I’m with @simon.pepper on this, and it’s been my understanding since reading the original description of MQAlong before it was offered on Tidal, long long long before Tidal started prioritizing its MQA versions over the standard Red Book ones.
I think there has been confusion over the use of “Hi Res”: some people (and almost all here) use it to mean “higher than CD” 44.1/16, but others (mainly coming from the “default lossy” world) use it to mean “better than whatever lo res MP3 you’ve been listening to up until now” (and it often tops out at Red Book). Lots of op ed articles on this out there, as it’s definitely a political stance…
In the case of MQA version of a CD quality source, the as-delivered 44.1/16 stream played back directly (ie no “unfolding” at allis indeed not a bit perfect version of the original Red Book file. Call that lossy, noisy, adulterated… whatever you like: it’s the same bit rate / but depth of course (so in that sense you might argue that “lossy” is an inappropriate word to describe what’s happening), but it is not the same signal. This is likely “bad” and audible…
But after the first unfold (eg software via Roon, or any other way you like), then the container (at 44.1/16) has been unpacked into a higher bit rate / bit depth stream for your DAC to convert.
As I read the claims, the compression algorithm from MQA fully covers the 44.1/16 space in the first-fold “triangle” (see the diagrams explaining the “origami” approach), and only partially covers the higher resolution (eg 96/24) space. The second “unfold” offers better, but still not complete, coverage of this 96/24 information space… hence even after both unfolds, MQA is a lossy compression / decompression encoding technique for original signals that are “higher res” than Red Book. But, whether after only one or after all of the unfolding has been done, the resulting higher bit rate / bit depth stream contains the full Red Book signsl (plus some noisy least significant bits that are below the Red Book noise floor and thus are irrelevant to the signal information content).
So in that sense, I’ve always felt that the confusion over “lossless” and “lossy” and “Red Book” and “Hi Res” has been massive and complicated and is at the heart of the (technical, but not the DRM) arguments for and against MQA.
TL;DR: for a CD source file, unfolded MQA is lossy but even a first-unfold-only stream has all the Red Book content; no amount of unfolding gets back a bit perfect data stream of a 96/24 source file, these are (tuned) lossy encodings for higher-than-CD res recordings.
As for Tidal replacing CD recordings with MQA, it is certainly tricky and a bit obscure…. Tidal definitely “push” the MQA versions now, but their website confirms both MQA and Red Book formats are offered. So if you search on an MQA title using the Tidal app, eg via that artist, then select “other albums” or “discography” you can still find the non-MQA version buried in there; favourite whichever ones you like, both are there. In the older Naim interaction with Tidal, I recall seeing both versions on display…. I don’t see that as often now. But @Stevesky has explained that the Naim stream pulls the non-MQA Red Book version for us since that’s how the hardware works best. In Roon, it is possible to see “versions” and choose the one you want, plus control whether the first (software) unfold takes place or not. Your choice, your call.
Possibly I’m horribly wrong… but I really don’t think so! Happy to hear from others with a different take on this!!
Regards alan