As I remember...
I s'pose it makes me a bit of a grey beard but I remember that the original Shuttle Concept was more like the shuttle on its way home from California, or Virgin's prototype/project.
There was to be a high altitude lifting body and a space going vessel. The lifter was to carry the spaceship to high levels and then it would fly away. IIRC this has the advantage that the ship can also vector on more of a tangent towards an orbit rather than having to make a lot of height with vertical velocity and then convert/manufacture a lot of horizontal (or strictly orbital) velocity.
Back when ITV was a proper TV station (and even Magpie looked at these things) I remember this all being in a "Space Annual" fronted or probably written in those days by a guy called "Peter ?????".
This design was deemed too hard/expensive in the end. And so the kludging started. The ship gained launch engines but no room for fuel, thus the external tank. This was still not enough so you get the fireworks boosters. As I recall even the failure of the O-Rings was at least partially due to a change in the operating temperatures/cycles between design and implementation and use. A lot like IT projects really but sadly more deadly (than most). Also lots of the ship had to come back, originally all of it. This was a mixture of early ecological thinking combined with a political/economic splash - ie. reusable = cheaper (even if its not) = faster turn around. The idea it was quicker to relaunch than to build and test. Like with aircraft - we fly those quite a few times in their lifespan.
There was also the US Military/NASA tradition of specifying most of the technology to crazy high levels, like those $M coffee machines on AWACs and Nimrods. Then skimping on the new fancy - hard for executives to understand stuff. (Hey, it is a lot like software projects!). Remember the design used several Bubble Memory Packs that were a) huge - small luggage sized and b) held about 64 KBytes. Laugh, but dynamic RAM either had not been invented or was expected to be vulnerable to alpha particles flipping the bits. As I recall even retail dRAM in 1982/3 indicated that you should use parity to protect against random-bit-errors! Not something we even engineer for in current earthbound PCs packing 4 GBytes. So times change. Although I think I'd like error correction or 3-way redundancy at 100 miles high.
Coat - mines the one with a James Burke special in the pocket.