With a D20, though, the numbers are as follows. A has a 35% chance, be has a 50% chance, C has a 65% chance, and D has 80% chance. You're dropping the low guy a bit, and raising the others quite a bit, and that's for a moderate challenge. You aren't just boosting proficiency, but you're boosting success rates. At least in the low and middle ranges. At the high ranges they drop considerably.
That master with his +4 modifier and +6 from proficiency/expertise has only a 1% chance to hit the DC 30 nearly impossible task, instead of the 5% the game would normally give him. During game play when I tell a player that he needs a natural 20 to succeed, he and the other players get excited and gather around to watch, because we've all seen the long shot happen. If I were to change to, "You need a natural 100(2 consecutive 10's) to succeed, they wouldn't get that way, because there's almost no chance of success.
I get the desire to remove some of the swinginess from the game, but it also comes at the expense of some of the fun in my opinion. Most DCs that will be encountered become easier for those with proficiency and expertise, reducing the challenge level of the game, while removing much of the chance of success from the dabbler, who then can't really participate in most challenges that they aren't proficient in or fall into their main stats. And the long shot success roll is pretty much kaput until high level.