I wanted to give this a 3.5, but rounded up because it's not half as bad as people make out, and I want Shadowrun to success.
Honestly, there was so much good in Anarchy, and I can see they've learned from some of it. But they made a lot of the same mistakes as the last two editions in terms of rushing the book out and not giving it a proper playtest or editing pass.
IMO, it would have been far better to shoot for an alpha playtest for their 30th anniversary GenCon, rather than a full release, and then they could've taken the time to file off the rough edges. Look at how Pathfinder, WotC (bleurgh -- OGL 1.1 was an awful idea) and even Onyx Path handle things. They take their time, sound the community out, and mostly deliver books that feel polished, well playtested and well thought out. Community reactions are nearly always much better, as well.
Whereas, companies like CGL (and White Wolf, when it was briefly resurrected as a game studio and not just a brand label) who rush books out tend to produce poorly finished and sometimes equally poorly received books as a result. I'd much rather the biggest controversy be whether signature character x's haircut looks good or not in the artwork, rather than whether a game is a dumpster fire or not. (And for the record, these games usually aren't dumpster fires, and were created with a lot of love, but love alone doesn't fix mistakes made by rushing and lack of resources.)
That said, there's a lot they get right in this edition. It *feels* breezier, and less intimidating, even if it actually ends up being almost as complex as prior editions in some ways (overcomplicated mechanics and, basically, the same crappy referencing and editing that we've got used to). The idea of just tossing out all modifiers and just using Edge would have been great, had they actually stuck to that. As it is, you have Edge, and then a bunch of modifiers that get used as well. Not to mention the extra step of comparing everyone's stats in combat to see who gets Edge in the first place...
6e is good for getting newbies in. I think, if you're an experienced (or even just confident) GM, you can hide the flaws and people will have a good time. If you're at all nervous and want to make sure you get things right, the holes will start to show, because as soon as you start looking for rule a on page XX, you're going to get lost.
If you want my advice, I'd probably just work out the average AR and DR based on those listed in the core rulebook. If your AR exceeds the average DR, gain +1 Edge. For every 4 or more over that average, gain another. Both sides can gain Edge this way, which evens things out, and doesn't require anyone to compare anything.
Next, remove the cap on Edge. Let characters accumulate as much as they possibly can on their turn, but only let them bank 2 points. That means if PC1 hypothetically generates 6 Edge in a social interaction, she can't drop it all to deliver a killer punch on her next go -- she has to spend the bulk of the Edge she generated on the action at hand. Carrying forward 2 Edge per turn allows your Edge pool to slowly increase, if you really want to save up and drop a big whammy, but most of the Edge you spend is going to go on the actions where you actually have an advantage.
I would probably also massively reduce the number of Edge expenditures. You only really need to re-roll dice, add dice and have some narrative editing. For the last, I'd have three categories: minor (gain a clue, coincidentally have brought along the right tool, etc), moderate (maybe an ally turns up at the right time) and major (you do the thing where Michael Myers looks like he's dead, but he actually gets up behind Laurie Strode/magically gets away before Loomis can look out of the window again). That should cover most things. Edge Actions are cool, but I think gating them behind Edge isn't the best, and most of those options should just be things you can attempt to do if you want to.