For about a decade now I've been building robotics software UIs in a context where we don't have a design team (internal software, tech savvy users, start-up pa

For about a decade now I've been building robotics software UIs in a context where we don't have a design team (internal software, tech savvy users, start-up pace). One of the things I've found to work so consistently well is the kind of test shared in this article. I regularly say, "to learn what we don't yet know about the problem space" when asked about why I like shipping things ASAP and then worrying about design characteristics later. I'm convinced that design regularly impacts a feature's usefulness, but is rarely necessary to make it useful at all. That is: it's important but not required. Of course this is for a captive audience. For something you're trying to sell, window dressing can matter.