That's a big complaint in modern CS research funding as well; traditional NSF and DARPA funding came in bigger chunks, with longer timeframes and less strong micromanagement of projects it was to be used for (there even used to be completely non-specific "center of excellence" type block grants, which gave a huge pile of money for a group of researchers to use on good research in a given topic of interest to the funding body). Now it typically comes in small chunks with very narrow parameters that have to be pre-approved and checked up on on a yearly basis, which not everyone thinks has improved the quality of CS research.