Very powerful analogy. However, with a small team I've always felt it makes sense to have one, maybe two "rafts" at a time. You need to execute something completely to know if its effective. I've found if you are splitting your time between too many rafts, they're all going to prove to be unfruitful, regardless of their potential
I think the key point that this article is missing, and maybe it's just assumed by the author, is that when you're sending out the life rafts in an uncharted direction you get enough user feedback to either work the feature into your product or not.
