Taking a look at the skill documents.
The first document on skills....my main critique is it doesn't factor in advantage, and I find competent groups find all sorts of ways to get advantage of skills, the help action being the easiest example. Then there's guidance, which is such a strong cantrip I consider parties without it the exception not the norm, so that's an extra +2.5 right there on many checks.
In other words, I think Mearls' math underestimates the bonuses that even low level parties will commonly have access to.
I also argue that DC 30s as impossible are anything but in dnd. Again with various other skill bonuses and your highest stats, I routinely see DC30s getting hit in a game, and I think 5e has routinely underestimate the "impossible" nature of DC 30. I think DC 35 is the true "impossible" DC range.
I think people are making a mistake when looking at maximum rolls.
Of course you can make a DC 30 check some times. But is it something you can achieve reliably?
Even a rogue with reliable talent and expertise struggles there.
10 (reliable talent base) +12 (expertise) +6 (attribute of 20) will only be 28.
If you allow guidance to be added after the roll (i guess that is how it works), you get to 30 75% of the time.
With advantage you are actually a bit better off as you roll higher than 10 75% of the time.
Does it seem impossible for the level 20 rogue? Hell no. But in the real world jumping close to 9 meters or running 100m in less than 10 seconds or 42km in less than 2hours seems so too.
But the best of the best can do so quite reliably.
In my opinion, you should rather look at the chance to fail. Not at the chance to succeed.
I am able, if I can do something but sometimes fail. Seldom on easy tasks, especially if I am naturally adept at certain tasks (above average ability scores (12 to 15)
So an easy DC of 5 seems right. At level 1, i can do them most of the time (85 to 90%). Sometimes I fail. If my natural adeptness is not that great (less than 10, I fail about twice as often (75%). Which still seems ok for an easy task.
I am base proficient in something if I can do easy tasks without chance of failure.
DC 5 does so (prof +2, expertise) .
The next harder task is DC 10.
The not very talented guy is doing that tasks 50% of the time. A coin flip chance.
The base proficient character has a +4 bonus. So a chance to do it of 75%. Also quite good. But we see, they sometimes needs two attempts to do it right.
In comes the lvl1 expert character: +3 from stats, +4 from expertise. Now we are looking at a 90% success rate. For a level 1 character, not bad at all. But still tense, as something bad can still happen once in a while.
So this should really be or medium difficulty. 15 is already hard for a level 1 expert. So if your game wants to encourage people doing mundane things without the help of magic, there needs to be a high level of reliability. Using DC 15s for normal tasks does not do it.
I do like mearls' level of success idea. Which is what I already use for a while. If you don't fail by 4 or more you succeed. But you pay for it with some problems.
I would also like it if prof bonus started at +3 and ended at +8. +2 is a tad too low in my opinion, even at level 1, and I'd like slightly earlier progression:
1: +3
2: +4
6: +5
10: +6
14: +7
18: +8
This progression increases at the same rate, but you get a head start compared to the 5e one. Makes proficiency mean more at low level.
Expertise would probably be a 1.5 multiplier instead of a *2 multiplier.