Looking at the ASCII chart, we find:
high low bits bits 0000 0001 0010 0011 0100 0101 0110 0111 1000 1001 010 ! " # $ % & ' ( ) 011 0 1 2 3 4 5 6 7 8 9This is why the characters !"#$%&'() appear where they do on a Teletype (thankfully, they didn't use shift-0 for space). This was *not* the weirdest variant of the QWERTY layout widely seen, by the way; that prize should probably go to one of several (differing) arrangements on IBM's even clunkier 026 and 029 card punches.
When electronic terminals became popular, in the early 1970s, there was no agreement in the industry over how the keyboards should be laid out. Some vendors opted to emulate the Teletype keyboard, while others used the flexibility of electronic circuitry to make their product look like an office typewriter. These alternatives became known as `bit-paired' and `typewriter-paired' keyboards. To a hacker, the bit-paired keyboard seemed far more logical -- and because most hackers in those days had never learned to touch-type, there was little pressure from the pioneering users to adapt keyboards to the typewriter standard.
The doom of the bit-paired keyboard was the large-scale introduction of the computer terminal into the normal office environment, where out-and-out technophobes were expected to use the equipment. The `typewriter-paired' standard became universal, `bit-paired' hardware was quickly junked or relegated to dusty corners, and both terms passed into disuse.