Only 5 trillion scientifically conducted tests have been done which show them all nulling.
That's only testing one aspect of the system.
A single WAV file will null itself. Yet when you take that WAV file and play it back on GT and then play it back in Sound Forge (no effects, no nothing) there is a subtle difference in sound quality.
I have no explanation for why that happens, it doesn't quite make sense to me. But my first-hand experience and testing resulted in that personal conclusion (I'll grant you that the testing was single-blind and subjective only, so it wasn't scientifically perfect. But I'll stand by the results.) The kicker was when we had two different mixers on two different systems in two different locations working on different aspects of the same project, which involved converting a project that was tracked in GT over to Audition (with finalizing in SoundForge). During the work, both parties independantly noticed a difference in playback quality of the original source tracks, but thought they were nuts until the subject came up in a phone call and they confirmed each other's feelings.
Again, I'm not claiming scientific purity with these tests, nor am I offering up an explanation for them. It goes against my understanding as well. I will also point out that this was with older versions of software (GT2, AA1.0/CEP 2.1, SF6, etc.). But I do stand by them as being solid enough to trust as there being something there more than just psychosomatic. Add in the anecdotal stuff from the classical community (FWIW), and I think it worth considering.
Now, does this have a direct effect on the resulting mixes? If the difference is in the playback engine and not in the summing engine, then perhaps not so much. It might be argued that the quality of monitoring is affected somewhat, though, which might have an indirect effect on the way the engineer mixes it. If he mixes identically in two packages, the results may null. But if subtle differences in the monitoring causes the enginer to make subtle changes in the mix, or perhaps moreso, premastering, then yeah, the results will be different.
Will they be THAT different? Maybe not. You don't see me rushing out to buy Samplitude
. Even if the difference were real, I personally don't see it as being big enough for my home rec situation to be worth the price. I'd rather spend the extra money elsewhere in the studio (or better yet, save it for sometihing that is actually important in life.) But if/when I move the quality of my project studio up to the next step - requireing a heavy 5-digit investment - a few hundred extra for the best in NLE software would probably be worthwhile...whether it soudns better or not.
G.