With some exceptions, literary agents who accept fiction queries by e-mail want just the query first. If they like the query, then they'll ask for a partial - 30 or 50 pages, say, sometimes with a brief synopsis included. If they like that, they'll ask for the whole book. If they like that, they'll offer representation.
Most submissions don't make it past the first stage, the query letter. Perhaps the query is badly written or the genre is one the agent doesn't handle or the description of the book doesn't grab the agent. If a partial is requested, the agent may lose interest after the first few sentences. That's not as harsh as it may seem. The agent knows that any editor to whom he submits a book is inundated with other submissions and must make quick decisions about marketability. Readers in bookstores make quick decisions about which new book to buy out of the gazillions on the shelves, each one screaming, "Snazzy cover! Blurbs! Catchy synopsis on back! Buy me, buy me, buy me, pleeeeeaaase!" If the first few sentences don't grab, then it's no sale.
So an agent reads a query, then maybe a few paragraphs or pages of the book, then maybe the whole book, and rejection can happen at any stage.
In the old days, when this entire process involved printed pages, agents understandably tried to limit the paper inundation by requesting one-page query letters and only rarely asking for partials (smallish packages) and very, very rarely whole manuscripts (large packages). But here in the 21st Century, most agents do all of this electronically. So why don't they ask for the whole shebang from the get-go?
In the electronic age, going through all the steps - query letter, partial, full - just wastes time. Why not ask for the whole ms. plus synopsis as part of the initial submission? Or, if the agent is wary of attachments, why not make that the second step - full ms. instead of partial? That way, if he likes the first few sentences, the agent can read the first few pages. If he likes that, he can keep reading. If he doesn't like it, he can stop at any point. It seems to me that this would take no more of the agent's time than the current system, but it would save lots of time for agent and potential client over all.
Storage space shouldn't be an issue. I assume (but perhaps this is a false assumption) that agents delete unsuccessful submissions anyway, whether query letters or full manuscripts. For those who hold onto submissions for reference purposes, external hard drives are cheap and immense nowadays.
Am I missing something? Am I looking at this like a techie geek instead of an agent? I wouldn't be at all surprised if there's some important consideration, obvious to an agent but invisible to me, that blows away my entire argument. If any agents read this blog (unlikely, I know), I'd love to have their input.
Probably I'll have to drift along forever in a state of bewilderment - regarding this question and so many others.