I agree with you that grants should still be firmly merit-based, and you make a good point that even senior researchers have to apply for multiple grants just to get one funded. The anecdotal number I've heard from various scientists at different points in their careers is 4-5 applications for every one success.

So that obviously makes it tough for young labs, especially if they are largely experimental - if you can't pay for equipment and grad students, very little gets done in the lab. A new investigator has to do all the experiments alone and write for multiple grants on top of that - not a good situation. So maybe small, short-term grants are the way to go, with enough money (together with a startup package) to get the lab to a point where a big 5-year grant becomes feasible.