From Dave Sirlin's coverage of GDC:
Day 1: See Rob Pardo's talk, paragraphs 8-9. Basically, players whined about randomness meaning that sometimes unlikely events happened, so they changed it so when you fail, the chance of success increases.
Day 2: See Sid Meier's talk, paragraphs 8-11. Basically, players whined about randomness meaning that sometimes unlikely events happened, so they changed it so when you fail, the chance of success increases.
It's really interesting that these guys independently came up against the same problem and came up with the same solution. I'm in two minds as to what I think about it. My initial reaction is: this is bad, they're dumbing things down, they should leave things done properly. Then I back this up with some solid reasons why it's harmful. But then I'm not sure.
So first, why is this harmful? Games are educational. One of the reasons humans play games is that it fulfils our desire to learn - our brains are built for learning, so we find it pleasurable. And one of the things games are able to teach is HOW PROBABILITY WORKS. Probability isn't intuitive. Sometimes you don't succeed, even when the odds are in your favour. Sometimes exceedingly improbable things do happen. (It's even worse in the infinite case - I've seen people have serious trouble with the idea that an event can have probability 0 and yet still occur, or probability 1 and not definitely occur.) So by twisting the probabilities in your game to line up with human intuition, you're preventing your game from teaching a valuable lesson - and, more importantly, you're potentially teaching a false lesson.
(
Here is a slightly related and slightly unrelated article by Tom Lehmann, designer of Race for the Galaxy, just because he's cool.)
But on the other hand, it's not really the case that randomness-where-failure-increases-your-chance-of-success is "fake" randomness. It's still random, just the random events are no longer independent, which is perfectly okay, both from an abstract mathematical perspective (it's still a valid formal system of game rules) and maybe even from a real-world-intuition perspective (plenty of real 'random' events aren't independent).
The popular board game Settlers of Catan suffers from problems with streaks of bad luck - resource generation is determined by a roll of the dice, and it's common for some players to miss out, or for certain resource types not to be generated much, sometimes resulting in a slow and tedious game. A fix for this which some of my friends use is to have instead a deck of cards with numbers in the same proportion as the sum of die rolls. This has essentially the same effect as what Rob and Sid described - if a number doesn't come up, it's more likely to next time, because the deck gets depleted and there are more copies of that numbers left in it. And this I have no problem with, partially because it's fixing a problem with game balance, rather than trying to conform to the intuitions of innumerate players, but also because the mechanics are explicit - you can see the cards being drawn from the deck, you know the card you drew is no longer in there, so you know you're not dealing with independent random variables.
So I guess what it comes down to is how it's communicated in the game. I haven't played either of the games in question, so I don't know how these things are communicated. If the interface gives the impression that the events are independent, but is secretly fiddling with the odds, then this should be considered Harmful. If it's clear that they're dependent, it's fine. For example, in WoW, if you're hunting bears in a forest, some of which have paws, it makes sense if there are a finite number of bears, and then finding a pawless one means that next time you find one there's a higher chance it will have a paw - you're drawing from a deck, not rolling dice. (I'm not sure what the reasoning could be in CivRev.. maybe a public "morale" stat that, counter-intuitively, goes up when you lose battles.)
Of course, videogames can hide their rules under the surface, and it's completely legitimate to say "you don't know what the rules are; a bunch of complicated interdependent systems affect everything that happens and it's up to you to experiment and figure things out". This is teaching something too - the scientific method.
