Comment by pserwylo | Hacker News Reader

Comment by pserwylo | Hacker News Reader

Parent Full thread

Firstly, thanks for doing what you're doing. Even if it is a hassle to get it all together, I hope that the more people engaging in this type of effort can only be a good thing.

I agree that the coherency of the text book would be poor if everything was written by different authors. In fact, it is often so jarring when I read a journal article with different writing styles, that it is one of the easiest ways for me to identify plagiarism in student work or when reviewing for journals.

I guess the answer would be to have particularly committed and experienced editors like you mentioned in your attempt.

Perhaps if the book started of as quite a basic book, with key concepts, written by one person. Then, over time, that person (or small group of people) could accept pull requests from the community, and modify them to fit the style of the original book.

This could work quite similarly to most open source software projects:

1) Start of with one or a few people

2) Build only enough functionality for a working/useful piece of software

3) Over time, build up functionality via community contributions (and further contributions from the original contributors)

4) The direction of the project is steered by a group of people with its best interest in heart

Note how in the example of software, there is a fully working thing (even if it doesn't have all the bells and whistles attached) quite early in the process. Also, the overall vision for the project is kept in line by the original maintainer, who runs an eye over each pull requests and is free to make suggestions for changes. If there is any major disagreements, people from the community are free to fork the software and make it suit their needs.

I don't know how practical this is, in that a text book without huge amounts of detail may have less usefullness than a piece of software without huge amounts of functionality. Perhaps there are some fields which would benefit from such an approach, while others really do need to have a comprehensive, "finished" book to be of any use to anyone.

> a basic book, with key concepts, written by one person [...] without huge amounts of detail

For math and physics you can get pretty far with definitions and key formulas. As for succinctness of the explanations, it all depends of the "level of description" for the coverage. The more math notation used, the shorter the text will be, but also less readable.

In addition to the notation, you must control the vocabulary and the level of abstraction which you use in the narrative. The language level in advanced math textbooks is level="grad student", most science textbooks assume level="some college", most science news articles are level="high school", general media assume level=middle school", except Fox and CNN who give primary-school-level explanations of World politics.

In my experience, writing explanations for the "some college"-level readers can be quite succinct so I think your idea of a pared-down minimum viable textbook is possible. For example, a short intro to machine learning could start from 100pp of short-form lecture notes that explains the basic ideas and formulas, solved example problems and maybe some code---godda have some code in there if it's about ML. Writing such a text is just 6 months of work, whereas writing a full ML book would take years. Once the basic notation, definitions, and math prerequisites are written down, there could be 20-30 separate applications chapters (10-30pp each) where a single idea of ML is described by someone knowledgeable about it (perhaps not a researcher, the best would be a grad student). Bonus points for complete code examples. Pull requests for typo-fixes and requests for additional explanations ensue. And Bam!, ML textbook v1.0 done. A textbook in a year.

(v2.0 one year later can add more applications, add applications of the applications, or additional introductory material (tutorials) to make the book more accessible.)

> it is often so jarring when I read a journal article with different writing styles,

I think this will be and important problem with book-collab projects... How do you meld the voices (and egos!) of 30 people in a single narrative?

> I understand that "getting your name out there" is not motivation enough for doing such things. However, surely just doing it for the good of education and teaching is motive enough?

I presume you must be under 30 from your statement since: Anyone who is under 30, and is not a leftist, has no heart; and anyone who is over 30, and is not a capitalist, has no brains. (http://en.wikiquote.org/wiki/Winston_Churchill#Misattributed)

There must be some sort of monetary incentive. Maybe "book writing" funding from fed-gov or prov/state-govs? Maybe paid by universities? Either way there must be money involved if we want to motivate people to invest one year of full time work to write a book on subject X. Making money of the printing seems like the best choice to me... and, interestingly, charging for printing is totally compatible with GFDL.

I think print-on-demand and self-publishing is a good first start---it puts more power in authors' and readers' hands, which are the only people that matter in the value chain. It's going to get very interesting in the book publishing sector in the coming years, I promise ;)

> perhaps not a researcher, the best would be a grad student

This reminds me of one of the other frequent failure points of having authors each write a chapter: they write a chapter that is really more of a survey paper about their favorite technique (maybe one they invented or have worked on). Usually that ends up with a choice of content and writing style that's a poor fit for a textbook. A grad student writing about something that is not their own research might be a way of avoiding that.

I agree some kind of incentive is probably needed. It could be monetary, could be advancement-based, depends on the people writing and their situation. On the other hand, money might help for both: if there were grants for textbook writing, which e.g. a professor could use to buy out some of their own teaching and pay their summer salary, it'd also help out with the angle of "if I spend a year writing this book instead of journal articles, will I still get [tenure | full professor | etc.]?", because getting grant money is usually counted as a plus for advancement.

There are some areas that have stronger internal incentives, mostly those in which there is significant controversy, and people have strong opinions about the Right Way to do something. The belief that existing books were doing it wrong and he could write one that introduced things correctly was what motivated Robert Harper to write his 2012 programming languages textbook, for example. Another variant of this is if there are no textbooks on a subject so far, and you feel it's as a result being overlooked: then writing a textbook can be a form of evangelism for your new field. There is less incentive to write a good textbook "for free" on established and uncontroversial subjects.

Reply on news.ycombinator.com