I've never worked pre-internet. In fact, I've had internet access in all of my jobs, but I've definitely worked pre-"internet was a useful source for searching for answers to programming questions". I mean, we had Usenet, and you could ask a question, but we didn't have search engines that were useful in that capacity.
I definitely had a lot of books, but actually the vast majority of my information came from man pages, RFCs and specs. I actually learned C++ by reading the spec -- I've never read any C++ book in my life, even though I was a professional C++ programmer for at least 10 years. For learning the STL, I did it by reading the source code!
Probably the biggest thing I miss in the age of Google is that offline documentation is getting hard to find. One of the reasons I chose Rust for my latest side project is that it's extremely easy to install offline information for virtually everything.
I like doing web searches searches and Stack Overflow is really stupendously awesome. Especially when I'm working in a new language or framework, I can find idiomatic solutions to problems without having to read through thousands of lines of source code. However, I do think that newer programmers today are missing out when they don't learn how to search primary sources to get their answers.
If you have a question about how the language works, it really does pay off to look at the spec. In doing so, you learn about things in a larger context. I'm always grateful for SO answers that have links to primary sources like specs in their answers, but I've noticed that my younger colleagues almost never click on those links. When they have the answer they want, they are back to their own code. Often they miss important nuances because they are too focused on getting the answer.
Similarly, I have found that there is often a huge reluctance to read source code. A colleague once joked that he used dependencies so that he didn't have to read the source code. It's funny because it's true ;-) However, if you at least try to answer your questions first by looking at the source code for a dependency, then it will tell you a lot about that dependency (mostly: OMG! We need to ditch this ASAP! ;-) )
I used to have a collection of hundreds of RFCs on my computer. Really, anything that is about the internet has an RFC (or did back in the day, anyway). These days, even I'm lazy and don't bother to maintain offline collections of this information. However, it is worrying that I often run into developers who build very complex internet applications and don't even know what an RFC is let alone use them to learn how they should build their systems. They will have very ingrained opinions on how build things, but they have no idea how the HTTP protocol works, for example. Discussions on the topic of how to design something usually results in snippets of blog posts obtained through Google searches, rather than pointing to the relevant RFC and saying, "It works like X so we need to do Y". Negotiation of how to proceed often involves considerable heated discussion about which notable internet pundit we should trust the most.
Yeah, this is an "old programmer rant" ;-) However, if I could wave a magic wand and get younger programmers to all add something to their arsenal of tricks it would be to read primary sources early and often.