Earlier discussion (earlier Wiki too): <a href="https://news.ycombinator.com/item?id=22794771" rel="nofollow">https://news.ycombinator.com/item?id=22794771</a><p>Repeats are fine after a year or so but we need more astonishment before then (<a href="https://news.ycombinator.com/newsfaq.html" rel="nofollow">https://news.ycombinator.com/newsfaq.html</a>).
This is an interesting principle which I understand the power of but feel is sometimes detrimental because it treats all surprises are equally bad.<p>Surprises can also be delightful. (It might be arguable that delight even <i>requires</i> surprise.)<p>That is perhaps the most extreme counterexample, but the point here is that we should treat very differently things that are "surprising, but easily recoverable" from things that are "surprising, and disorienting", "surprising, and makes you want to flip tables", or "surprising, and life-threatening".<p>Taken to an extreme, this principle leads you to stagnant designs, because you're afraid of modest innovations giving users a mild surprise. Or, your product fundamentally doesn't quite work the way most users expect it to work, so in order to avoid giving users a surprise, you do cosmetic things to paper over it...<p>Tldr: use at your own risk and remember you're at the wheel, not your principles