Both of those books were recently (as in two days ago on here) in a list of "Books people claimed they read but really don't."

I've tried a couple of times. Gearing up for another attempt soon.

But there's one thing I've found interesting. I'm still stuck in Chapter 1.2: the infamous Mathematical Preliminaries section. I am not proud of this. From what I can tell, the various aborted liveblogging attempts I've seen also all get stuck here. Knuth himself recommends skipping it if you get stuck.

But even though I haven't yet managed to finish that single part of the book -this part which comes before any actual computer stuff- I STILL come out of it a much better programmer. Every. Single. Time. I haven't yet written a line of MIX or MMIX, and I have STILL been able to use this book to level-grind, just by getting further in 1.2 than I was able to get before. It's awesome. Even when I fail, I come out stronger.

I do eventually plan to blog my own attempt. My plan, though, is to not actually blog 1.1 and 1.2 live: I will do these chapters, write the articles on them to give myself a buffer, and only put up the blog once I'm already through The Big Filter. I figure this will give me a much better chance of actually succeeding once I start.

> I'm still stuck in Chapter 1.2: the infamous Mathematical Preliminaries section

You might want to try starting with "Concrete Mathematics" (also by Knuth): it's a longer intro to the same material, but covering a lot more of the background that TAOCP just kind of assumes that you already know.

SICP is still a very good book for learning abstraction techniques, although it does sort of throw you in at the deep end in some ways.

TAoCP is great as a set of reference books that you can take a look at when stuck on a problem, but I don't think it's really designed to be read from cover(s) to cover(s).

I certainly don't think TAOCP is designed to be a reference. If so, it wouldn't be full of excellent exercises.

Agree wholeheartedly. I've seen people suggest from time to time that TAOCP should be used strictly as a reference book (like a dictionary or a thesaurus) - I suspect people who try to use it that way are going to be disappointed when they look up, say, the Trie algorithm and find out that the text assumes that you're already proficient in MIX programming, and back-references a half-dozen formulas from previous volumes (sometimes from the exercises!)