The Smurfiest Wizard Ever!
While a few people already responded, I have to say there is a big difference in every respect. Doing a walking handstand does require a lot of balance, but it also requires enormous strength. Few acrobats/gymnasts are weak by any standards. There are many times when Strength (Acrobatics) would be more appropriate than Dexterity (Acrobatics) and many circumstances when Dexterity (Athletics) is more appropriate than Strength (Athletics). That is why our group "unlinked" ability scores and skills. As the core books describe, everything is an ability check and sometimes a skill applies. It is up to the DM to determine when.As for passive skills, it works great and makes much more sense than the system as is. First, observant makes you better due to the +5 bonus when you AREN'T looking than when you are! How does that work?!? Secondly, if players state they are looking then everything works normally anyway, and if they aren't, a good passive score gives them a chance to notice something even when not actively looking. Nothing should be automatic in D&D IMO and it works well for our group. No one, and I seriously mean NO ONE, has ever complained or felt using passive skills RAW makes sense.The question isn't if you see a big difference between Athletics and Acrobatics. The question is if MECHANICALLY you see a big difference between Athletics and Acrobatics that aren't covered by the differences in ability scores? Many of the checks needed (escaping grapples, spells, etc.) right now can use either. And a bunch should. For example, right now someone with expertise in acrobatics with a 20 DEX has no indication that they can climb. That's STR (Athletics). I've literally had two different players curse that their uber-agile elves most likely would fail at climbing a tree or wall.This is double jeopardy. Just shy of half the people who could detect something will never get a roll, and of those that get a role many will still fail. Use one or the other, otherwise you are drastically changing the success/fail percentages in a way that messes up the math of the system.