In 1E, the ability to take a swift/immediate action "refreshed" at the end of each turn. In 2E, the ability to take reactions refreshes at the start of each turn. Although the 2E version is conceptually simpler, I think it leads to some undesirable situations.
In 1E, I can cast Feather Fall the moment I start falling. In 2E, I can do that... if I didn't happen to use a reaction on my own turn.
Or, just look at the stat block for a balor (as an example). It has two reactions - a standard attack of opportunity and one that triggers when it kills a creature. So, now, if it kills a creature and uses that reaction, it suddenly can't make an attack of opportunity for the full round? That could be crippling if some important spell gets cast and the balor "lost" its attack of opportunity because of this. If reactions refreshed at the end of each turn, then the balor's attack of opportunity would get "precedence" in a way, allowing it to respond to enemy actions before getting back to its turn and deciding if it wants to use that special kill reaction. If it had to make an attack of opportunity, then it just can't use that kill reaction - but that seems to me like a much better way for things to play out, since it puts a more sensible choice in the creature's hands ("Do I attack now and then not be able to use this other thing" vs. "Do I use this other thing now and then just hope nobody does anything I need to interrupt before my next turn").
Swift/immediate actions refreshed at end of turn for exactly this reason, always allowing you to "sacrifice" a future swift action to take an immediate action now. It seems weird to me that reactions don't work this same way. Reactions that actually react to somebody else shouldn't have to be sacrifice in favor of self-triggered reactions, but rather it should be the other way around.
Other than this, the new action economy looks pretty fantastic to me.
