Absolutely key ==> "Their “attention span budget” can thus be spent considering how your new thing can fit into their lives, rather than trying to recall ..."

Absolutely key ==> "Their “attention span budget” can thus be spent considering how your new thing can fit into their lives, rather than trying to recall ..." UI is all about minimizing user workload, and NOT requiring people to learn redundant knowledge. Take advantage of every bit of knowledge that they already have from previous systems. Every split second they must use to figure out your new way to do someting they already knonw how to do another way, is squandered and redundant. Eliminate it. Work hard to eliminate it, so that they can focus on your real innovation Note: even if you have actually figured out some truly better UI controls, they must be not merely noticeably better, but better by a margin that exceeds the cognitive costs of both the new learning & the ongoing disconnect from the main UI (and remember, they don't live in your app the way you do). E.g., the fad towards 'clean interfaces' with minimalistic or totally hidden hotspots may look prettier, but is a huge waste of user time and generates animosity towards the product. Just the example of Gmail hiding the Trash folder; I don't know how many times I've had to help people telling them that they need to scroll down, click on the "MoreV" word (not even transiently highlighted as a control), then scroll down again... I use it so little I've even had to Google it myself in a rush... That and 1000 other things are not an improvement, they are a waste of people's time for some design conceit. Of course, if you want to give off signals that your core product ideas aren't all that uniquely useful, and you need to create cute & costly UI differences as distractions, by all means, go right on ahead... If you have a real innovation, the best thing you can do is put zero redundant learning between your user and your key innovations.