Andre_geo: I had to retreat 3 times
Fenixp: Yup, that right there is exactly why retreating should be as dangerous as possible. While yes, stress penalties exist, they're just not severe enough to dissuade player from "I'll just try out the next encounter and see if I really need to abandon this quest" without giving it a second thought (perhaps save for having high enough stress level to risk a heart attack) and you're quite simply not forced into retreat often enough for stress to be much of a factor.
But... Well, if you dislike it, fair enough, that's the point of difficulty being adjustable after all :-)
I don't know. It's "risky" on paper. But what happens is that you're not actually taking a risk by retreating, instead it's giving you another chance at avoiding complete failure. Or, when you decide to retreat, you basically know you're going to fail. Possibly you've already lost a minion, etc. And like mentioned, retreating on room-battles might allow you to go another route to a different battle, or you can do the battle again and maybe get some less annoying mobs to fight.
So the only situation you're actually going to be using retreat tactically would be at the very beginning of a battle. And then having retreat possibly failing several times makes sense. While when I'm using retreat, it's to avoid losing the loot I have, and to save perhaps one or two of the minions. So then having multiple retreat failures usually means repeated death's door saves.
Or, having a retreat function like that opens you to attacks for so many rounds that if you want to calculate your chances, you need to retreat before you actually don't have a chance to win. That's basically the problem with it. I would probably leave it on if you for example had.. I don't know... 75% chance at failure if you've used three other characters this turn. And then maybe 25% if you've used none. And then if you fail the first time, then the failure chance could have been lowered by 20%. So you would never go more than 3 rounds, depending on how many characters you've used, and how many defensive abilities you've favored that you normally wouldn't, etc.
In other words, you basically remove the risk aspect, and replace it with an extra set of potential failure conditions outside the battle itself.
It's the same with the critical hits - you need to fight so many battles, and you are guaranteed to take several hits. So giving the ai a chance to wipe out the team in one battle in the first round is a metacondition that doesn't actually increase the risk, it just introduces an extra instant failure condition. Which.. would make sense if you could raise your heroes from the dead, or perhaps save them if you win the battle, and so on. But like it is now, it just adds another scenario where you can fail without really having taken any extra risks.
And it's weird to have that when you tend to be forced to put lower level characters in a group with higher level characters on a high level mission, rather than the other way around.
So the impression I'm getting is that the designers have been looking at people who play the game "safe", and who game the system to get relatively higher level minions into the lower-level challenges. And then these "punishment" conditions make sense, right? If you're actually overleveled, and are able to always heal, or possibly disable the entire enemy team every round. And that's.. maybe how a "veteran" player would play the game, after having played the game through the beta, etc.
But if you're playing risky, and want to take on new higher-level challenges as fast as possible. And gamble on getting in damage with a gravedigger/dodge setup while the other characters defend themselves or heal as much as possible, for example - having these extra failure conditions is basically just an extra loss-condition that eventually is going to beat you, no matter how good you are at calculating the odds.