Most cricket fans who are "online" don't need any introduction to CricInfo - the mother and father of all cricket websites, "The Home of Cricket on the Internet", as it has called itself for a decade and half now. It's a treasure trove of cricketana - live scores & commentary, reporting on cricket matches, articles, profiles, statistics, you name it. In fact the richness of its collection has made it somewhat hard to discover the hidden gems... most users probably use only a few regular features from its front page, never knowing what they're missing.
What makes CricInfo even more interesting, and perhaps a case study for business schools, is its origins. David Liverman has written up a very nice - and mostly accurate - history of CricInfo, which is definitely worth a read. In this post, and subsequent ones, I'll try and relive my little role in the setting up of CricInfo.
As I've mentioned in a previous post, I was a graduate student in Minneapolis during the 1990s. Our only sources of cricket news in those days were (week-old) Indian newspapers, and the Usenet newsgroup rec.sport.cricket, where some folks from England and Australia would post updates for the benefit of those like myself, stranded in cricket-blackout places like the US of A. We also used to hang out on IRC (Internet Relay Chat, one of the earliest instant-messaging systems), on a channel or "chat-room" called #cricket - hoping that someone from the cricket-enabled world would log on and give us score updates. On a lucky day, one of these kind souls would actually type in ball-by-ball commentary for us while watching a game on TV. Soon the word spread and IRC #cricket attracted hundreds of people, all clamouring for scores and updates the moment they joined the channel! We ended up having to make it a moderated channel with only the "commentators" allowed to write to it, and had another channel called #crickettalk for the masses to use for discussion, score requests, etc.
One of those IRC hangers-on was Simon King - whose IRC nickname flitted between ColdPom, CoolPom and occasionally even WarmPom depending on the Minnesota weather! For like myself, he too was at the University of Minnesota, as a post-doctoral fellow in the Chemical Engineering department. Simon was somewhat irritated at the constant clamouring for scores and scorecards on IRC - well, so were many of us regulars, but he actually did something about it. Simon himself had no computer science / programming background, but with the help of friends on IRC like Mandar Mirashi, he created the first incarnation of CricInfo - a "bot" on IRC. This was a program that would join the #cricket channel with the nickname CricInfo. You could send it a private message over IRC, asking for the latest scorecard etc. using a very limited language of keywords. In response, the CricInfo bot would send you the scorecard as a private message - thus avoiding cluttering up the #cricket channel itself with all these requests and responses.
Now where would CricInfo get its scorecard from? CricInfo the bot was simply a program running on Simon's workstation at the University of Minnesota. Simon had to keep a scorecard (a simple text file) updated on that machine, by watching the ball-by-ball commentary on IRC. Since it was impossible for one person to do that through the duration of a cricket game, some of us volunteered to help. One of us would log into the CricInfo account and keep the scorecard updated - not quite ball-to-ball perhaps, but pretty frequently. Thus we ended up forming a community of volunteers, all recruited from amongst the IRC #cricket regulars, who helped by creating the content for CricInfo to "serve". Soon, this went beyond the latest scorecard. We started populating CricInfo with older scorecards, match reports and even the Laws of the game, all painstakingly typed out from print references like the Wisden Almanack, Sportstar magazine, etc.
So, in its early days, CricInfo was only available on IRC using a very limited command language. It had a relatively small user base - only those who were aware of IRC and CricInfo, and were sufficiently comfortable with using that command language to request files from the IRC bot. CricInfo the bot would keep statistics on its usage, and I remember we rejoiced when the usage touched 1000 requests in a week. This must've been in early 1993. Still, even this level of usage couldn't be sustained on that workstation in Minnesota - it was apparently using too much network bandwidth, and Simon was requested by his system administrator to shut CricInfo down! Luckily, one of those IRC regulars, Prof. K.S. Rao offered us the use of a PC (an 80386-based machine!) in his office at the North Dakota State University. So CricInfo moved to tulip.ee.ndsu.nodak.edu, and we breathed a sigh of relief.
Meanwhile, I had become aware of a distributed information system called gopher - created incidentally at the University of Minnesota by some of its IT administrators. This was a precursor of "the web", much like the http-based web servers that were to follow soon. Around that time, many US academic institutions had installed gopher servers, making information available online through simple text menus. A gopher client (a browser, in today's terms) could connect to any of those servers, navigate the menus, and access files containing mostly text-based information. This seemed like the ideal interface for CricInfo, and I downloaded the gopher software onto Prof. Rao's machine and installed it atop the same directory structure that the IRC bot used. At one stroke, all those scorecards and articles that we'd been accumulating became available via gopher clients.
I "advertised" the new gopher interface by posting an article to the rec.sport.cricket newsgroup. Within days, the usage of CricInfo had exploded - apparently, many more people had gopher clients available than IRC clients. Also gopher was more friendly with bandwidth usage than IRC, and its response times were much quicker. So it quickly became very popular.
By now we had a motley collection of volunteers helping run CricInfo, doing all sorts of tasks - maintaining live scorecards, typing in older scorecards, keeping the 386 machine running (not an easy task with the load imposed on it), answering user queries at a "help desk" email address, etc. We called ourselves "The Management", rather grandly! Most of us were in academic institutions in the US, UK and Australia, either as students, post-docs or faculty. We "met" and talked to each other only on IRC and email - very rarely in person or even on the phone. As an example, Simon King and I have met just once (over lunch at a campus restaurant), despite being on the same university campus, and in adjacent buildings in fact! CricInfo was thus an almost purely online, collaborative venture.
One of these volunteers, Sridhar Venkataraman (at Arizona State Univ), had been playing around with this thing called an http server, and was raving about it. He and I chatted about it on IRC, and I tried it out as well. It seemed to be very similar to gopher at the time, except for the cool new thing called "hyperlinks" and the ability to embed images in text documents! We discussed it with the CricInfo Management. Given the kind of content we had - plain text scorecards, articles etc. - we decided that we didn't want to mess around with this http thingy! Gopher was doing just fine, thank you, and we had no use for these hyperlinks and images! Zero marks for foresight, I guess :-) Of course at that time, most of our users already had gopher clients installed on their workstations, and hardly anyone had http clients available. That was to change soon, and quickly. People had started playing around with lynx (a text-based http client) and then came mosaic, the graphical browser by Marc Andreesen that really launched the web revolution. Soon enough, Sridhar helped set up the CricInfo http server, and the rest, as they say, is history :-)