Skip to content

The history of subject wikis

Subject wikis began some time around December 2006. I thought of starting a wiki on the extensible automorphisms problem, a question I had raised in group theory. A month ago, my college had used a wiki page on Editthis.info to coordinate a multi-threaded discussion on concerns about the college moving: the page is still there. So I decided to create a wiki on editthis.info. After working a bit, I realized that I could make my wiki encompass all of group theory. Hence was born “groupprops, the group properties wiki”. “Group” came because it was about groups, and “properties” came because I was using a mode of organization that I preferred, and that I had, to some extent, developed myself: the property-theoretic organization (here is a page with my early thoughts on the matter).

I found the act of writing the wiki addictive. In time I moved from the editthis.info address to the Wiki-site address (both outdated now). I also started more wikis on wiki-site, including ones in topology and geometry.

About a year ago, when I’d already done considerable work on the Group properties wiki, and started other wikis such as Diffgeom and Topospaces, I started seriously considering the idea of “subject wikis”. To do this, I needed to abstract out the general ideas and underlying principles that had led to the group properties wiki, and look at whether these ideas held good in a more general context. Around March 2008, I started doing more extensive research on wikis. During the March-May 2008 period, I sought help from a friend, Harish (now working at a start-up, 4AM) to set up a website for subwiki.org, and host all the subject wikis that I’d already developed, along with many more, as subdomains.

With more control over the hosting, I was able to install a variety of extensions and configure the subject wikis better. I installed several extensions to MediaWiki, the underlying software, the most important among them being Semantic MediaWiki, that allowed for the storage of semantic relations between data, leading to very new possibilities for organizing data. It took time for traffic to be diverted from the old wiki-site address to the new location.

Another idea I had was that of a “reference guide” to subject wikis: an entry point to the complex maze of subject wikis. The reference guide would be a place where a non-initiated person could type in any term and get a range of meanings across subjects, with links to individual subject wikis for more detail. In May 2008, I started the Subject Wikis Reference Guide, though it took till late August to have a handful of entries on this wiki. Sample entries include normal and simple.

Looking back, it also seems that there were some important shifts that occurred in my approach towards subject wikis. When I began, it was just a way of getting my thoughts about extensible automorphisms down in a tangible way, given the huge maze of links. Soon, it had a general focus of putting group theory knowledge down. Through 2007, as the overall quantity of material increased, the original property-theoretic paradigm that I had wanted to showcase decreased in relative importance. What I mean is that I started to use a variety of different paradigms in addition to the property-theoretic paradigm to organize material. Later, in 2008, I introduced text boxes and quotations, and started the convention that all article-tagging templates (the stuff that displays stuff on top of the article) appear in text boxes and provide useful exploratory links. For instance, Template:Subgroup property on Groupprops is very different from its first version.

This shift arose for many reasons. First, my initial interest was in rapidly adding data to develop something nontrivial. But later, my interest shifted to ease of use and the way the model encouraged exploration. I was also interested in increasing standardization and making users/readers feel more secure. Further, this was a period of time when, after a hectic graduate student schedule, I had a little more time, and I was using this time to surf and go through other websites. This gave me a feel for what works as a user.

Also, my initial naivete about contribution and the way it works was also replaced by a clearer understanding of the motivation behind contribution. I understood that the idea that a resource could be collectively built from scratch without a vision was not viable. Rather, if I wanted to encourge others to contribute, the first step was building something that was either directly viable or was sufficient “proof-of-concept”. As of now, contribution by others remains low, but there have been many people who have contributed a few articles, comments and insights related to some articles, or provided general feedback about the content.

Another thing that I started building was a guided tour to group theory, meant as a full-fledged beginner’s course on the subject. In Autumn 2008, I was College Fellow for the Math 257 course at the University of Chicago, where the main topic is group theory (Chapters 1-6 of Dummit and Foote) and I used this occasion to get a clearer understanding of how students learn, and also shared some of the wiki content with students. The guided tour is still not complete, but I think it has huge promise and I plan to complete it in the summer of 2009. Once there is more basic content on the other wikis, guided tours can be created in those as well.

With the move to the new domain in 2008, I also installed Google Analytics on the individual subject wikis, and this helped me get a better idea of who was visiting and what pages they were visiting. Apart from a slight hump during the weeks surrounding Christmas, traffic on the group properties wiki has largely been rising. Current visit counts range between 50 and 100 visitors every day, with total pageviews ranging between 100 and 400 pages (this measures human visitors, not search bots and engines that add massively to the page count). On other wikis, traffic is considerably lower, ranging from 0 to 10 visitors per day.

What I find interesting is that the group properties wiki has gone in depth and coverage where few resources in group theory have. I don’t mean to say that it covers more material than any book in group theory — rather, that the way it covers a wide range of material of varying degrees of difficulty, with a great network of interlinking, brings the subject to life in a way that I think is unprecendented. While the typical book, constrained by space, contains largely a uniformly chosen subset even of the stuff it intends to cover (the rest being left to the reader), Groupprops, with its non-linear structure, can cover a lot more, giving each proof in full detail, highlighting relationships between simple facts and high-level facts. Many proofs are more likely to be available in completely correct version on Groupprops than in one specific book. And yet, when it actually comes to examining a specific proof page or a specific definition page, there is still a lot to be desired. A lot of holes and omissions are visible. That’s partly because the standards for what constitutes a page keep getting raised and some old pages don’t get updated to meet the new standards. And also, sometimes I am lazy, and don’t fill in a proof.

But what these standards about what each page should contain do is: they systematically expose my knowledge gaps. This has led me to ask many interesting new questions, some of which are obvious after a little thought, thereby adding to my understanding just that little bit. Others turn out to require more hardwork and I need to ask my advisor or others for answers. Of course, not all these questions are of natural interest, but some of them might be. (This is subject for another blog post).

I hope that 2009 will be a new era for subject wikis, with more growth both in usage and in contribution, and that it will see more experimentation with new forms of organizing knowledge.