I'd expand this a bit more. Firstly, I'd include non-scientific journals like business, politics, and humanities. Honestly, as a geek through and through, the best way to learn about the world is to figure out how other disciplines think. It gives you a whole new toolset and perspective.
The other thing I'd do is actually dive into a subject deep enough as if I need to pass a test. I'm doing that with statistics now. It's going very slowly, but that's kinda the point of this whole exercise.
Honestly, as a geek through and through, the best way to learn about the world is to figure out how other disciplines think.
Failing to do this is what a lot of CS people do that really annoys physicists, chemists, and other scientists. (Especially when we post comments!) We think that we can instantly grok anything that has good documentation. But not everything is an API. Other fields are full of very smart people, and they had to devote years to fully understand. It's like people who learn your native language, but speak it with a heavy accent and clumsy use of idioms.
The "how other disciplines think" as tools for your mental toolbox approach is articulated very well by Charlie Munger. It's fair to say that it's worked well for him.
A lot depends on how much you want to learn. I wrote a post about depth of knowledge in March (http://williambswift.blogspot.com/2009/03/learning-journal-a...) - lowest is recognition where you can remember things about it when you come across it again, usually just reading about something is enough for me to reach this level. Progressively more difficult, and requiring more work, are being able to recall information from memory without an immediate stimulus, understanding the subject matter, and being able to use what you have learned to make things or do original research in the field. I have several other posts in March and April about independent learning that you may find useful.
I think that learning is much more that "obtaining knowledge". I am almost at a point where I think that nothing interesting can be learned by only reading - anything that is worth learning involves doing.
If you want to learn something that is physical or social, this is a given but it also applies to, say, programming or even math. You don't really learn by reading - reading can only help you when you want to get started or get stuck at certain levels.
There is inestimable value in doing so. I find the approach especially valuable for learning new programming technologies — in practice, learning a new technology comes down to the questions: what does the code look like, and why? By looking deeply at the hardest parts of the source code, I find out how much I don't know about the software and learn more about the system itself.
The other thing I'd do is actually dive into a subject deep enough as if I need to pass a test. I'm doing that with statistics now. It's going very slowly, but that's kinda the point of this whole exercise.