In my Understanding Uncertainty presentation, I referenced the Huffduffer signup form as an example of refreshing the standard design of signup forms without changing the basic structure, and asked the audience for feedback. Not everyone liked it! Opinions were divided, both in terms of aesthetics and utility.
Well, if it makes any difference - there is now some evidence that this "mad libs" style of form design actually increases conversion rates. But before jumping to conclusions, there’s also evidence that it decreases conversion rates if it’s badly executed.
Yet another signal highlighting the importance of testing assumptions. Web design is not formulaic. The only “right” approach is tweaking a design based on a feedback loop of observation.