Pretty sure all these companies are heavily A/B testing every change they make, and would immediately abandon any new design if it decreased engagement or retention even slightly.
These changes are "hated" by a vocal minority who loves to complain about change to show how much the liked the old version. They don't quit using the product, though. And if you ask their opinion again even three months later (which is another thing big companies tend to do), they'll have completely forgotten what the difference was between the two.
Of course, some people do quit a service after any given change, no matter how innocuous—they just disappear without saying anything. But changes, even change for change's sake, also tends to increase signups. So in practice, it's just a question of whether the positives outweigh the negatives of the change. And only real usage data can answer that, not customer surveys—and especially not customer exit surveys.
You could avoid multiple input and multiple links with cookies. That wouldn't require a login.
Users might not even know what a cookie is, so they don't ask for one and instead they say the simplest thing that comes to mind: an account. They don't know what the options are in technology. They barely know what they need, and they most definitely don't know what they want.