My point, "it depends on the surface", is unchanged and simple. It follows naturally when one understands the 2 basic principles at work, mentioned above, which we both agreed. Here's another way to say it by example that may be more clear.
Pick a tire, for example the Conti GP5000 S TR in 700c. It comes in widths from 25 to 32 mm. Pick any 2 of them, like the narrowest and widest at 25 and 32mm. We can find a surface where the 25 is faster and more efficient - it's going to be very smooth. We can find a surface where the 32 is faster and more efficient - that surface is going to be rougher.
If my position seems obvious, in my defense it seemed worth saying because my first post was #6 and up to that point nobody had mentioned it.
Actually, that restatement is clearer, so I do think I understand it better. And that understanding gave me some pause because I had to think about it a bit as it put a different light on my original point.
If I could re-summarize and add your point in post #6 that people are (presumably unwisely) "disregarding it," I believe you are saying:
--Surfaces vary (true on normal roads most people ride on).
--Each surface has an optimal tire width/pressure (this is true per physics)
--Therefore, tire choice should consider surface roughness (not unreasonable), and people are disregarding that (true)
--Rather than matching tire to surface, people sometimes overgeneralize “wider is better” (true, and it's exactly what I do)
So, all of that is completely true, and it is addressing the question of my OP at a higher-principal level in order to constrain the lower-level individual cases. This is highly rational and exactly the approach I always tend to take-- so I gotta respect that!
So, I had to think a bit on why I'm going against your (totally correct) set of principles as I am indeed saying that "wider is virtually always better (or, at least no worse)." It's like my point disagrees with your principles but I can't really contradict any of them.
But here is the resolution of the apparent contradiction: My contention is not that you are wrong, but that the wider tires cover enough of the range of surfaces such that there is no need for the granular level tire/surface matching possibility that your principles correctly states exists, but I contend are not needed.
So, I would say:
--Yes, every surface has an optimal tire width.
--But the range of road surfaces cyclists actually ride is narrow enough that 35–48 mm supple tires are optimal or equal to any "matched" tire across the entire real‑world spectrum.
--The surfaces where narrower/fat tires are faster are so smooth/bumpy that they don’t really exist in normal road riding.
--So, the general principles you state are totally true, but in practice wider tires cover everything without penalty
--Thus, your principles, while true, are mostly moot on real roads if one simply chooses wider tires
So, with that, I don't contradict any of your principles but rather contend that wider tires cover all the bases such that it is actually fine to "disregard" the issue.
In any relevant situation for the normal road cyclist riding on typical surfaces, wider is better (in any practical sense). This is what the data say, and this is why they trump (not refute) the point you were making.
Note that your point is one of the best I've encountered on this forum, so good on you for that!
