Optical connections, TOSLINK in particular, are subject to a couple or three things ...
The first is attenuation of the signal over distance (true of electrical cables too). Plastic cables typically have higher attenuation rates than high-quality glass and tend to become problematic at shorter distances. It's quite possible to get to a point where, and I'm grossly simplifying here, "1" and "0" become unreliably distinguishable (specifically, some of the "1" pulses may simply not have the amplitude to register in the receiver). There's not really any subjectivity to this effect - and it's why optical fibers carry attenuation ratings.
How that actually manifests itself, from an audio perspective, is dependent on the receiver implementation, as well as how severe the effect is. It may result in drop-outs, it may result in audible corruption, it may result in static or it may get smoothed over, or it may just result in the receiver refusing to sync and play (i.e. the cable just won't work).
It's worth noting that as the transmission frequency increases (or as the sample rate goes up) this will tend to have a greater impact.
The degree to which this is an issue at typical cable lengths in an audio system is another matter, of course.
The second is related to how an optical signal actual travels along a fiber. I'll skip the gummy physics and math, but essentially the light wave bounces along the fiber under-going total internal reflection (ideally) at the boundaries of the medium. As a result, and again I'm simplifying heavily, if the optical fiber is bent, then wave can be partially reflected at different points along the medium boundary and this results in a time-based distribution of the signals arrival at the other end of the cable. At high enough clock rates, this can smear the signal edges and make it harder, or impossible, to properly recover bit-states. It also results in signal jitter, since the signal edges are, essentially, how the clock is determined (and it's why TOSLINK uses a bi-phase clock).
Using lots of very small diameter fibers helps address issues from bending the cable. Glass generally has a higher refractive index than plastic and as a result total internal reflection is more reliable. So a better cable will general be multi-strand and glass.
In general, the fewer bends you have, and the larger their radius, the better off you are.
Finally, the polish on the ends of the cable makes a VERY big difference to how well they work - and if the ends get damaged (which can be as simple as touching them) it'll often severely impact the speeds the cable will work at, or even if they'll work at all.