There was also an option of sitting down with paper and pencil and actually solving the problem, reinventing the wheel for the thousandth time. Or you could reach out to specific people you knew were likely to know something about the problem and socialize a bit by the way.
Both approaches had some good sides to them and it's a shame that they're not even considered as options now, which GP seems to demonstrate.
Yeah, it occurred to me that way back when, as I was learning C on a C=64 (though later I was able to buy a super powerful 386 machine) that as I was trying to code up alife and neural net programs my problems were rarely of the "What's the workaround for a bug in alife.js or alife.rb" sort.
I wrote out a lot of pseudo-code and drew diagrams and when stuff didn't work I could often figure it out by poking through the K&R book or the manual for my compiler.
As it became easier to spread code and ideas it also became easier to spread more sophisticated but sometimes opaque libraries and tools, and perhaps ironically it is those new libraries and tools that most require the increased communication because while they offer more power they also introduce more complex bugs and quirks.
It wasn't the Dark Ages.
Maybe more like the Early Dawn Ages.