Are you sure that the design changes between the 1.x to 2.x were "popular" with the 1.x userbase when they were originally introduced? I don't remember it that way. And the parent article here gives a little snarky head nod to the fact that we went through a lot of this emotion before in the 1.x to 2.x transition. It is really worth going back and taking a detailed look at the 1.x to 2.x transition...really look at the initial reaction to the changes being planned and implemented which fell out of the usability study you hold up as a good idea. I bet the reaction really was not all that great and not all that different to what we are seeing this time around.
Moreover, throwing money at the usability issue to purchase additional expert manpower will not solve the perception problem. Changes are disruptive...even changes that adhere to expert state-of-the-art usability design considerations. Because fundamentally the userbase really does not appreciate what is and is not good usability in the same way that those trained in the art and science of usability do.
Experts in usability think differently about usability than untrained users do. And its far from clear to me that we as a userbase appreciate or even understand the value of usability experts. As users, we like what we like, we use what is familiar, we are seldom prepared to stop and think through the issues of usability as a product design exercise to meet the needs of anyone other than ourselves.
There is absolutely no way that you could take my personal preferences, design a product based on them, and then get many other people to enjoy using that product. And here's the secret that nobody in the linux using community wants to admit to themselves publicly. I'm typical for a linux user in that regard. As a breed we are conformist in our non-conformity. We are stereotypically individualists to a fault. Catering to any one of us, means not catering to vast sea of other permutations of personal preferences.
And... unlike actual shipping retail product lines, where changes can be introduced at physical product boundaries and not as software upgrades to existing physical devices...we've chosen a model where software changes are more fluidly applied to existing computer devices. We've created an additional mental burden for ourselves by taking advantage of the ability to extend the useful life of any computing device by continually upgrading its functionality with new software. But the cost of that is disruption to workflows as new software development moves forward and new UI concepts are introduced. Introducing new UI at time of purchase of a new computing device is less of a burden on the user, because there is a honeymoon period with a new device where users are willing to work with new UI as part of getting familiar with the new hardware device itself.