time.
YOu may have to do this, but I certainly don't - in fact, I believe essentially the opposite of this!
> As an individual you can check you perceptions across time, with several people you remove yourself as a variable.
That could depend on the nature of your relationship with the people - it's possible, but not guaranteed (and knowing which situation you're in isn't easy).
> I've been extreme in the formulations, but no, you don't need other methods, repeatable experiment with accurate predictions is enough.
This is only true if it is actually true though - how do you confirm that it is true?
> The heuristics are quite reliable.
...the heuristic process suggests.
> It's just that we tend to forget/ignore that everything that what we think of being true is just a large compound of beliefs, and that the only way to validate those beliefs is to use them to make prediction, and validate it through experiments.
It is true that there is value in science, but there is also risk.