Interesting. I’d be more inclined to pick a programmer who read anything outside of his field.
Both of your approaches are interesting, but I don't know if they are in any way helpful. In my beginner's days - which was around 1983, on 3rd year in university, I read a lot of computer books - about IBM/3[67]0 hardware, about languages, about PDP-11 and its hardware and the languages, about UNIX, about DECUS C and other C compilers, Pascal, FORTRAN, PL/1, parsers, grammars, etc. etc.
But over time I discovered that I'm largely past that point. I don't think I read any hardcopy book about programming in last decade, even though new languages like Java and C# showed up as strong players (and I used Java, and use C# currently.) But somehow I managed to learn them without reading books. Probably it's because I have some basic knowledge, and whatever extra is required to code in this or that language I just get from the Net.
Often reading a book is plain counterproductive. A full description of a .NET API may easily exceed TAOCP by volume. But can you really learn a thousand classes, each with hundreds of public members? At best you can remember a few concepts, but they aren't that difficult, and they already changed a few times between WinForms and WPF. Your best book is under that F1 key, especially considering that it is linked online to even larger depository of sample code and documentation and "best practices" articles. When I needed to attach a custom font as a resource I learned the way to do it not from a book but from Google.
Today I read a lot of SF. I don't read about Perl. I don't code much in it, and when I do I have bookmarked everything that I need to do what I need. Today I'm not in a position of a fresh, green graduate who has to spread thin to meet requirements of many potential employers. Today I know my sphere of comfort (analog/RF, microcontrollers, firmware, and PC software to operate hardware.) I don't need to get outside that and study stored procedures in Oracle, for example. There is a lot of demand for someone who knows the difference between S[2,1] and S[1,2] and can quickly put together a system that measures both, simultaneously :-)