This is my impression as well. Kinda similar in its "bragging rights" to The Art of Computer Programming.
GEB was a frustrating read. I mean, it's interesting in places, but it's just all over the place, jumping between many different topics. The central theme is meant to be the strange loops, but it's IMHO not very interesting concept and his application on the cognition is just author's personal conjecture.
It's utterly unlike TAOCP. One is a comprehensive algorithms reference full of (hard) technical problems. The other is an extended personal essay. (Neither one is worth "bragging" about reading in my opinion.)
"Reading" all of TAOCP would take literally years of intense effort even if you set aside all other activity. There are a lot of great problems inside, and plenty of dry humor, and I would recommend people try to at least skim sections of TAOCP which seem interesting or relevant to their work, but very few people are going to even nominally work through the whole thing, and the people who might are professional scholars of the topic.
Reading GEB can be done leisurely over the course of a few days or maybe weeks, depending on how much time someone spends reading every day. It's not quite as easy a read as a pulp novel or comic book, but it also doesn't take any inordinate amount of work to make basic sense of, or require any special skills or background understanding to start on. It's a fun book to hand to a ~13–16 year old.
You are absolutely right. It was a great book to hand to a 21 year old me.
I've often read the hate on this site for this book. At least for me, I find the discussions and analogies to help me in thinking about, and eventually understanding the material. I contrast it with a graduate intro to Recursion Theory which can leave a reader feeling that they followed all the precise arguments but still somehow missed a lot.
I compared the two in the sense how it's fashionable to have them on your bookshelf, but IMHO few people actually enjoy them and understand them beyond the surface level.
This discussion is evidence that some people really liked GEB and other people found it boring or too unfocused. It can't be that many people who bought it just to look cool on a shelf. The people who found it boring should perhaps try to appreciate that sometimes other people can genuinely like things they don't like (and vice versa I guess).
Again, if you do any work with computer algorithms, it's worth checking out TAOCP at the library and skimming the sections relevant to your work. If you might need it as a reference, it's not a bad source to have at hand; I look things up in there maybe a few times a year for the past decade. Some parts are now a bit outdated in this fast-moving field, but it's still the best available survey source about some topics, and there are some nice explanations and a lot of great problems in there. Knuth is a pretty funny writer if you enjoy dry humor.
GEB was a frustrating read. I mean, it's interesting in places, but it's just all over the place, jumping between many different topics. The central theme is meant to be the strange loops, but it's IMHO not very interesting concept and his application on the cognition is just author's personal conjecture.