Secondly, use links as shorthand. Kouwe and Lehrer were both brought down by the fact that they felt the need to re-write what had already been written elsewhere. On the web, you never need to do that. If you or someone else has already written something well, just link to that, rather than feeling the need to repeat it.
Thirdly, use the blog to interact with your peers, rather than just primary sources. There are hundreds of great science and ideas blogs out there already; start reading them, and be generous about linking to them. Your readers will thank you. When you see an article you wish you’d written, link to it and say so. When someone finds a fantastic paper and writes it up in a slightly incomplete way, credit them with the great find, and then fill in the blanks. When two or three people are all talking about the same thing, sum up what the debate is, and explain where you stand.
Fourthly, iterate. Lehrer is a big-name journalist at a major publication: when he writes stuff, people respond, often on their own blogs, and often with very keen intelligence. Link to those people, learn from them, converse with them via the medium of blog, and use that collaboration and conversation to hone and further develop your own ideas. Treat every blog post as the beginning of a process, rather than as the end of one.
As the editors of the American Chemical Society write, self-plagiarising is fraud, because it is “an intentional attempt to deceive a reader by implying that new information is being presented”. A blogger should never feel the need to do that, because blogging is not at heart about delivering new information, so much as it is about finding and linking and connecting and conversing. Once you internalize that, self-plagiarism becomes a non-issue.