KidSnide
Adventurer
The big thing for me is derived ability score mods (10-11 grants no modifier, 12-13 grants +1, 14-15 grants +2, etc.). I have come to dislike this mechanic, for two reasons. First, the modifiers are too small compared to the size of a d20. If you take an average person (Int 10) and Einstein (Int 20), and have them both make Intelligence checks, the average person will beat Einstein slightly more than 1 out of 4 times. That's way too big a variance. Second, derived mods are confusing to newbies. I would much prefer to have ability checks be "roll 1d20 and add the ability score." So if you have Int 14, your Int check is 1d20+14. It's simpler, it solves the odd stat issue, and it means high stats really mean something. But it's rather too late for this now, I'm afraid.
Agreed. We've had this mechanic for a while, but at least 3e (and in a different way 4e) mitigated the non-ability check version of this issue by providing hefty skill modifiers. Now that we're stuck with tiny proficiency adjustments (the lucky rogues and bards exempted), this problem shows up all over the place.
Early on in the process, WotC decided that it was less fun if skill checks (and "skill-like" ability checks) were auto-successes or auto-failures depending on the skill (or ability) of the character. Maybe this results in a more "fun" result for some tables, because everyone always has a chance of success. But it makes total hash of reality. Most tasks that are challenging for trained professionals (jumping, surgery, stealth, code breaking, historical knowledge) are essentially impossible for untrained amateurs of average ability.
It's like they decided that realism was the problem with skills. Personally, I agree with them when it comes to games with an action movie sensibility. I don't want that kind of realism when I play Dungeon World. But a lot of my D&D games are in a more grounded universe, and I don't like it when the skill rules produce essentially nonsensical results.
-KS