Tejwant ji & Seeker9 ji,
Thank you for your response! I can see a few possible responses to it, but (spoiler alert!) at the end of the day I was wrong.
First, I could hold fast and say that the type of paradigm shift we see in science isn't present in Sikhi. While we certainly learn new things and correct what we take to be mistaken beliefs, a complete paradigm shift in this technical sense would be something entirely new, something that isn't Sikhi properly-speaking. This new religion would be something where we could see the ways in which Sikhi was right, and it would also let us understand how Sikhi could lead us astray. Our new vocabulary would better describe the world, but would also let us exactly describe just what's wrong with Sikhi. But this response fails. After all, adherents expect that the Sikhi of long study really is a radical rearticulation of the neophyte's Sikhi, and that even though the terms might look the same in each model, the meanings are radically different. Since we certainly want to say that both understandings are articulations of Sikhi (though the wise one corrects and accounts for the errors of the naive one), this response fails.
The second response would be that Sikhi is unique, and uses a fallible methodology where other religions do not. But surely this is false, too. Naive Christian understandings are radically incompatible with existentialist Christianity, and "Adonai" in storybook Judaism means something far different than "Adoni" on the lips of a negative theologian. So this cannot be right, either.
A third response (similar to the second) would be that religion proper is practiced non-scientifically, and Sikhi or Buddhism or whatever becomes a sort of interior/spiritual science or something when practiced fallibly. In this case, the practices of Sikhi (and other religions) wouldn't count as religion at all. As they approach science, they become scientific (as opposed to religious) inquiry. This is the direction I'd go if I had to defend my prior view, but I think the violence done to the term 'religion' is probably too much. As such, I think it's best to abandon my prior definition, as it simply doesn't match with what the word 'religion' actually means and how people use it!
I should have realized this when I said that religion can be practiced scientifically and fallibly, but I did not take the full import of this into account in my definition. If religion can be scientific, then something can be both religious and fallible. This obviously conflicts with my definition of religion as dogmatic, and was a mistake!
What, then, is the right definition of religion? I very much like the start that seeker9 ji gave us. To alter it a bit, perhaps religion is primarily about the subjective, the internal world as opposed to the external world (that is, all the evidence we have comes from us). Then the difference would be that only we can tell if a set of religious practices is right, because that is indexed to us. In any case, I certainly concede that my previous definition was wrong.
I think that most of my comments in the prior post can be preserved. Science really IS defined by its methodology, so that part at least was a-ok. There's still a distinction between science and religion, but now there's also room for an overlap of the two, in fallible religious inquiry. Which is much better than what I had before!
Thank you again Tejwant ji and seeker9 ji for your help in wrestling with this issue and for correcting my mistakes!