Membrane pores aren't irreducibly complex.

Check out this somewhat odd-seeming PNAS paper (don't worry, it's open access) about diffusion across pores. Using two macroscopic models, one real and one virtual, they showed that a concentration gradient can be maintained across a membrane with leaky pores in it without the need for gating, antiporting, charges, or anything. All that's needed is a physically asymmetric pore - wider at one end.

When I first read the paper, I thought it was so obvious that I wondered why the research even had to be done in the first place. But then, I realized that it's not so intuitive; it seems to go against a lot of what we're taught about diffusion equilibria. Simply changing the geometry of the pore shifts the equilibrium to one side, without even the need for bigger particles to block one end of the pore. The reason it's relevant, the researchers suggest, is that it shows the potential for an extremely rudimentary metabolism at the very beginnings of life. It's possible to maintain a particle gradient of, say, sugars and ions, across a membrane without any of the fancy multimeric gated two-way channels that we advanced eukaryotes sport. An early ion pore could easily have evolved from a protein that already bound that ion and underwent a mutation that bound it to the membrane.