How can we say something is the "identical code" when they were changing code throughout the entire development process?

All you have to do is a quick search on here to see all the comments about Viper sounding just like a Virus, before the DSP56300 emulators became available. Now we can load a ROM file into a software synth and give it the benefit of the doubt. You can even look on here at the development threads of these DSP56300 emulators where people were saying one version sounded different than another. Code optimization being done to unburden the whopping cpu hits they first had, etc. Simply loading a ROM file is just that, loading a file. This thing could be Viper with a different GUI and a .bin file requirement to allow a instance to be loaded by the DAW and people would try claiming the same things.

Put a dither plugin on a 24bit audio file in your daw and lower the bit depth. Does it still sound identical, because I did not alter the file loaded onto the channel?

 

It does not mean "identical" code is going to result in identical sound, and that the user saying he noticed a difference is an impossibility. Even a supposedly perfect dsp56300 rom emulator is still only as accurate as the emulation, timing, in/out. If the differences or "errors" are not obvious; he could still notice differences using effects like reverb, the various filters, the vocoder, other modulation.

It requires a good amount of assumptions to say his observation is wrong out of hand, without any actual sound files as proof.

 

Mate, are you serious now? If it nulls it is identical. It's perception that can play at fault. You can listen to the same audio file couple of times and it might sound different based on your changed mood or your focus