I think the shit hits the fan especially when you are evolving a platform continuously, adding features and improving on design. Of course sometime it is a need to change your 90s interface to something more current but sometime you have managers who make it as their goal, each year to come with something completely new so that their managers can get from their managers an appraisal. You know the drill. The higher you rank, you need to pretend you are doing something.
In this special example, the problem was that every year you had some overgrown weasel with god-like self impression and middle management chair aspirations, who wanted to push something, something new. It did not matter if it made sense or if it was breaking the rest of the concept, or even if it was fitting the concept, not to mention how it will play along with the other features. Can you get the picture now? Can you spot the mistakes?
That being said, the afore mentioned weasel was pushing the ideas forward, with the only notable reason: "It looks modern and I like the looks". Well, fuck you too! Good arguments, best ones after Moses has split the sea in two and Jesus walked on water. Why? Because they could!!! When you have a complex interface with lots of possible interactions and elements you can't simply afford changes because is cool. Let's see some issues.
You ditch a certain format where then you need a huge menu of some sort, preferably side menu because is too big, but you hate the idea of side menu, is so 90s. Well, you hate the idea of an L shaped navigation too, somehow you put everything in a side menu, because was the last option. Your next step is to create "virtual" desktops because they are so cool on Mac .... later you try to transform all the menu items into icons and put an icon bar at the bottom of the screen. Does someone still follow me? Can you see already what it is? Well, for the more visually impaired ones .... it is becoming slowly a MacOS interface .... for a web application. Are you serious? The only answer is "It looks nice". Go buck a fuffalo you .... very intelligent person.
And no, this is no joke. Imagine also that every year you change things and your users need to constantly learn how to use the same application. Of course if you are an Apple fanboy you are used to change stuff with each update, to discover that some apps will not work and so on. And you love that because you love Apple. But ... you do not work for Apple and your users are not as brainwashed as you are.
So one of the biggest dangers is that you can end up messing your UI every time something becomes trendy, and the worst part is that your users will be confused too.
Another unfortunate trend is to jump on new technologies because "if we miss the train, we would never be on the market". First of all it is a completely wrong mindset, as long as you are not a company who produces always bleeding edge whatevers. But let's see, shoes are "designed" since centuries, would you not be starting a shoe manufacture if you would want to? Would all the clients been taken by the competition? Coming late but bringing a more mature and refined design will not have benefits? Maybe the metaphor is hard to be understood but if you ditch the old and jump on the new framework, which is not mature enough, has no proven reliability and some things you need to invent for yourself because are not yet available, what do you expect? Usually doing this breaks your app in countless ways and some more ways you cannot even foresee. I wouldn't mind to do this on small experimental projects, where the users know that is experimental and bleeding edge. Take the beta testers for example, they assume the risks for having that new feature which is not implemented for the rest of the mortals. They do this just because they want to separate themselves from the foot-folk, but are you willing to transform all your users in beta testers?
As an ending, please use trends wisely and think also from the user perspective, the real user not you imagined user, that persona who loves technology, because you will have a large majority of technical challenged users too, and they should be also your target.