Asking them to read during the meeting usually results in a performance, they skim or get through as much as they can until people start saying they are done, and then they don't want to hold things up so they also announce they are done. This can happen with all slow readers, even those who would have high comprehension with multiple passes or at their own speed.
The other issue is the asymmetry of effort, which the author alludes to with "no doc => doesn't happen". Proposals require writing, but criticisms can be made verbally. You would expect a lot of the feedback to actually make the proposal worse because the proposal has been through a few write-edit cycles and the feedback is generated on-the-spot during the meeting within minutes of first reading the proposal.
If critics don't also have to write then feedback becomes a game of politics. If you ask people to give feedback on something without time to think, you will get noise. No one wants to appear bad at reading in front of their peers.