Edit: The short version of the long post below is, I believe that if one of us won the lottery and handed @jofemodo a budget that covered a couple years of his and 2-3 more talented people’s full time work, that he would deliver something every bit as good as the device being hyped in OP, and do it for less.
This is true of so many. My friend Scooter, currently a Eurorack addict, said the other day “I’m glad we cannot access life statistics like time spent setting up gear vs time spent playing said gear.” We would both go down as huge fails, I think, especially in comparison to some of my three-hours-of-scales-per-day bluegrass friends.
No, synthesizers are no indicator of musical ability, and there is indeed an entire subculture of people IMO who are doing something with them which is one degree removed from musicianship. Teenage Engineering is a great example of this - they make these weird Fisher-Price devices and charge ludicrous sums for them, the build quality is not there, and it’s thoroughly enshittified, calling home to their servers, probably putting all your performances through their AI training.
Mostly, though, they make toys, not instruments. But if Teenage Engineering is drinking a six pack and smoking a joint at a party, Eurorack is fentanyl. Yes, that is how synthesizers used to be done, because - and this is the important bit here - there was no other way of doing it. We moved on from that model quite joyfully, much as society moved on from war with muskets. But some people like putting on Union and Confederate uniforms and running around in fields shooting blanks at each other.
As to synthesis power, I think that’s a nothingburger. From the looks of it, you can stack up 16 voices with that Montage device, and yes, that is some serious CPU muscle under the hood, to be sure.
But, supposing you only need two, three, maybe four of those voices, not all sixteen at once?
I think the computing gap, if you do an engine-to-engine, single-voice Pepsi Challenge, is not remotely as stark as you’re making it sound. If we could look at the code we would know, of course, but we have to take their word for it that their “advanced” algorithms are actually some new, deeper technological leap, with accompanying new, faster technology, vs. being merely representative of another few years of work on the code and the standard Moore’s Law leap in available compute that you would expect in the number of years since the last version.
This isn’t even so much me being a Zynthian zealot as me pointing out that these “elite” devices are no longer really that elite, now that everything is a digital model. The engineering behind the analog devices of the 80s was genuinely impressive, and indeed those devices had some real problems when they went bad as well. But engineering the original Prophet 5 was absolutely the work of a lot of strong brains.
The Montage was more a set of practical concerns. Decide on feature set, spec out a CPU and supporting chips that deliver that feature set, do circuit design just like Jofemodo does, put the software together, most of which is the same software in the last three to five devices of this class, with a few updates.
Sequential Circuits, Moog, they had engineers. Modern devices only require IT guys and programmers. The difference between Montage and Zynthian is available budget and human hours.
PureData vs Max/MSP - is one better than the other? Not really. If Max had a genuine advantage, I’m positive the organelle would use Max instead. We are moving away from the age of needing capitalists to get things done, hopefully, cause they really can’t get things done remotely as efficiently as a bit of open hardware and the community of Libre devs (pianoteq excepted) who provide all the engines in Zynthian.