Not the GP, but, in short, because space is very, very big, and very, very empty.
By way of extremely loose analogy, consider a large pea (1 cm in diameter) hanging by a string in the middle of a football stadium. Let's say that represents our solar system (minus the Oort cloud) and its environs. That's a slightly scaled-down version of Stross' analogy, where the diameter of the solar system (again, minus the Oort cloud) is about an Imperial foot. Now, without aiming, start firing individual molecules around the space. Those molecules are roughly to scale with a rogue planet.
On average, how long do you think you'd have to keep shooting molecules around before one of them passed within an inch of the pea? Within a centimeter? What about actually hitting it?
Even that's stretching the actual probabilities to the point of breaking, because your molecules are already in the neighborhood. For a more realistic example, take our 1cm pea and put it somewhere in some hypothetical space the size of the Milky Way, itself 6.4 * 10^9 AU in average diameter. At our scale of 1cm = 30 AU, the Milky Way would be a little over 200k KM across, or slightly more than half the distance between the earth and the moon. Now start shooting molecules around that space — from random locations, in random directions, again without aiming — and give me a call when one of them hits the pea...
Gathering all the interstellar hydrogen in a swath of space 1x1m wide and 1 light year long (i.e. everything you can collect with a 1-meter scoop from here to ¼ the way to Alpha Centauri) yields 0.01 grams of hydrogen.
A sobering thought regarding the inevitable "interstellar ramjet" suggestion.
