It's not the relation to other categories I'm interested in; that is just an intuition pump for MB (which I'll return to presently). I agree that the application is the main issue, but the level of abstraction is what I am highlighting. To run with your examples from zoology:

"Snakes are venomous" - specific, low abstraction, useful.
"Living things are venomous" - more abstract, sort of true but possibility of being unclear or misleading.
"Life is venomous" - high level of abstraction, very unclear, borderline meaningless without clarification.

The reason I made the point about theocracy/democracy is to give MB a sense of how it's possible to jump down several levels of abstraction and then say "Ha! Well you'd accept that this proposition is meaningful" but it fails to address the point. I agree that at such a lower level of abstraction, the propositions become meaningful.

Finally, as you agree that the application is key to this, please respond to my request that you outline a methodology (feel free to make up any data) that would make sense of a quantitative analysis of the claim "society is a net negative". It really is easy to change my mind on this as I'm essentially making the claim that I simply can't make sense of the proposition without reducing the words "religion" or "society" to a more specific term / lower level of abstraction. You deny this, so it should be easy change my mind.

As another intuition pump, what I have in my mind is some sort of graph with labels on the x and y axis; it's easy to imagine ways to do this for smoking (number of cigarettes smoked lifetime vs average age at death would be one possible way of doing it) but I am at a loss to imagining any such quantitative graph that has "religion" as the label on an axis.. what units are used.. is it nominal data e.g. using "religion" as a count noun?.. that sort of thing. Remember to do this without getting any more specific than "religion", as to do so would be to accept my entire argument.

This is a very odd definition of "religion".

I don't claim that survey data is without value. But in terms of its usefulness for determining something like "religion is a net negative" I think it's pretty useless.

This is now straying quite far from where you started. You came into the conversation with the perspective that:

So as you post this "happiness survey" are we supposed to invert the axis just because you want it to be that way?

Actually, for MB this is pretty standard. In many conversations in the past, he's taken very general statements and insisted that they apply very specifically to religion and not other modes of human thought or human behaviors.

He has also created very distinctive categories like "religious parenting" which is not how religious parents actually raise their kids but how he believes they should be raising their kids (which he then objects to).

Would be an even better zinger if both books had the same author. Which they should by the way.

I'll refrain from commenting, apart from noting that some people certainly do themselves a favor by serving their wit with pictures.