> Using optical rather than RF doesn't really change the game that much.

The precision required for aiming is directly related to the wavelength. Iridium NEXT satellites use Ka band with a wavelength around ten millimeters where anything light related has a wavelength measured in hundreds of nanometers.

The forward/backward links are a lot easier than the inter-plane links, but it's still not trivial because you're trying to hit an object the size of a small car with a laser from over 1000 miles away. Not impossible by any means, but there's not a lot of margin for error when they're looking to be able to transfer around 100 gigabits per second over this link. Other FSO systems work at significantly lower bandwidth and/or shorter range. That's not even getting in to the inter-plane links, where the target is constantly moving even in a relative sense.

Doesn't RF just get shot in all directions and get picked up by the satellites easily, whereas the lasers need to aim precisely when transmitted?

You'd need an excessively powerful transmitter to use an omnidirectional antenna. In the context of a satellite, where power efficiency is crucial, it makes much more sense to use a lower-power transmitter and a directional antenna / beamforming.

Hmmm... but less precision is still required than optical lasers, right?

A bit? It doesn't really change the nature of the problem, just the tolerances. It's nowhere near the intractable problem that some people make it out to be.

I mean, we are all aware each extra 9 of precision/uptime/etc. is far more expensive than the last. If the lasers require an order of magnitude more precision (and I could imagine it being higher), it would be a far far harder problem.