Yes definitions are important and are there are no clear standards here which is part of the problem. Generally a 44.1/16 bit CD has been considered lossless and anything less than that is lossy and anything more is “HiRes”
MQA is lossy in the sense that it does not preserve all the bits in a 94/24bit file. That is obvious if you look at any of the “origami” descriptions of how MQA works. In the same way of course CD is also lossy in that it does not preserve all the bits in a 94/24 bit file either. However MQA can be considered lossy in the same sense that a CD is.
Most digital audio processing these days is done in formats such as DXD which is 352.8/24 bit so anything less than this will also be lossy in the bits is bits definition. Indeed any file that is less than the full original bit resolution will be lossy and this means even 192/24 is lossy in the pure bits is bits sense. DSD/SACD is also lossy in the pure bits is bits sense.
So how do we judge what is truly lossless? Archimago has tried to get to the bottom of a lot of MQA stuff, he is no fan of MQA and their claims overall but in this
Archimago’s Musings: COMPARISON: Hardware-Decoded MQA (using Mytek Brooklyn DAC)
he concludes “I can say that MQA does “work” as claimed to reconstruct material >22/24kHz with reasonable accuracy.”
Basically it seems the main downside of MQA over files greater than 44.1/16 bits or 48/16 bits is a small increase in the noise floor, so small that it will be in-audible.
So MQA is less lossy than a CD in presenting what is there in HiRes files. Compared with other 24 bit formats it is at least reasonably accurate as judged by an independent critic.
This just addresses the compression aspects of MQA, however MQA also adjusts what they consider to be “flaws” in the original analogue to digital process/fliters which they claim will make things sound better. This is the more critical aspect of MQA and what will impact the sound more than the lossless/lossy debate yet this is rarely discussed whereas “lossy MQA” is bandied about all the time.