Scene luminance has nothing to do with exposure.

No one who knows what they're talking about has said that the in-camera histogram is not accurate for the camera's conversion or a default conversion. It is BASED on the camera's conversion parameters, so how could it be wrong for that purpose? What it doesn't show you is what is going on in the raw data, and depending on the camera and WB and ISO setting, the amount of raw headroom can vary tremendously. If you want to venture safely into those extended raw highlights, then you would want a raw histogram; not a default conversion histogram.

This is so common with wildlife photographers. They will swear that their bigger-sensor, bigger pixel camera gives better IQ in low light, even in focal-length-limited situations, when their smaller-sensor, smaller-pixel cameras actually have more resolution and less visible noise. Lots of people have no idea what they're looking at. We see the same thing with people who give up on TCs, because they did something like go through their photos, investigating 100% pixel views, and concluded that the "IQ" was consistently better without the TCs. Humans have not evolved to understand sampling and scale; there is a visceral drive to see acuity as angular micro-contrast, as confirmation of the eyes focusing, and when we can't find that acuity, our lizard brains see a failure, even when we have much more actual detail. We have to learn to ignore this primitive evaluation.

I didn't either and do not see them in practice with a7r4 but after reading 1000+ posts a74 seems to be a different animal.

Those three things are completely different and unrelated. Only Sony zebras can be used for ETTR.

If Sony zebra thresholds are specified as IRE levels, that's not working with the raw data, so I don't see how that tool will reliably give an ETTR exposure.

Well, and to my point, none of those tools support a real ETTR exposure assertion.

How?

Sony zebras show where the clipping occurs, not exactly as in raw but is closer than most other clipping warnings, AFAIK. Very useful to maximize exposure.
Fuji Natural View disables WYSIWYG in order to make detail in shadows more visible, but the histogram is not valid. No change in metering occurs. Fuji should have done it as Nikon and disabled the histogram when in that mode.
Nikon’s Highlight weighted is just a different metering mode that helps in situations like theaters, it has nothing to do with ETTR. It may help avoid clipping some highlights by underexposing. You can read detailed analysis in Thom’s e-book.

That is true. But histograms, zebras, highlight clipping are often used to improve exposure. Fuji’s natural view and Nikon’s highlight weighted cannot be used for that.

If your definition of "improve" is just to sidle up the data to the saturation point, yep. But I'm "improving" my exposure with highlight-weighted matrix metering by not blowing highlights, at the expense of a stop or two of headroom.

Wrong ,you need to actually test the cameras you think you know something about. then get back to me.

Yes they are

Most highlight weighted metering can blow highlights or will unnecessary underexpose.

What one camera model does does not necessarily negate what another model does.

Indeed it does, both of those things. I've been using my Z 6's highlight-weighted matrix metering as my default metering mode for almost two years, and it definitely will let specular and in-scene illuminant highlights saturate, I'll take that. But, I only have to use highlight reconstruction occasionally to color those highlights, never to reconstruct meaningful content in the scene. I do have to mess with a tone curve to shape the rest of the data, but I've set up my workflow to make that simple and quick.

"Necessity" depends on the situation. I've tried most of the ETTR stupid pet tricks, just don't work for me when I'm in the middle of a railroad yard in the early morning where the light is changing by the minute and I'm in close proximity to large moving machines. Yeah, I'd like to have that stop or so of headroom back for my image, but I've not run into a situation where a bit of tone curve shaping won't bring the things I care about into view.

I'd much rather have a raw-based tool to readily assess clipping, but I think the camera manufacturers are sensitive to having histogram/zebras that are consistent to the image in the viewfinder so I'm not holding my breath...

Excellent, Donald, I just now read your terse one-liner amongst all the obfuscation.

It is fully supported by this article en.wikipedia.org/wiki/Exposure_value.

Well done.

It's impossible for your camera to always give the same RGB headroom in both raw and display RGB, just because of white balance. Your camera would literally have to alter its filtration of color during exposure, or the raws would have to have WB applied to them (wasting headroom) for your assertion to be roughly accurate. Also, the only time you see headroom in the green channel approximately the same in the raw and display RGB is when the raws have limited headroom, like the "extended" ISO 50 in many FF cameras.

I just compared the Sony A7R2, Canon R5, and Nikon D850 at DxOMark, to see the saturation-based "measured" ISOs, and sure enough, the A7R2 has the least absolute highlight headroom at ISO 100. So, you may see the two histograms looking more similar with your camera than with the others, but that does not mean that your camera histogram is a substitute for a raw histogram, as the red and blue channel histograms will diverge due to WB. You'd have to have a white balance for a magenta light tint, and greyscale subject matter, to expect the two histograms to be as close, consistently. Add highly-saturated highlight colors and WB, and the two histograms will diverge.

As I read it, it is sarcasm and a bit more :)