Space junk travels at supersonic speeds
...not to detract from your point, but... That's a meaningless descriptor, as sound doesn't really travel at any speed (and much, much slower in the bits that have fringes of rarified exosphere/whatever)
and does not do it justice. In LEO, where the Starlinks are, satellites travel at a velocity that, in atmosphere, would be roughly twice that considered
hypersonic in an atmospheric projectile.
And there are certainly problems from space-junk. Partly for this reason, the orbits chosen for Starlink constellations are quite low (less line-of-sight coverage, more additional satellites needed, but deorbitting times <5 years[1]) compared to those of Iridium (66ish active satellites providing near total coverage, but idealised decay time
could be[2] 10s of thousands of years, well beyond their useful life in other regards if not sent into decaying/outer-graveyard orbits while still capable of being so controlled).
Most interesting, as comparison, is the Iridium that was hit by the (defunkt) Kosmos in 2009 at hypervelocity closing speeds and made one of the few well-tracked debris fields (and I think the most significant of all the purely accidental ones). The combined mass of initial debris was ~1.5tonnes, well above the ISS. The scattering effect meant the ISS was prepared for debris in the vicinity (most notably in 2012), but such fragments of the debris was then more likely to deorbit itself naturally... I'm sure there's a good assessment of how much is still in a stable orbit ('high', circular, LEO), how much is just eccentric enough (initially up or down) to be troublesome but not yet retarded into re-entry at closest approach, what proportion has succumbed to the upper atmosphere after various passes.
Starlinks are 350kg (or less), each, not expected to last much more than those 5 years from the last self-boost[3] and are placed closer to the ISS (a future shell of "VLEO" Starlinks would be well
below it) but in an orbit where comparatively little consistent traffic criss-crosses the shell apart from the many other Starlinks, with current tracking stipulations meaning that long-range forecasts of possible collisions are likely to provoke whichever of the sats can still be adjusted (if not both) to be so done well ahead of time. By no means without danger[4], but much shorter-term, making the instantaneous hazards (spread/deprecated over time, divided by mass, not overly multiplied by the number of responsible parties, but ...yes... multiplied by relative constellation sizes) possibly somewhat lower, without necessarily being too generous in that assessment.
Yeah, sorry, I just wanted to clarify on "supersonic", and I seemed to get sucked into the details. Many
interesting details! Again. Maybe most of this best sits in the Space Thread, rather than the thread about deleting a thing that I never even use myself, but here it is now...
[1] Depending on space/Earth weather fluctuations, some have suggested increased interactions with the atmosphere
could deorbit more suddenly than expected for those not yet beyond their midlife when they start to encounter far less tenuous exosphere at their current part-decayed altitude, if their Krypton-based boosters/station-keepers fail. It's a bit of a crapshoot, which Musk clearly intends to ride by just firing more replacements back up to ~350 miles.
[2] Any number of failed-in-service Iridiums didn't take this long to deorbit, and others are 'obligingly' decaying quite nicely, thank you very much, so I'm quoting a clear top-end estimate here. Interestingly a "currently decaying" Iridium (or maybe a few of them, from the same batch) is helping with communications at the South Pole, last I heard. Being a polar-orbitting swarm, the coverage is limited but at least fairly usable, unlike the "mesh toroid" of the Starlink constellation, that isn't really capable of servicing the poles because of the local horizon and limits to range. But this is(/these are) in a
decaying orbit, presumably no station-keeping or de-orbitting fuel left, but has been usefully providing communications for quite a while longer than you'd expect a Starlink to stay up there in similarly adjusted circumstances.
[3] Or the first one. I understand they're sent to <350 miles at first, spat out of the "bus" to spread them out a bit and then tested. Only those that function well are then commanded up to the 350ish-mile slots that time and trajectory best suites them. Those that failed immediately after de-bussing (or even failed to de-bus) are going to decay much earlier, with either active or passive deorbitting.
[4] And I'm sure there's plenty of room to imagine Musk going all actively Bond Villain and actively threatening/creating a Kesler-shell upon one touch of a button he handily keeps in his pocket, for some suitably Hollywood reason that he has been leading up to all this time, if only we had known his true motives..!