In networks there'll always be failures.

So I think it's still necessary to deal with the problem of your distributed database becoming partly disconnected and partition.

And as long as there is consistency (though eventual) in case of such severe problem I personally could live with it. Better than staying available by all means but having to fix up inconsistencies and vex clients.

Good arguments, I like in particular the insight that different business circumstances call for different database designs, and different compromises between consistency and availability. Google is of course a different player than most others on this planet.