Nobody producing these products will ever invest in the type of studies that one side demands - it's simply never going to happen because it's not financially viable or worthwhile to do so for the market involved.
I'm not sure what objective criteria can be brought into a discussion about the M-Scaler. One side seems to believe that a specific set of measurements are the definitive answer to whether or not it does something and the other think that perhaps there's more going on than that. The funny thing about it is that it's those purporting to support a scientific approach that have the closed minds and belief that all that needs to be known is known and fully understood.
I agree. It's asking for the impossible, there simply isn't enough financial incentive to do large and expensive studies on the countless subjective aspects we experience or even the audibility thresholds (or preference trends) of various objective aspects.
This does mean though that manufacturers are able to make claims about the audible effects of their products without evidence, but also has unfortunately meant that a lot of people use the lack of evidence as proof that there is no audible effect at all, which is not scientific or objective at all. Lack of proof is not proof, in EITHER direction.
In terms of the MScaler, the thing is, it's easy to definitively prove exactly what it's doing. It's a very high performance sinc reconstruction filter with a windowing function optimised for transient response. Meaning it will provide extremely good time-domain accuracy and immensely sharp stop-band attenuation.
These things are what the marketing claims, and these have been demonstrated. It's not 'snake oil' as it is demonstrably doing EXACTLY what it says on the tin.
The issue is then whether those changes are audible or not, but to my knowledge there has been 0 study done on preference or even audibility of reconstruction filters. (If I'm wrong please correct me as I'd like to read this)
Though there is some study on the closely-related issue of whether high native sample rate recordings are audibly distinguishable from redbook, and the indication is they are to at least some listeners and therefore reconstruction filters should be too.
So whilst some might claim that the differences are inaudible, that is at the end of the day, just an unsubstantiated assumption. And in fact it's much more sensible to assume the MScaler does make a difference given as the little evidence we DO have is in its favour.