Or, alternatively, maybe their vision was "let's see what mechanics people like the best." Like I said, it wasn't a beta play-test, it was a design play-test. In fact, they made quite clear with the first play-test that it was still in the design phase. You propose they only let you see the rules after they've decided, for themselves, what mechanics ought to be used. They took that approach with 4E and decided to take a different road this time.
The problem is that they didn't ask what happened in people's games. They could have, it would have been trivial to include a few check boxes in modules for the GM to tick for which characters made saving throws at Point X(Y,Z), what they did in Room F (negotiate, sneak, trick, fight), etc and then get GMs to report that along with giving the subjective feedback. If you don't know what the "typical" result is, how do you judge whether people who like/dislike the game do so because they had typical or atypical results? It may be that you'll get results from that which tell you what preconeived notions people have about what should happen, and you can then if you want design to that, but it's probably not a good idea to design based on highly vocal people who saw an unusual series of results.
Alternatively, you can decide that the "feel" of D&D is solely in the fluff used rather than on what happens in the game, and design based on that. You can do all the work that will determine what happens when that's finished, tweaking maths is easy, and if people then turn round and complain you can point out that they liked the fluff and were told that the maths still needed tweaking, and it's hardly WotC's fault that those tweaks made a big difference to what happened - they were after all told by the playtesters that the "feel" was right.