What this discussion is basically saying is "business vs safety". Since the launch is unmanned, there's really no safety issue either, just the potential loss of the payload. SpaceX doesn't insure launches, so they don't stand to loose if the launch goes bad. The way I see it, the more they can test per launch, the better. I'm sure in a few years when crewed launches begin, Elon will be far more conservative considering most launches will be crewed, and SpaceX will have less opportunity to push the envelope and so the big risk taking should be done now. With that said, I think there's not much chance of anything going south with this landing attempt. I think the same software that has been working great so far will be the same until post staging when the newer code will take over, by which time the upper stage will have ignited and left the first stage behind.