I see a lot of arguments that seem to focus on one or more variables. Often an argument that neglects to mention some element gets taken as a dismissal of it. It's good to keep an open mind and open ears. Some things are appropriate to dismiss. Some are worth paying attention to!
A lot of technically imperfect audio storage and delivery systems work at a high enough level that unless you have a very professionally tuned listening space and reference quality gear, you'll never be the wiser. That's a good thing right? And even if said elements of loss ARE perceivable on a well tuned system, it might be one of those 2% discussions. So we're usually in a really good place nowadays for hearing music as intended by the artist.
One of the points I try to make is that the mix quality and mastering quality are still the big factor. That's not a dismissal of any format discussion. Conversely, I'll suggest that since we HAVE lossless HD consumer delivery formats now that they should be used 'just because you can'. Might as well eliminate any format container variables if possible right?*
I've heard various lossy formats result in significant loss with telltale sonic signature. I've heard lowly mp3's of some things where I would have been none the wiser. I've heard excellent mixes really destroyed by 'portable device' style mastering (heavy limiting and then boosting and with heavy high end eq boosting) to the point that any loss from even the lowest quality formats (not just mp3... like poorly reproduced 8-track tape cartridges even) would be a moot point.
When I hear someone critique things like reducing 24 bit program to 16 bit, I'm with you in that I've heard many examples too. I think it should be avoided unless catering to older equipment that requires it. But then when someone talks about this in a way that suggests they hear loss more along the lines of what that poorly reproduced 8-track cartridge might do, I have to suggest that this isn't the variable that's the main cause of trouble in the scenario. Sample rate conversion is even more subtle than that generally. There might be some ringer recordings that are strikingly more susceptible to this than others but we're still having one of those top 5% discussions here. If you hear something more worthy of a 90% discussion for example, it's almost got to be a different variable at play than what you thought.
* I'm not critiquing every HD recording that comes along and having a shootout between 24/96 PCM vs. 24/192 PCM vs. DSD and thinking I'm hearing things. It's actually the other way around honestly! Damaged recordings are what tend to make further generation loss perceptible and sometimes to a surprising extent. The budget or consumer level AD & DA converters that run cleaner at HD vs. SD. 16 bit reductions. These are usually still pretty high level containers for audio but you get the odd case of hearing some loss. Starting with introduction of 24/96 PCM, I haven't heard a single example of any perceptible generation loss due to the container format. Even badly damaged recordings that tend to crumble into dust at the slightest touch are preserved intact. High fidelity recordings in DSD format that I've heard as well as DSD to 24/88.2 PCM conversions of them suggest to me that DSD is every bit the full quality container HD PCM is. If I hear damage, the root cause is always something else.
Following that, my suggestion is to use this format to avoid any "container loss" just because you can and then just set up the best monitor system that is reasonable to invest in for you. Further, when you hear something off, look at the bigger variables first.
My critique of DSD is simply that we didn't need a different digital "language" and to make the disc storage format that uses it (SACD) spin the opposite direction in the drive. None of that is aimed at fidelity improvement. It was a greedy power play move to get consumers to buy into a different system that would make it difficult to play anything from "Company A". Now if you want access to both, you need to invest in more equipment and/or learn some tech skills to deal with it which translates to less music and/or lower quality reproduction for many people. Thus I think it deserves to be ostracized.
Hear something funny?
When I do, I go in this order:
Is it just level differences between the two?
Is it the same mix?
Is it a mastering issue? (Or was one obviously mastered for portable device listening?)
Is one in a grossly lossy format? (mp3, lossy dolby, core dts)
Loss from sample rate reduction to SD or 24 bit reduced to 16 (not including snowballing degradation of already challenged recordings) is more of a comment like: What did you do that for when we have lossless 24/96 containers? I don't see 16/44.1 vs. 24/96 as going from 0 to 100. More like closing the gap from the high 90's to 100%.
When someone then comes along and suggests "Well, my speakers are only such and such so this lower setting is good enough." Spending money is spending money and choices need to be made. But when it turns into stubbornness with a switch setting... come on now! Now who's more interested in arguing vs. actually listening to some music?
When purchasing music, I see SD and lossy formats as a calling card for portable device style mastering and avoid them for that processing work which eclipses pretty much all discussions on container quality. Whereas HD formats are usually a calling card for well handled and presented recordings. There are exceptions in both directions of course but that's the general pattern. (I'm sure there's some mp3 out there that sounds better than some bluray disc release of the same for a ringer example. Which would be an example of a mastering disaster for said bluray.)
Well, have fun with that!