One minute blocks that are of size ~1 KB (most of the blocks right now) double the likelihood of a miner getting a block. Selfish mining only becomes a severe issue when blocksize becomes large.
You're the one who is concerned about the reward being too small in 10-20 years??!!
Obviously the goal is to create a design that is sound for the long term, not focus on a couple of weeks of easy mining when there are hardly any transactions on the network. As was pointed out before, solo mining for small (one computer) miners will sooner or later (most likely sooner) become non-viable whether blocks are one minute or two minute blocks. Maybe not in the first week, but if the coin thrives, it won't be long at all. This was a very short-sighted reason to speed up the blocks.
I'm not referring to selfish mining in any case, just regular (random) orphans. If it takes 3 seconds to propagate a block across the network, then with 60 second blocks you will have about 5% random orphans (where two different people solve a block at the same time; only one will survive). This gets much worse when the blocks get bigger (take longer to verify and forward at each hop). If it takes 10 seconds to propagate you are looking at 16% orphans.
Pools reduce orphans by concentrating hash rate and directly communicating with the hashers in a star pattern instead of a p2p. You won't be solving a block at the same time as one of your pool-mates very often (due to longpoll/stratum) and the pool will solve on top of its own blocks more quickly than any foreign blocks, so it will win more races. These effects combine to give pools a huge economy of scale.
Even with those 1k blocks you describe, there are already plenty of orphans, just look at the daemon output. Bytecoin with 2 minute blocks still has a lot of orphans too.
Satoshi was no idiot when he picked 10 minutes, and didn't just pull that number out of his ass. That may be slight overkill but 1 minute is going way too far in the other direction.