Happened upon a page on UXMovement.com where one contributor posits a "better" order of OK/Cancel buttons, and one comment from a Penny stuck out:

I’m impressed – I found this to be an exceptionally well-reasoned and well-written article. The reasoning rings true with my overall life philosophy… Doing anything solely because “that’s how it’s always been done” is a surefire way to guarantee stagnation and the halting of progress/advancement/improvement.

@Chris McGrath: What you’re talking about is pandering to the stupidity of the masses. You gain a user’s trust first and foremost by developing a useful and stable product. If well-reasoned differences in UI design could cause so epic a who-moved-my-cheese devastation in users, then said users wouldn’t last long on this ever-evolving and adapting planet in the first place and would most likely be up for a Darwin Award in the very imminent future.

Since the moderator has not yet seen fit to post my response, I provide an expanded version of it here.

Pandering
Hmmm... "pandering to the stupidity of the masses"—I find that a disquietingly condescending statement, suggesting that the speaker is somehow superior to the masses referred to. This is not the kind of person I want deciding my UX. I would prefer somebody with empathy for the masses, rather than so flippant and dismissive an attitude, and one that doesn't so readily poison the well of an opposing viewpoint.

Alternative View
I agree with Penny in that this page is well reasoned, and well-discussed. I do, however, beleive the thesis—that of reading left-to-right necessarily dictates placement—to be flawed. The presumption is that as one progresses through the elements of the dialog/form/webpage, that one is moving the pointing device to keep up with one's center of attention. Were that true, the argument that OK should be on the right might hold more sway. Absent that, it's just a matter of battling theories, neither one of which is demonstrably superior (or else there wouldn't be as pronounced a bimodal distribution of opinions.)

A countervailing theory—my take, and that of my colleagues—is that the user's attention is essentially a cone or pyramid, and that the most important elements need to be presented as close to the center of that cone as possible. Arguably, the OK (being the affirmation of the interaction and the far more likely choice) is reasonably placed closer to the center of attention; the Cancel, being the denial of the interaction, may be placed farther in the penumbra of attention.

As an aside: from a practical—not so abstract and theoretical—standpoint, most users still use a pointing device, be it mouse, track ball, or even digit touching screen. That pointing device is most probably in vicinity of the most recent edit, so the mechanics of moving that to the OK would justify the OK being in the more central location, affording an economy of movement.

Innovation vs Novelty
I agree with Penny, insofar as I abhor the unconsidered status quo, and I agree that innovation is vital. Let's break new ground, discover new things, and make the world a better place. Innovation is necessarily change, but change is not necessarily innovation. I don't think we need to figure out new and improved ways of suggesting our consent: that much is settled, given the affordances we have. As an aside, I'm reminded of a scene from the movie Jabberwocky, where the young apprentice Dennis goes into the blacksmith's shop, and citing "improved efficiency" takes it upon himself to rearrange things, to catastrophic effect. There are gnarls and burls in everyday life. Personally, I hate that the Reverse gear in my automobile is forward of Neutral, and that Drive (the Forward gear) is to the rear. But citing the cognitive inversion is not likely to get the automotive industry to change that: people are used to it being that way. Penny may counter that this, too, is pandering to the stupidity of the masses.

The response from Detroit/et al.: deafening silence.

The QWERTY keyboard is another one of those strange artifacts of history, originally designed to impede typists. The Dvorak keyboard is far superior in terms of words that can be typed with letters on the home row. But for every Dvorak keyboard out there, I can show you ten thousand QWERTYs. Stupidity of the masses, or settled art, this too is unlikely to change.

Penny-Wise Future
Regretably, Penny goes on with more contempt for others with her darwinistic references. Maybe the future is headed for such dystopia, and the masses are held at the whims of ivory-tower UX designers.

But my vision of the future—where true innovation lies—is that the machine will conform to the desires of the user, not, as at present, the user must conform to the desires of the machine (or, more precisely some removed designer making unilateral decrees of what is best for them.)

My vision, and that of my colleagues? The User is Boss. This is as it must be. Do you want the OK button on the right? Fine; you, the user are master; let it be so. On the left? Excellent; your wish is my desire. The current one-size-fits-all mass-produced ethic is better suited to the wearysome factory of 19th century, than to the computer-filled 21st. I've always been enthralled by the things that technology can do for us, but I have no desire to be a thrall to that technology. The truly innovative designer will figure out a paradigm that facilitates this, rather than quibbling over two UI affordances that perhaps in a decade will be rendered moot.

The clever designer may also concede that, as the shifter on an automatic transmission, and the QWERTY keyboard, there are unfortunate casualties that only the most quixotic (and unimaginative) designer will champion, and instead look to the next big thing that renders the whole tiresome conversation irrelevant. From a practical standpoint, Windows boxes far outnumber macs and linux, so though it may stick in some UI designers' craws, perhaps OK/Cancel is the way to go... but only until the truly imaginative and innovative figure out a whole new way of relating to the electronic world than by a matrix of momentary-contact switches, and an illuminated analogy for a sheet of paper. So let's move on from this pointless debate, and tackle that problem. That, my dear Penny, will be real progress/advancement/improvement...