The Forgies have a fairly extended discussion/debate about the place of randomness in gaming. I'm not about to cover even a tenth of that here, but I did want to zip down a few ideas on an aspect thereof.
I've been giving some significant thought to the idea of Fortune at the Front, the idea that one makes rolls/tests during the initial phase of resolution, and only after are the results reified into description. Note that this is explicitly different from the usual course of events, Fortune in the Middle, where you declare intent, narrate attempt, work out what the rolls/tests should be, resolve, then narrate the reification of the test.
Exempli gratia:
Fortune in the Middle:
"I'll be attacking Pertwee this round."
"Thrag swings his mighty axe at Pertwee, screaming his clan's battle-cry: Yippie-ki-yay!"
"Pertwee aborts his action to dodge the axe."
"OK, that's Dex plus Melee(Sword) plus +2 for the battle cry versus Pertwee's Dex plus Dodge(Melee). A hit! Damage comes out to ... a moderate wound."
"Thrag's axe bites deep into Pertwee's shoulder as he fails to get out of the way in time!"
"Thrag and Pertwee are fighting, so, conflict of Warfare?"
"Thrag gets a Superb, Pertwee only a Good. That's two steps different."
"Thrag swings his mighty axe at Pertwee, screaming his clan's battle-cry: Yippie-ki-yay! Thrag's axe bites deep into Pertwee's shoulder as he fails to get out of the way in time!"
Thrag has Warfare 5, Pertwee has Warfare 2. We use a modified Sorcerer/Donjon resolution, and they both roll their Warfare in d10s. Thrag ends up with 2 dice which beat Pertwee's highest of 6, and Thrag's highest is 3 greater than that. So, not only did Thrag hit Pertwee hard, but extremely well.Reifying the results of the test with this kind of mechanic is considerably easier, if only because one has the material to work from to make the result sound meaningful. Further, its easier to do so with Fortune at the Front-style resolution, because you don't begin to set up for a very specific action, only to get a result that doesn't reify sensibly without a stretch. That is to say, it feels less gratifying to kick off an action with an elaborate description, have the dice come to an awkward result, and then reify it with an unsatisfying change. If Thrag ends up fumbling his axe attack in the first case, it sounds more awkward and less satisfying in retrospect, rather than finding out the test was a fumble and setting up to describe one in the aftermath."Thrag's axe comes screaming into Pertwee like a banshee; it bites deep, catching the smaller man completely unawares."Had Pertwee and Thrag tied on their highest die and Pertwee beat Thrag's second highest only by 1, then it'd be a much, much closer event."Thrag's axe swoops in, from Pertwee's blind side, but at the last moment the mage's head turns and the axe only leaves a shallow bleeding gash along his cheek, hairs' breadth from his exposed throat."
One advantage is that 'winning' allows a great deal of control, which narration then flexibly outlines. That is, great, the mechanics tell you you win. But HOW do you win? That is a second issue, and separate element.
I think this solves some of the problems people have about 'rewarding good description.'