But LessWrong is currently a set of articles.

And also an e-book (which is probably not finished yet, last mention here), that is still just a set of articles, but they are selected, reordered, and the comments are removed -- which is helpful, at least for readers like me, because when I read the web, I cannot resist reading the comments (which together can be 10 times as long as the article) and clicking hyperlinks, but when I read the book, I obediently follow the page flow.

A good writer could then take this book as a starting point, and rewrite it, with exercises. But for this we need a volunteer, because Eliezer is not going to do it. And the volunteer needs to have some skills.

And how many people do you think the article on akrasia has actually cured of akrasia?

Akrasia survey data analysis. Some methods seem to work for some people, but no method is universally useful. The highest success was "exercise to increase energy" and even that helped only 25% of people; and the critical weakness seems to be that most people think it is a good idea, but don't do it. To overcome this, we would need some off-line solutions, like exercising together. (Or maybe a "LessWrong Virtual Exercise Hall".)

do you disagree with "any decent therapist would try to provide many/most of these tools to his/her patients"?

Yes, I do. Therapists don't see teaching rationality as their job (although it correlates), wouldn't agree with some parts of our definitions of rationality (many of them are religious, or enjoy some kind of mysticism), and would consider some parts too technical and irrelevant for mental health (Bayes Rule, Solomonoff Prior, neural networks...).

But when you remove the technical details, what is left is pretty much "do things that seem reasonable". Which still would be a huge improvement for many people.

On the more scientific side, the idea of optimal scientific inquiry has been refined over the years, but the core of observation, experimentation and modeling is hardly new either.

That's the theory. Now look at the practice of... say, medicine. How much of it really is evidence-based, and how much of that is double-blind with control group and large enough sample and meta-analysis et cetera? When you start looking at it closely, actually very little. (If you want a horror story, read about Ignaz Semmelweis, who discovered how to save lifes of thousands of people and provided hard evidence... and how the medical community rewarded him.)