The difficulty with revolution, at least since the advent of professional armies that could easily defeat any peasant force, is that of target selection; people who won't line up for the Maxim guns. You can't kill them all, because you need some of them, but at the same time they're hiding among the people you do need. What's the acceptable false positive rate?

Now we're talking about a situation in which everyone can be watched for suspicious behaviour by computers, every minute of every day. Where you need fewer and fewer people, so your acceptable rate for false positives is going up rapidly. And in which the military force that can be applied by per dollar is increasing very quickly.

At an extreme, if you can automate everything, you and your mates hold up in a bunker and just gas every other living thing in the country. Before that, you send your robots into the ghettos if the peasants start getting uppity and kill every second person or something like that - you don't have to worry too much about false positives there because the people you need are a bit richer and get nicer digs. Before that you drag those who defect up before the courts and there's no revolution, there are just criminals harming the wealth of society - the war on drugs writ large.

You don't get a civil war when the disparity between the rich and the poor is very great. The poor don't get to organise and arm. They just get arrested at the low end and slaughtered at the high end.

No-one's going to get into a field with you and have a fair old dust up. At least, not unless the dust up involves nerve agents and ghettos.

What about government backed by robotic AI? At this point you no longer have cops and soldiers who are part of the masses, and thus feel empathy to the 99%, but you have robots that will do whatever they are told.

Robots change precisely that.