Makes sense. IMHO since there is an ITU standard the default stereo downmix should be made from a 5.1 (or 5.1 render from Atmos) as:
L = L + (Ls-3dB) + (C-3dB)
R = R + (Rs-3dB) + (C-3dB)
But as discussed, that's not always the case. I consider deviations from this to be "wrong" so I use this when checking my own Atmos or 5.1 mixes for stereo compatibility. But if done differently by the consumer deliberately, that's fine by me. It's more of a problem if it happens accidentally and the surrounds aren't attenuated enough.
A solution would be for studios/services to provide a dedicated, properly balanced stereo mix as a streaming option.
Couldn't agree more; it's certainly how it used to be done.
While none of this helps the OP, and going somewhat off-topic for a bit of brain dump, FWIW... I stopped doing so a while ago because what actually happens is they make a 'mezzanine' file from which everything else is derived. Which means they use our Atmos mix for everything. To prove it to myself, I did make some minor tweaks to some stereos I sent, and then used a stereo device on a few different streaming services (mostly web browser on a laptop through dante vsc) to compare and contrast what I got from the platforms with my masters. What I got in most cases was an exact downmix of my Atmos. So, that's good and bad. It means even if we send a stereo mix, they do almost certainly just throw it away. But it also means that if mixers monitor their Atmos mixes in stereo in the studio, with a default downmix, then what you get from the service in stereo is exactly that. Almost sample accurate, I would say. At least, it is in cases where the service knows the end device is stereo. And, if an AVR or TV is doing the downmix, it should really be using the same parameters since this is both an ITU thing and a Dolby default too.
I'd speculate that an exception seems to be Hulu, or possibly all of Disney+ which I think might use some kind of spatial processing on/from the Atmos to give a wider sounding stereo downmix than the ITU 5.1 downmix. I don't think this is necessarily a bad thing, since if anything, I'd imagine widening out the surround image in the stereo field will work in helping maintain perceptive loudnesses and preserving dialog. The downside is, it's a closed box (at least to me) and not something I can audition while mixing/mastering. Like I say, I should probably look into it more for this particular service.
What all this means for me is that I always make sure my Atmos mix works in stereo and 5.1 (sometimes 7.1, if there's time, but that's closer to Atmos). It's my belief that some mixers still end up spending time crafting a stereo version, and then complain when it doesn't sound so good on TV; sometimes even blaming it on some imaginary downstream processing applied by the streaming service / blu-ray author / broadcaster. Based on my limited tests so far, I think they just didn't understand that their stereo mix gets discarded and everything comes from the Atmos master. Or they simply want an excuse for missing the runway with their mix in the first place, which let's face it, given all the ridiculous pressures, schedules and working hours in the industry, is hardly surprising sometimes.
As a result, my Atmos mixes are now often compromised slightly to make the stereo work, since this is super important. Usually it's not a big deal, like, it might be moving some panning a bit so it doesn't lump up as much in stereo, as well as starting with strategies that minimise duplication of phase coherent sound in multiple speakers in surround. Unless as
@krabapple says, we get to do different mixes for different formats, sadly I will have to stick with this strategy for the foreseeable future.
However, none of this really has tons of effect on total dynamic range, which I understood to be one of the primary complaints. This is such a personal (or circumstantial?) preference. The downmixing issue is more one of whether the way the music/fx are done means they get more push in stereo than is wanted, but we're only talking a few dB here and there, which while likely hurting dialog in places if the mix is quite dense already, IMHO isn't on it's own likely the direct cause of "I had to reach for my volume control".