Y'know, for a random BBC news article, that was pretty well written, and set out pretty well the advantages and disadvantages of these sorts of things. I stron

Y'know, for a random BBC news article, that was pretty well written, and set out pretty well the advantages and disadvantages of these sorts of things. I strongly reckon that the best way to learn anything is in a one-on-one conversation with an expert. The second best way, I've always thought, is a small interactive lecture where you can ask questions. The third best way is with a well-written book, and the fourth best way is in a non-interactive lecture. However, I can see that there's an opportunity for a new second-best option just below the one-on-one conversation with the expert, and that's one-on-one interaction with software that simulates an expert. (An infinitely patient expert, in fact.) It can explain something to you, test your knowledge of it, and if you don't understand it then it'll go back and explain it again. A really well written teaching system like this could make learning things far easier than it currently is. Writing these sorts of things, though, is incredibly difficult. I may be an expert on certain subjects, but I can't anticipate all possible lessons in advance. It would be a huge time and money committment to put such a lesson together -- you'd write it, test it, improve it, and slowly get to the point where it really worked well. In order to make this worthwhile, you'd need an assurance that the lesson, once finished, would be widely used for some time to come.