Different space missions have different orbit requirements, so there is a wide range of inclinations, altitudes, eccentricities, and ascending nodes. Take a look: http://stuffin.space/. It's a very dynamic system with a dizzying amount of pieces in it.
The issue isn't so much mega-constellations in and of themselves, since presumably with good operations automation and standardized communications between operators we can thrust and avoid potential collisions as they arise, even with very large constellations. The issue is reliability: applying a small burn to nudge out of the way isn't an issue when one or both of the two spacecraft in a potential collision is operating and has propellant (although accidents still happen [1]), but if they're both dead, that's a recipe for a hypervelocity trainwreck.
That's why it's important to ensure reliable fail-safe de-orbiting of end-of-life spacecraft. The FCC is moving in this direction, but unfortunately the rulemaking is still pending.
High circular orbits move a lot slower than lower circular orbits (eg: geostationary orbits basically don't move at all relative to the ground). It's the highly elliptical orbits that have massive velocities.
I'm not sure that being stationary relative to the ground is very relevant since we are talking about stuff colliding in space. The more stable orbits seem like they might be less prone to issues, though.
Humans are great at saying, "Look at all this amount of space. Let's dump junk in it!" until it becomes untenable.
A quick check of Wikipedia[1] told me that the orbit altitude of Starlink satelites is 550km. The earth has a radius of 6371 km. Based on some trig calculation (which can be done on this[2] little php web applet I found), the visible percentage of the Earth from 550km is 4%! That's a lot. Of course, that's a significant overestimate, since it counts surface area which is just barely technically visible at a nearly 180 degree angle, and assumes a spherical Earth. Let's be generous, and say that only 5% of that area is both reasonably visible from StarLink (I apologize for this crudeness, I'm too lazy to do the math) and above trees/buildings/mountains. So that's .2% of the Earth visible from each satelite at an azimuth which is possibly visible for an average human observer on the ground. That gives an average number of 84 satelites observable, assuming a normal distribution, and likely many more for an average observer in an above-average-density area. Obviously these aren't going to be as visible and bright as the ISS or Iridium flares, but they aren't gonna be sparse.
[1]https://en.wikipedia.org/wiki/SpaceX_Starlink
[2]http://www.neoprogrammics.com/spheres/visible_fraction_of_su...
Even better, SpaceX's new Starship super rocket will make it vastly cheaper to put mass into orbit. That means it will become possible to put up a bunch of satellites that will collect dead satellites and other space debris and return them to the ground.
https://orbitaldebris.jsc.nasa.gov/measurements/radar.html
> NASA's main source of data for debris in the size range of approximately 5 mm to 30 cm is the Haystack Ultrawideband Satellite Imaging Radar (HUSIR). The HUSIR radar, operated by the Massachusetts Institute of Technology’s Lincoln Laboratory, has been collecting orbital debris data for the ODPO since 1990 under an agreement with the U.S. Department of Defense. HUSIR statistically samples the debris population by "staring" at selected pointing angles and detecting debris that fly through its field-of-view.
> The data are used to characterize the debris population by size, altitude, and inclination. From these measurements, scientists have concluded that there are approximately 500,000 debris fragments in orbit with sizes down to one centimeter.
https://orbitaldebris.jsc.nasa.gov/faq/#
> 3. How much orbital debris is currently in Earth orbit?
> More than 23,000 orbital debris larger than 10 cm are known to exist. The estimated population of particles between 1 and 10 cm in diameter is approximately 500,000. The number of particles larger than 1 mm exceeds 100 million.
Orbital debris is already causing damage. Here's some examples from Hubble: https://orbitaldebris.jsc.nasa.gov/measurements/in-situ.html
The photo gallery shows other examples of impact damage as well as examples of debris: https://orbitaldebris.jsc.nasa.gov/photo-gallery/
Given that the GEO ring space is indeed finite & should be kept free of debris (even crossing one) at all costs I can imagine eventually reverting to the idea of large communication stations instead of individual satellites.
The stations will likely still not be manned will support automated & possibly even manned servising. It could look likey large truss, kinda similar to the one used on the ISS. Inside it would host power & fuel interconnects. Attached to it would be the communication payloads, solar power modules, engines and fuel tanks. Periodically a resuply/repair craft would visit, replenishing fuel & swapping any failed modules. Depending on your launch costs broken mpdules would either be tossed to a safe graveyard orbit or brought back (possibly to a LEO space station) for refurbishment.
The first steps of this might be already happening with the recently launched Mission Extension Vehicle:
https://en.m.wikipedia.org/wiki/Mission_Extension_Vehicle
The MEV will grapple a satellite that is low on fuel and will act as it's replacement propulsion unit. In this way it can serve at multiple satellites and ppssibly even serve as a tug for satellites that have failed & are are likely to fail soon, to make sure they reach the designated graveyard orbit.
And other similar & more ambitious missions are already planned and being built.
The sky is huge. And though it's probable that one thing will hit another up there, it doesn't seem very probable, even if something were to explode into a billion tiny bullet -like projectiles.