One of the major errors in the Space Transportation System (STS) design was the lack of a crew escape system usable from launch to landing.
Escape systems are costly, take up mass allowance that can be used for payload, and can cause problems on their own. The STS designers had an idée fixe that STS would be an "airliner to space" and would operate with airliner levels of safety. They concluded no escape system would be required due to what they considered a completely safe design.
Note that SpaceX's proposed Starship has no escape system, using similar rationale.
After the failure of STS-51L which destroyed the Challenger Orbiter and killed its crew, the rationale for not including such an escape system was heavily scrutinized by the press, public, and government. Studies were done regarding retrofitting escape systems into the existing Orbiter. Much of this work is mentioned in the NASA report IMPLEMENTATION OF THE PRESIDENTIAL COMMISSION RECOMMENDATIONS which summarized the work done after the STS-51L failure to return the STS to flight operations.
Note that the Rogers Commission recommendation re: escape was only...
Make all efforts to provide a crew escape system for use during controlled gliding flight.
...which was a 'softball'. However, a recommendation to provide a launch to landing escape system would have shut down the program, which at the time was supposed to be the sole US launcher of payloads...
The study, completed in September 1986, considered ejection seats, tractor rocket extraction of seated crew members, bottom bail-out, and tractor rocket extraction through the side hatch. Each option considered the crew size, the required orbiter modifications, and the implementation schedule. These options are summarized in the following paragraphs.
An ejection seat concept that would extract up to five astronauts was assessed. During operation, this concept would jettison the tops of both the crew module and the forward fuselage before propelling the crew out of the opening in individual ejection seats. The addition of ejection seats would require major structural modification of the overhead consoles, flight deck floor, crew module structure, and forward fuselage structure.
A new ejection seat design would be required because the ejection seats used during the orbital flight test program are very large, and installation of five seats would affect orbiter aft flight deck payload station usage. The estimated first availability of the ejection seat concept is mid-1990. This concept is not being pursued because of late availability, extensive vehicle modifications, and crew size limitations.
Another extraction concept investigated was a tractor rocket system that would extract up to six seated crew members. Once activated, this system would jettison the tops of the crew module and forward fuselage and extract the crew using tractor rockets. This concept would require modification of the crew module and forward fuselage structure, the flight deck floor, and overhead consoles, and would affect payload station usage. The earliest availability of this modification is mid-1990. This configuration is no longer being pursued because of late availability, vehicle modification requirements, and crew size limitations.
A bottom bail-out concept that would provide safe egress for up to eight crew members was also assessed. In this concept, a panel would be opened on the bottom of the orbiter to deploy a guide chute, permitting the astronauts to exit the orbiter through the chute. This concept would require extensive structural modifications, including installation of a deployable panel and pyrotechnic devices to open the panel, design and installation of the chute, and relocation of some subsystem components. Modifications could potentially be completed and certified by 1989, but the concept is not being considered because of the highly complex vehicle changes required.
The final concept evaluated was escape through the side hatch using tractor rockets to propel the astronauts out of the orbiter. This method, which could safely extract up to eight astronauts, would require early venting of the crew module to equalize the crew module internal pressure with the external pressure. After venting is completed, the side hatch would be jettisoned. The crew members would then exit sequentially by using the tractor rockets. Required orbiter changes include addition of a cabin vent capability, modification of the side hatch structure to allow for hatch jettison, addition of pyrotechnic devices to jettison the hatch, and installation of the tractor rocket system.
The CAIB Report chapter 10, paragraph 10.2 Crew Escape and Survival also contains a useful discussion of the situation, recommended for further reading.
I personally remember proposals to incorporate an escape capsule containing the entire crew compartment; this is not mentioned in the report although this newspaper story alludes to them.. Also note that the proposals mentioned would likely be ineffective for ascent.
In summary, a launch-to-landing escape system should have been provided from the beginning; IMHO if the design could not support it, the design should have been reconsidered. Retrofitting a launch-to-landing escape system into the existing vehicle was not practical.