Alan Perlman endorsed the Timewarp 2600 quite a few years ago, when utter accuracy wasn't really expected from plugins, so it's hard to infer how the current version is likely to stand up against either the Arturia version or a real 2600.
Also, even with ARPs of the same version, there were variations in the sound between individual synthetic. Joe Zawinul had two which he used for different sounds because of these variations
Yes vintage synths did'nt sound excactly the same, even if it was the same model. Also in many cases different models was made. (ARP 2800 has quite a history.) Some artist also made personal changes to the instrument.
That's a reason why i believe the characteristic sound from a synth is the most important thing, when we talk about emulation.
Even if the sound can be a little different, then there still is some characteristica that they share.
Also i believe sounds that made a synth famous is a thing to emulate. Most people only know the sounds from music.
If a emulation can emulate the characteristica and the sounds that made it famous (prefereble using the original parameter settings) excactly or very close, then i would say it's a good emulation.
Among the characteristica can also be playmodes like the speciel way the filter envelope in a minimoog work, and that Arturias version i think is the only emulation that try to emulate this behavior. Other things can be important for the characteristica, including how modules in a synth work.
But yes it's hard to compare without having the model for the emulation. Even though you could compare with all the models. Arturias V3 have only the 2601 GUI that's the last, and the filter in that model is different. The black models with the Moog filter they had to stop using, i think has the most valued sound.
I think Arturia perhaps try to emulate all the models from the blue model to the black/ orange, as the filter has many possibilities. A emulation can try to emulate both, and be compared with all original models. But yes it's perhaps not fair to compare with something it don't try to emulate.
How would you compare? You don't think it's possible?