IIRC, award seats were originally meant to be seats that would otherwise go unsold. This is fine when load factors are low, but as loads creep up, it becomes harder to predict on which flights seats will go unsold.

To improve award seat availability, DL has to:
a) lower load factors --> not good for business, or
b) give seats away at the expense of revenue seats --> not good for business, or
c) develop a better ability to predict on which city pairs loads will be low.

One thing DL could do to lower load factors is to discourage resource-consuming mileage runners (of the: JFK-SLC-LAX-SEA-ATL-MCO-JFK for $232 all in variety). Probably not popular here.

For item (c), it is much easier to predict availability of awards as one gets closer to the date, but then they could be diluting revenue from walk-up Y fares. For this case, I like the suggestion of putting a request in ahead of time, problem is that people need to be able to plan vacations. Some feedback of the "chances are good", or "fat chance bubba" variety would be useful to add.

The other thing DL could do is to NOT issue inventory at 331 days out, instead issue it only at say one to two months out. I would not like such a scheme, as it favors use of miles for more spontaneous trips, versus well-planned vacations.