On checking:
When I started at FLIR Systems in 1994, if a mechanical engineer wanted to get a drawing onto the floor he had to have a draftsman draw it up, then the draftsman had to submit it to not one, but two checkers before it could be signed off. Thus, four sets of eyeballs had to look at the work before it was released.
If an electrical engineer wanted to get a schematic approved for production, he had to draw it up, then one checker would look at it before it could go to layout. Then the layout guy would draw it up, the engineer would sign off, and it'd go into production -- so, three sets of eyeballs.
When a software engineer wanted to get some code into production, he would toss a floppy disk in the general direction of the ladies in Documentation, sign the line that said "responsible engineer", and off it'd go. I always felt that if it were software, the line should read "irresponsible engineer".
I brought this up as a potential quality issue, as a software engineer, and was shot down. I concluded that while mechanical and electrical engineering were, indeed, regarded as engineering disciplines, software was seen as magic.
(To their credit, FLIR did, eventually get their act into shape -- you didn't even get to send your code off to SQA until it had been reviewed by your peers, then SQA tested the snot out of it before it went into production. The system still wasn't perfect when I left, but it was better and getting better yet.)