I’ve spent all yesterday playing around with tex4ht. I wanted to find out how hard it is to produce PDF and web documents from one LaTeX source without sacrificing quality of the PDF. So I worked on converting the diffy qs notes into HTML. The result is
here. I don’t agree with several default decisions that tex4ht makes so it took a while to get it right. For example, tex4ht does some equations (mostly inline ones) as regular html and some as pictures. The problem is complete lack of consistency. For example the Laplace transform symbol looked different, so the student would have to figure it out. Plus the html equations are hard to read sometimes if they are complicated.
Unfortunately tex4ht is another example of literate programming effort, which is a wonderful idea which in my opinion does not work. All the literate programming projects I have seen have terrible documentation. The problem is nonlinearity of the documentation and many times circular logic. Terms are not defined before they are used, plus there is no obvious starting place. The right information is hard to find. This is a problem with all things LaTeX related. Getting started is very hard because you have no clue where to find the right info. All documents seem to assume you know everything in all the other documents and you know where to find all those other documents.
That is not to say that traditionally documented things are better. That is mostly because traditionally documented things are undocumented. And many times traditional documentation suffers from the same nonlinearity and circularity. For example, trying to implement adsense was also a pain for lack of good documentation. The documentation is a series of FAQs essentially. Each answer is vague and refers to language defined somewhere else which is not linked. And even if linked, the structure is hardly linear. There is no one place to start which defines all concepts and then goes into detail in a tree like manner. Instead the overall document is missing, and you have a spaghetti of a not simply connected graph.
On the plus side I have made $1.59 on adsense since saturday. While that may not seem like much (it is not) it is the first kind of monetary support for the notes. Actually over $1 was from the genius page which is not related to the notes and is far less visited. But apparently it has more relevant ads, or the visitors are less stingy with clicking ads. There’s about 500-1000 hits a day on the diffy qs notes webpage, though on average only 20-25 new people per day download the PDF. Still it says something about how the server is nonbusy that my notes account for about 30% of the traffic.
There have been over 3000 unique IP addresses that downloaded the notes so far. 3000 copies for a math textbook is not bad I guess. Obviously, it cannot compete with porn in the number of hits. But it seems that I am only getting cheap-o visitors. Not a single person bought the paperback even though it’s very reasonably priced. I get $2.50 from that. So thus far, adsense is getting me the equivalent of a little bit over half a copy sold. Outside of time, I have spent about $20 putting the notes together. So I guess I am still way underwater.
So … BUY THE BOOK!