I absolutely agree - that's the whole point of the article and my comment. If you're in an experimental space, and 9/10 of your ideas are never going to make it to market, I'd rather spend 1 units of time on each to figure out which one's going to work, and then 20 units of time redesigning it the right way (i.e. 21 units of time to be done), rather than 10 units of time on each to make them all ready for production (i.e. 100 units of time to be done). But those ratios will vary depending on how sure you are about what needs to be done next, so yes - you've gotta weigh the cost / benefit.