It suffered the only LOCV event in history that was the result of a launch vehicle failure, because, unlike any other launch vehicle that has ever flown manned, it had no provision, and, because of its design, could have no provision, for crew escape during the relevant phase of flight.
It also suffered the only LOCV event in history that was the result of debris from the launch vehicle impinging on the spacecraft, once again because of a launch stack arrangement unique among manned vehicles (Buran doesn't count as it was canceled before a manned launch was attempted), and something that could not have been corrected without turning it into another vehicle entirely.
That's what I mean by "inherently" dangerous, BTW. The design flaws that caused the two LOCV events were not correctable by any means other than building a different spacecraft, whereas the failures that caused the Apollo and Soviet LOCV events could be corrected with incremental changes to the spacecraft.
And? Would it be safer if it would just suffer from the same failure modes that we already know? And did you even bother reading about the "abort envelope" of a capsule before ranting about the Shuttle? A LAS effectively means, that you are now sandwiched between explosives, which will kill you more effective than the rocket behind, if you trigger an abort in 70-80% of the first stage flight, because of aerodynamic and structural constraints. Yes, a LAS gives you more abort options than no LAS. For a capsule, it is also the only way to abort because it can't fly away from the rocket by other means. The Shuttle can pull a Nz maneuver from SRB separation on, if you need a contingency abort. If you are not scared of damaging the heatshield more, you can even do a risky abort from 92 seconds in flight on (which was really an option if the orbiter will crash into the ocean anyway). The more you try to land the Shuttle intact, the less abort options you get and the smaller the abort windows get. But the window for ensuring crew survival is not much smaller than it was for a Saturn V launch. The pad abort capability was the only thing that was really missing, but most astronauts agree that in case of a Saturn V explosion, it would have been barely adequate. The Shuttle on the other hand was much safer in that category because it can't explode like the Saturn V. It can only catch fire because of the low density of the hydrogen (Think of the Hindenburg there).
Yes, the Shuttle has a larger heat shield which is more exposed to debris from the ET and the SRB nose cones. By design. It needs this heat shield for its mission, which includes being able to travel 1500 km cross-range, since it was suppose to land on one of two primary runways for faster and cheaper recovery (which it achieved. Even a transfer flight with a 747 is cheaper than renting a full aircraft carrier battle group for multiple weeks). But the real big issue is, that the known shedding of debris by the ET was simply accepted and assumed that it will never hit one of the RCC panels (debris strikes on the black tiles are actually no big problem, it takes massive damage their to compromise the heat shield). The heat shield was never the problem, actually 99% of it was way more robust than the heat shields of previous spacecraft (Apollo 13 for example would have failed fatal, if the heat shield was really cracked by damage at the outer edges - luckily the damage on it was maximal superficial).
The debris by the ET was the problem, since the ET dropped way more debris than previous launchers, and it did so because management wanted to keep costs on the only fully expendable part of the shuttle as low as possible. It was a negligence by management for years, that neither the heat shield was improved to be more robust at the fragile RCC, nor the ET improved before STS-107 (which had massive effects on the debris events then), nor that a EVA repair option was possible. Even simply creating a dual layer RCC panel would have been an option (which would still have been lighter than the OBSS in its sum) when the first debris events had been noticed - instead NASA management decided to resort to "don't ask, don't tell" - and after STS-107, pressure to finish the ISS was too high and the end of the Shuttle decided, so a long-term solution to fix this problem ultimately was no option any more.
The big issue of STS-51-L was not the lack of abort options. No abort option would have helped there. Similar accidents cause LOCV even in capsule+LAS simulations because you can't trigger the abort as fast as the rocket disintegrates around you. Don't fall prone to the fallacy "If the capsule sits on top of the launcher, it is safe". On top of the launcher is not the safest place, though it often buys you extra time that you can badly need. When rocket tanks disintegrate at the bottom dome, you can't fire the LAS as fast as the capsule is pushed into the rocket debris by drag while the disintegrating tank accelerates by Newtons third forward. Yes, a LAS is there better than just letting the accident go on. But a side mounted spaceplane has actually more chances to evade such an accident, if you get the abort signal at the same time and with enough time to react at all. Not really given on any SRB (they fail rarely, but when they fail, then it is fast and violent).
STS-51-L was doomed already on launch. Even a capsule would have failed to protect the crew on such a situation, any crew survival of a capsule in such an accident would have been plain luck. A proper emergency detection system in the Shuttle could have triggered an abort 3 seconds before, when the chamber pressures in both SRBs diverged massively. The problem is just: Neither a LAS, nor the Shuttle orbiter had been in their window to allow an abort. The LAS would have had too high dynamic pressure to fire, same for separating the orbiter in emergency - the LAS would have been available again about 85 seconds after launch, the shuttle orbiter could have separated earliest after 92 seconds.
And sorry, but "incremental changes" is something that does not even work out easy in software and is often impossible in real hardware. In the fairy tale world, engineers will sit together and improve the hardware with magic. In the real world, engineers will sit together and often simply write into the manual "Don't try this again.".
As you can again see: Just being the first to fail in a specific mode, does not mean that others would not have failed so. Every design decision has its price in different chances of failures to happen and the danger of these failures.
