> "New UIs are worse because of screenshot driven design."
This. (I have high hopes for Apple, now that J. Ive is gone.) However, mockup screenshots had always been the medium of communicating designs to the client. But the more prominent role of screenshots may further the concentration on just a few paths of usage, as we may observe it in the common reduction of interface options – which may be fine for the few selected usage paths and may provide the base for a "cleaner" design (which should actually be a design job, not a content issue), but necessarily fails a number of use cases.
Another point may be made about an overly application of metrics. The issue here is that all metrics that may be actually useful are in the qualitative domain, which is rather labor intensive, complex and expensive. (Especially the process of transforming qualitative data into quantitative metrics.) So everyone is going for quantitive data, which is cheap and supposedly easy. That only few of those involved have a background in social research and would actually know what they are doing, doesn't especially help. So we may add data driven design to the list of possible issues.
(Much of this may be understood in the old dichotomy of "art or engineering". While the engineering approach is fine for managing exchangeable labor, it's failing in many other aspects. Meaning, what you get is an exchangeable product and chances are that, if you actually have a unique product, this approach will fail your needs – or, at least, the needs of your customers.)