Failure is Advancement: A House Rule for Unknown Armies

I've been thinking about my UA game and listening to the recording as a kind of after-action report, and I've come to a few conclusions:

1. PCs are going to suck at most skill rolls in a stress situation unless they're either really high or are fore an Obsession skill, and
2. Advancement, while regular, is pretty slow. 

Neither one of these is inherently bad, but in concert they can be frustrating.

For example, Adrastia has a Drive skill of 25%. That's not the best in the group, but it's pretty good and better than the default score of 15%. Now in Unknown Armies, there are three kinds of rolls: Minor, Significant, and Major. 

• Minor skill checks are trivially easy and the kind of thing that we do all the time, like driving through traffic. With a minor check, if you have a 15% in the skill, you succeed automatically, so you'd only fail if you were learning the skill, didn't have the skill, or were impaired from drugs or damage. 
• Significant skill checks are harder than trivial but there's no stress to them. These are what I call "job checks" because you do them on a regular basis, like at your job. If you roll under your skill then you succeed strongly, but so long as you roll under your stat, you succeed weakly. The average stat in a UA game is 55%, so that gives slightly better than average chances at success. 
• Major skills checks, though, are serious life or death things. They happen in combat and any other kind of stressful situation where a screwup could mean injury or worse. Players MUST roll under their skill rating to succeed, and this is where disaster lurks, because while Adrastia is better at driving than most people she isn't a trained combat driver, and in situations where such driving is necessary she has a 75% chance of failure. 

Like the combat rules, these numbers are harsh and probably realistic. It's one of the things which make UA a horror, or perhaps terror, game. There's a good chance that if the shit hits the fan, you're gonna die. 

Now all of this is cool. I can enjoy high-stakes games, and UA characters aren't especially hard to make. But I wonder, I worry, if that realism paired with slow advancement and limited XP will result in players frustrated at their inability to get things done. 

So I think I'm going to take a cue from my game Unknown Ponies: Failure is Awesome (which, ironically enough, is based off the UA game engine, hence the name) and use the concept of "We learn more from our failures than from our successes" to give my PCs a slightly faster form of advancement that won't make them too powerful too quickly. 

Here's the rule: Every time a player character fails a Major skill roll, the player puts a check next to that skill. There can never be more than one check next to a skill, so multiple failures earn nothing. At the end of game session, each skill with a check next to it increases by 1%. 

Here are the limiting factors:

1. It has to be a Major skill check, meaning in combat or similar. 
2. The skill only gets one boost per game session. 
3. It cannot raise a skill above the hard cap of an attribute stat, and those can only be raised through XP. 

Yes, I am aware that this can be abused by player fishing for skill checks in combat. If they want to waste their turn whoring for advancement at the risk of failing the mission and/or the lives of their characters, that's on them. 

I expect there will be rapid improvement in low-level skills that get used a lot during stress situations -- as is proper, I feel -- which will taper off once the 50% level is reached. Of course, if the players spend XP to improve a skill further, that will only accelerate the drop-off. 

There may be other ways to game this system, but I'm not thinking of any. If you can I'd like to know about them.