Wire wrap was indeed the way to build even very complex digitial systems in the days before multilayer PCB fabrication and layout technology was as highly developed as today. In such systems, a simple 2 or 4 layer board was used to provide power and ground plaes only, but no signal routing. Sockets were inserted, and picked up the power connections. CNC machines then wrapped the boards automatically from a netlist file. Note that the problem of automatically routing insulated wires is much simpler than routing PCB traces, since cross-overs, layers, etc, do not enter into the problem.
A lot of hobby-grade stuff has given wire-wrap a bad reputation, but in fact done properly with automated equipment it was demonstrated to be more reliable than soldering, multilayer construction, etc. were at the time.
A big issue was that the lengths of the wire wrap pins significantly limited the packing density of circuit cards. In an effort to addrses this, Kollmorgen Corp. developed a process called multiwire, in which insulated wire was atached with adhesive to the surface of a bord, using CNC machines. The wire ends terminated at plated through holes in the board, to which IC could be soldered in the usual way. The PBC etch itself contained only these pads and power/ground planes. This reduced the thickness of the assembly over wire wrap, by eliminating the pins. Reliability was also increased because the chips could be directly soldered, rather than socketed. Kollmorgen sold off this technology in the late 1980's at which point it was a serious money looser. Astonishly, to this day Hitachi continues to make Multiwire boards, except that they now imbed the machine-laid wires into an internal resin layer of the PCB. See www.hitachi-chemical.com/products_pwb_05.htm
I have many memories working with wire wrap. On one of my first design projects, we built up three racks of equipment for software and test to use. On each rack were four assemblies with four or five wire wrapped cards and wire wrapped backplanes. We had three technicians to support all the changes we were making to the boards. Two of them were extremely fast and could install the changes fairly quickly, but then I would spend 2 or 3 times as long fixing their mistakes. The third technician was slow but thorough. He might take twice as long as the other guys, but I hardly every had to fix his work. I got to be pretty proficient at looking at rows of wire wrap pins for bent pins and loose wire. It was a real adventure for me and I learned a lot, but I wouldn't want to do it again.
aaarrrgghhh indeed! I too remember doing an ISA plugin card with wire wrap -- my first experience with using a PC as test equipment and test controller. I remember being extremely grateful for those little flag labels for the wire wrap posts, which eliminated a lot of excessive pin counting and wiring errors.
Wire wrap aaaarrrgghhh!!! The worst I saw was a wire wrap backplane for a tank gunnery simulator that had what looked like thousands, but was probably just hundreds of wires. Soon after it was finished and tested, someone inadvertently dropped a heavy wrench from quite high that fell through this mess and ripped quite a few wires out and probably made some other connections unreliable. They chose to rip the whole lot out and start again....
The most complex job I ever did was making a PC ISA plug in card. That was hell enough for me.
Wire wrap can be reliable. I paid for several semesters of college wire wraping NASA ground support equipment. After graduation I worked for Sperry Univac. Their 1110 mainframe computers had wire wrapped backplanes
I once worked on a high speed image generation system (6 RGB channels of 1080 line video) using wire wrapped boards. Probably 200 - 300 boards about 18 X 9 inches, each holding 100+ ICs. It was housed in 12 84 inch high racks. Unbelievable amount of wire.
I've seen an entire terminal controller - probably about 15-20 boards, each about 15 x 7 inches - done with wire wrap. I got given an old one and it was a rich source of spare 74LS chips, and (once I had the wire off, which was a pain to do) of IC sockets which I did not have to buy for a long time.
I guess it was one way of building boards, especially short runs where the PCB design time would have been longer than the wire wrap time. (This was in the days before PCB design programs. Before PCs, in fact...) And I'd hate to think what the signal crosstalk would be on modern designs running at over 10 MHz...)
An open CMOS gate input connection will indeed cause problems eventually. And yes, wire-wrap has been used in quite a few production devices. I even heard of a system including an array of logic built on those quick-prototype plastic breadboards. These got into the plants because either there was way to much of a rush, or because nobody responsible understood what they saw.
I learned about the CMOS inputs very early in my career, and after that every assembly had 10K pull down resistors on all external input connections. But for logic running on 12 volts we used 27K pull downs.