If you're trying things in a vacuum, do not get reviews from people in know of how things work, do not do research past "this seems to work for me, time to disseminate", you will probably be missing on a lot of content.
People who get to the task by actually forcing themselves to learn first, get some peer reviews, go straight to sources and take a broader mindset of "I have to get this right because it costs more time and energy to undo bad teaching than teaching in the first place" might do better. In fact, they might get the benefit of teaching while still having a fresh perspective on a given topic, something extremely lacking with some teaching material.
Many universities have some kind of "teaching assistantship" or "teaching fellowship", that is, they pay students who've already taken the class (or a similar class) to teach and hold office hours for other students.
This was my first semester doing it, and I learned quite a lot on a subject I thought I already knew very well. The thing is, when you have to teach something to someone during a section (or recitation, or whatever you call it) or during office hours, you must be prepared to answer unexpected questions, or explain the difference between two concepts that are very close. Best of all, you must come up with different examples and mental representations of your subject, hoping that one of the approaches will resonate with some students.
Now some mentioned the fact that certain bloggers where going live way too early. Well, my guess is that they can do it because they don't receive real feedback. And no, a comment is not the same as a student telling you that you don't know what you're talking about. You don't want to become a TA to early, lest you make a fool of yourself, and trust me if students can burn you, they will (although most of them are really nice people, they tend to loose patience when confronted with incompetence, as most of us do).
But the process of writing down some kind of explanation in a clear way is still a good thing. Often, when I'm programming and come up with a neat way of doing something, I write a blog post about it, but I usually don't publish it for the same reason many of you mentioned. Publishing a post would require thinking more about the subject, trying different approaches, testing the code for edge cases, etc, etc.. all things I don't have time to do. But it doesn't matter, writing is enough to wrap my mind around the subject.
- I started blogging recently, and found that having to actually write my thought in an organized way forces me to really learn and understand the topic
- When I was a student preparing for a competitive exam (one where your rank determines the school you go to and the other students in your class take it "against" you), some people refused to help others, thinking they would beat them in the exam. In fact, I found that helping other actually helps you even more, and overall, the "helping students" fared much better in the exam
Blogging about my progress has reinforced how little I knew. In going to explain a topic, then suddenly realizing I didn't know how to explain it, I just knew how it was supposed to work. This has put a spotlight on what I didn't know, and I find great value in it.
The other upside is that people apparently think I'm kidding when I say "I'm just learning this" in my blog, because I routinely get questions via IRC of a much more advanced nature from people trying to extend what I've taught. I like to help, so I always give it a shot, and usually that means diving into source and learning more -- win/win.
Of course, sometimes I'll go answer a question on something I thought I knew really well, only to find out that I don't. I found out the other day that I don't really know how Django's form binding works. Somebody pointed that out, and now I do.
A sub-topic about this could also be that "Discussing something is the second-best way to learn something." I've never been one of those guys who hung out in chat rooms all day, but honestly, I didn't know what I was missing. Lurk around in #python for a day and I almost guarantee you'll see something you didn't know.
5% from lecture.
10% from reading.
20% from audio-visual.
30% from demonstration.
50% from group discussion.
75% from practice.
90% from teaching.
(above copied from somewhere but there is a lot of information on this subject: John Biggs is a good source).
I would have students lecture classes for me (I would have them lecture to me the day before to make sure they had the lessons figured out).
Learning is an incredibly complex combination of content, medium, background, strategy, mental state, mental ability, and more, and it varies widely based on subject material. There appear to be some advantages for certain teaching strategies under certain conditions for particular populations, but there is very little research that clearly establishes a more generalizable trend.
I've tried to find the original scientific research myself and have concluded that it does not exist. I'm not the only one who has come to this conclusion either.  I would love to be corrected on this matter though. These days, I tend to use the following instead, which also matches my intuition about learning:
 J. P. M. Lalley, “THE LEARNING PYRAMID: DOES IT POINT TEACHERS IN THE RIGHT DIRECTION?,” Education, vol. 128, no. 1, p. 64, Fall 2007.
The more I teach others, the more I learn myself.
Still, it's a good reminder to take the teaching approach wherever practical.
But... Probably the absolute best way to learn something is have a deadline where you have something to lose.
That's why I have moved to volunteering to teaching training courses whenever I can.. It forces me to learn it better than anyone, and the pressure of losing reputation if you get up and make a fool out of yourself is a great motivator to get it right. When you are done, you find you learned it faster than anyone, and are now the expert.
Being able to handle that setting is also a useful skill, of course, so you could argue it's fair to test. It's not the only useful skill, though, and it's not clear to me if it's being overweighted (even when I actively try not to). For example, in a computer graphics course, there are students who could do brilliant work if you let them sit alone for 60 minutes with a compiler (or with a piece of paper and some formulas), but who don't shine when examined orally, and vice-versa. There are some students who will do well in all modalities, and some who'll do poorly in all, but I think there's a significant number who will do differently depending on whether your exam is a 30-minute oral exam or a 3-hour take-home exam, so that choice really changes what you're testing for.
Given a sufficiently inquisitive/rigorous examiner? Very yes.
However, I think it's weak that the 'best example' is to 'hazard a guess' that Sal Khan learns more from teaching than just reading. There's no actual evidence in that statement, it's effectively just backing up 'I think' with 'I think.' I find it disappointing because I'm sure there are good arguments that could be made for his point.
I've embraced this. Yes, by gum, I will stand in front of a room full of people who have paid thousands of dollars each to be there, I will plow thru the material with gusto and enthusiasm, I will create content on the fly if I have to based on my experience and skill, and if I make a mistake or slam into an embarrassing dead end then I will accept it and use it as a teaching moment to explain that even the best make mistakes or get stuck and that I'll get them a correct answer ASAP and in the meantime we will either start the topic over or move on to the next one. This is reality, this is me, this is what real developers have to deal with, and I'm not going to apologize (per se) for it.
No fear. No apologies. Do the best you can in front of an audience because you love the material. If you're going to be inaccurate, then be _definitively_ inaccurate.
In the last training section I ran I hijacked 7 of my employees to sit through the talk. For me this is a no pressure scenario and nothing to lose. It went well, and when I moved to a larger possibly hostile audience I knew what to expect... As my employees were already brutally honest and I survived.
Also, if you are wrong, better to learn it now (that's why you learn so much faster as the teacher) than complete life and never know! When I'm wrong I take that as a positive thing, the case where I probably learned more than the student did.
Teaching something to show you are retaining it isn't the best way to go, though teaching something to show you have retained it is better.
Yeah, a lot of people who do have something interesting to say remain silent through fear of something irrelevant being picked up on and used as the basis for abuse. Something I don't recall seeing here.
I'm in the school of Bloom's Taxonomy for learning objectives. It goes like this:
4. Analyse, evaluate and create
This grading goes from the known and concrete to the unknown and abstract. But, as said, there are many ways to look at this. Some say that the fourth point can be separated into three points, other's argue they are of equal level of skill. Just take a look at the Wikipedia article for Bloom's Taxonomy (http://en.wikipedia.org/wiki/Blooms_Taxonomy) and you'll see that it even portrays several different perspectives to one (actually two) theory.