Repeated patterns in digital signals can cause errors in several ways such as DC bias (likely the case here), or a buildup of energy on the edges of the signal's fourier spectrum which then gets filtered out and shows up as signal degradation on the oscilloscope.
Nowadays the lower layer transmission protocols all re-code the signal to ensure frequent edge transitions, and after a few layers of that the odds of these patterns causing problems goes way down.
And then compression and encryption (hopefully in that order!) make it go away entirely.
Buuut, 25 years ago network equipment wasn't as layered and sophisticated as it is today, so that sort of thing would crop up now and then.
Nowadays the lower layer transmission protocols all re-code the signal to ensure frequent edge transitions, and after a few layers of that the odds of these patterns causing problems goes way down.
And then compression and encryption (hopefully in that order!) make it go away entirely.
Buuut, 25 years ago network equipment wasn't as layered and sophisticated as it is today, so that sort of thing would crop up now and then.