A "1" on a d20 is a 5% chance.
But your logic follows my general movement in gaming, as I've grown less fond of games in general where "the cops showing up" don't mean anything. Even in D&D, the "cops" or town guard or whatever, very quickly become meaningless threats, and it shows with how prevalent murderhoboing is in the system. In systems where "the cops" (average joes with average weapons and armor) are a very real and present danger outside of the most insane of characters, I find players behave less like gamblers and more like rational people.
I get the feeling that there's a happy medium in dice-pool systems somewhere between 3 and 6 dice. I've generally found that large dice pools don't really produce that much more success. Sure, 5/15 successes can really wallop an enemy, but it's still statistically about the same as 2-3/5 successes. Some kind of reduction in the number you need for every 5 dice you'd have beyond 5, I dunno or something, I dunno. More dice, like higher modifiers, should mean more success, but it really doesn't.
And I find 7+ dX to be kinda wonky to hold anyway.
As above, for the people who understand this math, and I mean REALLY understand it, unless there is a flat 0% chance of success or failure (depending on the case) it leads to poor decision-making via gambling. When a d20 has a 5% chance to roll a 1 and a 5% chance to roll a 20, even if your only chance of success/failure is on one of those numbers, I find people more likely to gamble at "winning big" rather than take approaches that may lessen the difficulty but consume resrouces.
As I mentioned before, when I tested out 2d10, it was a much smoother game, with far more successes for the players, but far fewer "OMG A NAT 1 I'M SO DEAD!!!" and far less "OMG A NAT 20 I WIN OMG OMG OMG!!!!" Which is an unhealthy (IMO) gamblers mentality.
Not to say risks shouldn't be taken, but if you're always gambling on the "big wins", the game itsself becomes background noise to the extremes.