Yeah— that's one of those tomes that takes a superhuman amount of will for most people to push through if they're not being compelled by university course work. Very well-written but it's just a lot of information to absorb.
When I was a teenager, just before I started more seriously getting into programming on the school computers, a Russian friend asked his dad— a soviet university trained software engineer— how he could learn to code. His dad handed him a his stack of Knuth and said "read that and then get back to me."
Of course, the kid read 10 pages of Fundamental Algorithms, put the book down, and never tried to learn how to code ever again. Why would he have? As far as he was concerned, the very first steps were a big pile of confusing theory that wasn't bringing him closer to his goal of doing some fun projects with a computer at any point in the near future. He had to absorb all that just to start? Forget it.
Looking back, I think his dad probably thought it was funny and probably even joked with his coworkers about it... but I wonder how different my friend's path would have been if his dad handed him a copy of Learning Perl (it was the mid 90s) and helped him get ActivePerl installed on his machine. If your nerd machismo is so intense that you're alienating your own teen kid with it, you really need to reconsider your life strategy.
You should just pick a random section and start reading; or better still find a section relevant to something you have worked with before. Knuth is a clear writer and pretty funny, and there is a lot of great insight packed in there.
Note that TAOCP is more a reference book for professional researchers than a book to skim quickly cover-to-cover.
