Steve,
I can't say your arguments are without merit. OTOH, how many times in the history of computing have we heard, "Nobody will ever need more than X?" Yet somehow, we keep managing to find ways to use the excess capacity. You may be right that we've finally reached the stage where the next increment of better is not worth the cost, but history suggests otherwise.
Sure is hard to argue with history..
I think you're right in that we manage to keep using more power as it's made available. And probably most of this is progress in our ability to do a better job (of whatever we do) in a more attractive environment. But history also shows we progress in cycles where we're computer heavy and then software heavy.. and lately it's been mostly hardware heavy. And software developers take advantage of this by writing sloppy or inefficient code.. which if you think about it is very much like stealing a portion of the computing power we're paying for.
As an example, lets say a fictional software developer decided to stop selling perpetual licenses and forced it''s vested customers into expensive subscriptions. Not nice. But now, you find out their code is sloppy and inefficient and where if written efficiently you would have been perfectly happy with the software's performance on a MBP.. but because it isn't you will now need $10,000 worth of new Macpro stuff to run at a satisfactory speed? History suggests this is happening pretty much with most major developers.
Sometimes sloppy inefficient code is rushed to market to compete with another fictional character.. and we notice glitches, install problems, sticky sliders, maybe a certain brush is very slow, miscslow performance.. but we can make up for most of that with pure computational power (that we pay for). And then the next non-release version comes out and performs much better.
How much did this fictional company save by not bringing on enough programmers to meet their competition imposed deadline.. that of course was never passed on to the consumers who blunted the effects of their negligence with the power of their hardware.. or worse their limited patience stores of today's stressful world.
This is all part of doing business for modern software developers and I'm okay with it. At least until they get too damn greedy and take advantage from even more fronts. Eventually it breeds ill will and even anger and general distrust. Not what you want your customers feeling.
But back to is the new MacPro overkill for the overwhelming majority of still photographers when we have such wonderful and economical tools as Haswell and Ivy Bridge available at a fraction of the price and in a much more expandable and user serviceable platform? Hmm.. Let me ask you this. How many still photographers have spent the money on workstation GPU's? Some I'm sure, but how about TWO of them? See how I'm getting the "overwhelming majority" level?
A more common bit of advice we're telling still photographers is with today's integral graphics (to the CPU) they can get by with just the integral graphics for average workloads.. and a wonderful example of this is the new 4 core Mac Mini.
I recently threw a 256gb Samsung 840 Pro into a 2.5g model (with the help of OWC's kit) keeping the 1tb HDD as storage, upped RAM to 16gb for under $100.. and what a screaming machine that is! It's an Ivy Bridge 4 core, 8 threads... everyone who used it couldn't believe the speed from a now $900 mini.. One client wouldn't take no for an answer and ended up taking it home. He didn't want to build his own, or have me do it. He wanted that specific one like it was magical.. This was the sort of reaction I was getting when current MBP users tried this machine.. they haven't had that level of Mac Power available outside of a pricey Macpro..
So I really think.. Video Professionals will be well matched and served with a new Macpro. But still photographers? There will be exceptions with their gigapan heads and who stitch a zillion images at a time.. but for the vast majority of still photographers it's up, over, on, and just plain beats down the point of marginal returns. These people would be better served, overall (economics and performance) by a Haswell or Ivy Bridge platform at a fraction of the cost. And get a much more expandable and user serviceable platform in the bargain.
We live in great times..