I have no idea how many posts I've dedicated to NSF BIO's preproposals and I'm too lazy to check, TBH. However, they have significant impact on my community and many of the scientists I know and work with. They are the Gate Keepers to getting a shot at that golden ring. Successfully navigating this stage of the process is critical to having the chance to even apply for money.
As a reviewer I generally like preproposals. I feel like I can get a decent sense of what the PI(s) are trying to get at and I can decide whether I believe they can make it happen, for the most part. I like reviewing the compact format - I often find I am interrupted too much when reviewing full proposals if I don't retreat to solitude. I can get through preproposals in between obligations. However, you fill in a lot of gaps as a reviewer in the preproposal format. It's also a different type of writing with a different target audience and I think it lends itself a bit more to over-promising with the intention of generating that last bit of data before you really have to prove it in the full proposal. The fact that the membership on the pre- vs full panel does not completely overlap has always left me uncomfortable for exactly this reason.
As an applicant to NSF BIO I'm concerned that preproposals have significantly increased the number of applications both DEB and IOS, mostly from new applicants. There's two ways to look at this: 1) By lowering the activation energy, NSF is now able to capture a broader diversity of science that was not previously being proposed. 2) Alternatively, there's more noise in the system as people throw applications in so that they can report them in their annual review. Unfortunately, this is A Thing in some places and there are universities that even pay PIs per submitted application. Reality is likely somewhere in between, but the competition at the preproposal stage is fierce and not getting better.
As a preproposal writer, the format creates problems. Foremost, the limit of 2 proposals per division means that I often have to choose between a promising collaboration and a core project my lab is invested in. Considering the short turn around of NSF grants, it seems like I've always got a renewal and a new core project going in. One strategy is to, say, re-pitch a DEB proposal for IOS, but that never works out as well as it seems it might. Whereas the short format makes it easy to propose new ideas, the limits on numbers kills a lot of potential collaborations. Obviously there has to be some way to control the number of applications that are now easier to package up and send off, but damn, I wish I could fit in some of these cool side projects with collaborators.
Unfortunately, none of this is getting better without more money in the system. However, I would honestly prefer a compromise where we funded shorter proposals (say, 8 pages), did away with preproposals and limited people to 3 total aps during a year with 2 cycles. I fully recognize that this solution works best in my world, whereas someone at a PUI might think this is a terrible change. I don't know, but I can't help but think there's a better solution that splits the difference between the old and new system. And yes, I realize I just recently wrote about the futility of making tweaks to the proposal system.