If you think that the criteria that Arrow's theorem gives are important, then I think it's important that we realize that these criteria cannot be met. It is good to know what it not possible so as to not waste time trying to achieve the impossible.
However, as I pointed out on Wiki:ArrowsTheorem, I believe that a voting system can be "good" even if it doesn't meet the Arrow criteria. Specifically, I think it is okay to violate the Arrow criteria when a vote is "close"; a close vote to me means the "will of the people" are almost indifferent between the given options, so not much would be lost even if the result was chosen randomly in that case (except the illogical but sometimes necessary appearance of absolute legitimacy). – BayleShanks
See also VotingIsEvil
I don't know ArrowsTheorem. I haven't studied it.
And again, it seems a little too "absolutist" to me- "either you have this property, or you don't." I've seen cases where a little "take" in one place gives overwealmingly "give" in another.
It's also clear to me that some voting systems are better than others.
Here's a page justifying Concordent, it seems good enough to me.