While it is true that crafting a good FOIA request is a subtle art on a good day, and one that requires subtlety, cunning, and overt legal threats on a less good day; does that not strike you as the real problem?
The fact that targeting an FOIA is quite tricky isn't some fact of nature, like burning yourself if you touch hot objects, it's a deliberate product of the incentives and processes that govern FOIA systems. Shockingly enough, those tend not to veer in the direction of 'pleasant, helpful, and obliging'. If you are lucky, it's more like holding a senate hearing with a spook: If you know exactly what question to ask, and how to ask it, you may receive an answer that is strictly true within the context of the question you asked; but if you don't know, good luck. This doesn't exactly help the existing trend of actual interaction with your government being something that is either a full time, slightly obsessive, hobby or something you need a legal team on retainer for.
Aside from that, there's also the consideration that while some 'sunlight' type efforts are quite focused on a specific person, case, interaction, etc. others are essentially 'epidemiological' in approach. Unless there is a single "This Is The Smoking Gun, The One Dirty Deal Where All The Big Players Showed Their Hands", which there typically won't be, your task is much more one of population level survey, trend-spotting, analysis of how influence is transmitted through an organization or organizations, and so on.
This sort of study effectively demands the ability to obtain public records at considerable scale without massive transaction costs. Depending on the (almost certainly horrible) realities of legacy document systems, this may simply not be possible in the immediate term(even with an ample budget and actual goodwill, serious document management and conversion programs are tactical nuclear nightmare fuel, perhaps second only to ERP rollouts); but it must be the goal if 'freedom of information' is to retain and expand its utility.