I agree. That's my experience in this industry, and in big projects. Designers are hired to make things look pretty.
> In the same way that a full-stack Linux expert isn't going to understand how someone without 5+ years of command line experience approaches a computer, a trained designer with an experienced eye is always going to struggle to point themselves in the place of a naive user.
... but with this I can't agree. It's not a reasonable comparison. The full-stack Linux expert won't understand the user mostly because it's not his job to do so. He doesn't have the knowledge and methodology to get to know and understand the user (as well as his context). The UX expert, on the other hand, should be trained specifically to do this in the most systematic and unbiased way possible. He doesn't need exactly to role-play the user (frankly, I don't like this conception of UX, but that's a pet peeve), but he should know and understand him. It's not "I think the user will like this", it's "I know users want this".
> Good UI needs to be tested for design assumptions and for user taskflow. The testing needs to be repeated every time changes are made.
And in this we fully agree again. All the knowledge the UX expert can harness about the user and his context should be validated, as early as possible.
In the same way that a full-stack Linux expert isn't going to understand how someone without 5+ years of command line experience approaches a computer, a trained designer with an experienced eye is always going to struggle to point themselves in the place of a naive user.
Good UI needs to be tested for design assumptions and for user taskflow. The testing needs to be repeated every time changes are made. For any business with a web site that sells to customers, this is at least as important as unit tests in code.
IME designers aren't often interested in this. Some are, but many simply don't get it. And even if they are, there's pressure to see UX as a machine for driving conversions (hence the same Bootstrap template with the same social proof and the same calls to action everywhere), and not so much as a way to make the customer experience more positive.
The proof is that many sites are broken in some fundamental way. I was on a logistics site yesterday trying to organise a redelivery, and the site simply didn't do what I needed. The graphics were pretty, but it obviously hadn't been tested properly, and many elements were broken.
Yesterday's example was unusually bad, but similar issues aren't unusual.