Your caveats seem to make the rest of the post a bit pointless, although it is very interesting! You say that dictatorships have not been stable up till now, i.e. in conditions we are used to. But the central xrisk concern about dictatorships is their stability in new conditions i.e. world government or advanced surveillance. The stability of dictatorships in the past or at present doesn't seem to have much bearing on their stability in the future.
Or rather, it seems to have bearing in the more limited sense that we can examine particular causes of past instability and consider dictatorship-stability without them. So for example isolation seems key. If the USSR was the sole state in the world, it would probably be around today. A major reason for its end was elite exposure to outside ideas, mass exposure to differences in living standards and the cost of competition with the West. If an isolated totalitarian state is able to keep a strong enough grip on its education, media and power structure to prevent people with alternative ideas from gaining power or influence, it seems like it could last a long time.
This seems to be the main problem with a singleton. I sometimes imagine the future as threading the needle between increased control and global coordination (to prevent nano- and bio-risks) and the possibility of dictatorship.
Orwell's Nineteen Eighty-Four is brilliant, terrifying and useful. It's been at its best fighting against governmental intrusions, and is often quoted by journalists and even judges. It's cultural impact has been immense. And, hey, it's well written.
But that doesn't mean it's accurate as a source of predictions or counterfactuals. Orwell's belief that "British democracy as it existed before 1939 would not survive the war" was wrong. Nineteen Eighty-Four did not predict the future course of communism. There is no evidence that anything like the world he envisaged could (or will) happen. Which isn't the same as saying that it couldn't, but we do require some evidence before accepting Orwell's world as realistic.
Yet from this book, a lot of implicit assumptions have seeped into our consciousness. The most important one (shared with many other dystopian novels) is that dictatorships are stable forms of government. Note the "forever" in the quote above - the society Orwell warned about would never change, never improve, never transform. In several conversations (about future governments, for instance), I've heard - and made - the argument that a dictatorship was inevitable, because it's an absorbing state. Democracies can come become dictatorships, but dictatorships (barring revolutions) will endure for good. And so the idea is that if revolutions become impossible (because of ubiquitous surveillance, for instance), then we're stuck with Big Brother for life, and for our children's children'c children's lives.
But thinking about this in the context of history, this doesn't seem credible. The most stable forms of government are democracies and monarchies; nothing else endures that long. And laying revolutions aside, there have been plenty of examples of even quite nasty governments improving themselves. Robespierre was deposed from within his own government - and so the Terror, for all its bloodshed, didn't even last a full year. The worse excesses of Stalinism ended with Stalin. Gorbachev voluntarily opened up his regime (to a certain extent). Mao would excoriate the China of today. Britain's leaders in the 19th and 20th century gradually opened up the franchise, without ever coming close to being deposed by force of arms. The dictatorships of Latin America have mostly fallen to democracies (though revolutions played a larger role there). Looking over the course of recent history, I see very little evidence the dictatorships have much lasting power at all - or that they are incapable of drastic internal change and even improvements.
Now, caveats abound. The future won't be like the past - maybe an Orwellian dictatorship will become possible with advanced surveillance technologies. Maybe a world government won't see any neighbouring government doing a better job, and feel compelled to match it by improving lot of its citizens. Maybe the threat of revolution remains necessary, even if revolts don't actually happen.
Still, we should refrain from assuming that dictatorships, whether party or individual, are somehow the default state, and conduct a much more evidence-based analysis of the matter.