It's a difficult issue. Feedback and re-implemenation cycles require time, money, and a pretty high level of professionalism (and thick skin) on the part of both the developers AND the playtesters.
In the early stages of development, a game is often a terrible mess, with myriads of placeholder elements, and a lot of features missing or half-working. Many important aspects (game flow, look & feel, consistency, continuity, dramatic power, dialogue flow, difficulty) are quite difficult to test until all the necessary elements come together in a near-final version. Before that stage, testers AND developers will need to fill many voids with their imagination. (And interestingly, when imagining something, it often looks and works much better than the real thing
).
Once a game HAS reached beta stage, and all elements are there, the picture becomes clearer. But at the same time, the game has become by several orders of magnitude more complex and vulnerable to changes. The assets have been produced, the money has been spent, the time has been used up. The team is probably exhausted, and with the release deadline approaching, they may now also be, well, unreceptive to fundamental redesign requests.
This is why beta testing often focuses on eliminating bugs, tweaking user friendliness, and adjusting difficulty. Feedback and change requests regarding fundamental design decisions need to be brought in long before Beta. It is simply not a good idea to introduce brand new engine features, interface elements, characters and subplots while in Beta. This will VERY likely result in A LOT of unforeseen trouble (and work), and ultimately push back the release.
