A bot called Cricinfo

We were graduate students in the US. I was at Cornell University. The problem was getting hold of cricket scores of any kind.

Today to imagine a pre-internet era is really difficult. There was no television that broadcast cricket, the [foreign] newspapers were received a week to ten days after the matches actually happened. No internet, of course, though the university had internet connectivity and email.

Cricinfo was the culmination of a series of attempts that were made to simply inform a guy that a match had happened a day or five days back and this was the result. That was actually it, nothing beyond that. The idea of "live" coverage was not there to start with.

A few attempts had happened. One very credible attempt was the Usenet newsgroups, which had started happening around 1986.

To start a group there is a process. So if somebody proposed that a group be created for, let's say cricket, somebody could come and oppose that, saying that there is no need for it. There were always a bunch of crazy guys who would oppose the creation of any group - people who would be called trolls today.

There was canvassing, saying we're going to start rec.sport.cricket, and all of us will have to vote otherwise we will not get this group. We are talking about a period of scarcity, where there was pressure on resources. Spaces had to be allocated by system admins, who looked at it as a pain and so said that a certain number of people should vote.

There was a very primitive level of cricket information that used to be discussed in the newsgroup but some of what you would call good cricket writing as well. Someone would write about a match that he saw in, say, 1970, in which England won amazingly or whatever.

Mailing lists, IRC and all that
Then the mailing list comes in. That is where KS Rao, who was a professor at North Dakota State University, comes into the picture. He had access to a large computer. He created two mailing lists where anyone could send in their email id to subscribe. The mailing lists were created maybe a little after the newsgroup. Content from the newsgroup was distributed to subscribers of the mailing list.

In 1992, during the World Cup, our main consumption of cricket was through rec.sport.cricket and cricket s, which were the two mailing lists. cricket s was for scores, for the guys who said, just send me scores and nothing else. Both lists were running out of the North Dakota State University computing facility through Prof Rao.

The same content was also posted in the newsgroup, which was done by a chap in Australia called Robert Elz, who was part of the project to link Australia to US through an undersea cable. So he was the guy with the best internet connection in Australia. He used to do five-over summaries for every match that he could watch, and have it posted to rec.sport.cricket. That was the closest to the first ball-by-ball commentary.

The BBC used to have short-wave broadcasts of cricket commentary, broadcasting it through a Caribbean station, and lots of us who were close to the east coast of the US had it quite clear. They even had a replication station in Syracuse [New York state], if I'm not mistaken. So there was some kind of radio coverage for it, particularly when people toured West Indies. But the BBC stopped all short-wave cricket broadcasting completely when their funding got cut, and our desperation levels really mounted.

The cricket commentary was available in the UK through BBC medium-wave and FM, and some guys in the US - paid grad students, subsidised by research grants, who could take a day off for doing cricket commentary - figured out some kind of encoder and receiver and shipped it to someone in the UK to connect it to an FM radio. That device would encode it and broadcast it through the net to the US, where someone would be able to listen to it and do a score update.

It was in the end of 1993 that I got more involved. IRC [Internet Relay Chat] became very popular during the Gulf War. I do not know whether Simon [King] was the first one to get the idea that it could be used also to provide cricket information - the idea that a person can sit in front of a network and give updates.

The early days of IRC were just an ego game of people trying to create groups, give themselves a higher status, friends giving friends status. It seemed like a silly environment but it was in such a silly environment that Cricinfo actually emerged.

It was Simon's genius that he looked at the bot that gave privileges and said that if they improved it, it could do other things also. At this point the group was such that if you joined late, you could only see what was happening from that point onwards and nothing before. The first thing that people asked when they joined was what the score was. Someone would reply with, "It's 140 for 3." And then the next fellow would come and ask the same thing. And if you looked at the transcripts of the group chats, you would see a continuous repetition of "What is the score?" "140 for 3", "What is the score?" "140 for 3."

That's when Simon came up with the idea of a bot to provide the information. It was a query-based thing. It could communicate with you if you put in certain keywords. Simon went by the nickname of "Cool Pom" on IRC, and the bot was under the name Cricinfo. Others helped him certainly - the other programmers who were creating this bot, they were much better at doing the programming, but they had no idea of what to actually do with it.

In every IRC channel you could set a title, like the title of a webpage or what you would call today a status message in Facebook. This was the "topic". When you joined the channel, the topic would flash. The Cricinfo bot would regularly set the topic with the latest scores of all the matches, if someone gave it the information, which anybody could do.

Because there would be all kinds of trolls who would give wrong information, a set of approved people was made. And there were other privileges. You could ban somebody from entering the group, you could throw out somebody for a while if they caused too much of a ruckus. It was like a schoolmaster managing a bunch of kids. If there are some transcripts from the IRC channels, you will see the kind of conversations that went on.

I remember one particular incident from those days. We had an Indian kid from Hong Kong who was following cricket somehow, and he started covering one of the many one-day internationals India was playing at that time - I think it was one of the Sharjah matches. This kid started providing cricket scores and no one realised that he was a high-school kid. And at some point some people were goading him and making fun of him and he got so angry that he started shouting: Give me money. I've spent so much money giving you scores in the IRC channel.

It kind of gave you a flavour of what was going on. I will not give you the score unless you give me money. How do you give money? There was no way, even if you wanted to. No payment gateway or anything. You could see the kind of fragile people inhabiting that crazy world.

To convert it into a more sophisticated thing, it took a long time, and it took place only because of Simon King. He was very tenacious and he saw some kind of potential - I don't think it was business potential but a community potential. He clearly emerged as a single leader. It became sort of a dictatorial model, where Simon King decided that that is what Cricinfo would be. If you disagreed with him, you could go do something else. It was for the better. I didn't have any problem with it.

Proto-Google
The major limitation of IRC was that nothing was archived, and the networks were very fragile. Nobody was responsible for the internet and nobody was responsible for IRC.

That's when the University of Minnesota computer science department came up with Gopher. The same university as Simon, but Simon was in a different department. It just so happened that the University of Minnesota's computer science guys initiated this idea, and it was a server-client architecture model, which automatically provided some sort of stability as compared to a peer-to-peer IRC networking model.

Gopher was developed for maintaining a structured, ordered tree-based info storage and retrieval mechanism. The idea was that you keep this info in various university networks and you try to put together some sort of repository or indexing system. This was pre-Google, or you can say proto-Google. You needed a Gopher client, a browser. WWW or the W3C models and web browsers had already been developed at CERN, but Gopher had a graphical interface. I wouldn't say it was a great graphical interface but you could use a mouse and click on links. Neeran Karnik, who was a student at the university, was following all of Cricinfo's development on IRC, and with Simon helping, the Cricinfo Gopher was built by them.

Simon started building a back-end database, which was great engineering because the system was not developed or defined for that. He modified the Cricinfo bot so that people could not only give live scores but also archive scorecards. What you see today as the Cricinfo scorecard structure, its precursor was evolved at that time loosely by Simon.

He spread the idea and he would talk about it on rec.sport.cricket, encouraging people to throw scorecards into the system. Slowly a few volunteers started doing so. There were incentives. If you submit one scorecard, you will get access to 100 scorecards, or something like that. This was done to encourage more people to contribute. It was through that system that I started contributing.

Once I sent in a scorecard, Simon immediately pounced on me - and every other new volunteer - telling me and them to do more. In this way he was slowly organising a bunch of volunteers. I would send in a card once in a while just to maintain my privilege of getting access to more content.

Simon needed more volunteers in a managerial capacity because he did not have enough time to contact and communicate with everyone and get them to contribute. At that time, two of us agreed to be volunteer-managers. The other one was a chap called Shashin Shah. So now two of us had some kind of access to the system, along with a third guy, Murari Venkataraman, who did a lot of work and then again worked for Cricinfo in India and headed Cricinfo India's operations for several years.

"We should have paid Allan Border royalty"
The donkey work was put down to me, but I liked it, so it was not a problem.

That was when a bunch of statisticians, people who had access to Wisden Almanacks, started coming in. I remember one name, John Hall. He came up with this idea: why don't we type and archive every Test scorecard ever? This idea was floated on rec.sport.cricket and volunteers came and joined. I volunteered to type all India matches and John volunteered to type all England matches. I think Travis [Basevi, who later went on to build Statsguru on Cricinfo] was doing some Australian matches. He was a high-school kid then.

At this point everyone was typing the scorecards in his own format. Some ideas were introduced on how the formatting of the scorecards should be. The first attempted scorecard had only the score. It was only later, in 1995, that we said runs, minutes, balls, fours, sixes. At first there was just a fall of wickets as 1-100, 2-173. Later on, adding the name of the dismissed batsman came in.

The life of Cricinfo stopped at the scorecard. It was seen as an end in itself.

In mid '94 or end '94, Simon found some sort of code that was available and written in some horrible language called Tcl/Tk, which could be used to take the Gopher site and completely convert it into a www and http format. Simon was still not very comfortable with the http protocol, more because we didn't have good software to view it. Around then, NCSA Mosaic happened and we sort of disconnected Gopher, since http gave us a lot more facilities.

You could actually present an image in Mosaic, which Gopher couldn't do. Not that we had access to many images, but you could at least throw a logo. Simon developed the logo for Cricinfo, which was an image of Allan Border playing a shot, and it was slotted and blue paint was poured over it and that was the logo, which he was very fond of. Cricinfo written in Times New Roman, with both C's capital and everything else in regular size. We should have paid Allan Border some royalty for using his name!

Once the scorecards started pouring in, the value of Cricinfo went up. That was also the time that the attempted live scoring of matches started happening. A few guys in South Africa had attempted live scoring even earlier, probably in '93. They had developed a score program and called it "Dougie" because the name of one of the South African scorers was Douglas. He and his wife Lennie were always seen at cricket venues, scoring matches together. Dougie was a very sophisticated score program, written in C language, and I worked on the code subsequently, along with several other guys. In the spirit of the internet, it was a completely open-source code.

Obviously now you probably have far better versions of the software, but at that time to say that you could actually create such a scoring program in C with no graphical user interface and just key in input, it was dramatic.

The thing was that they were developing not only a score-crunching machine, they were developing an internet-connecting protocol to convey it, which was the difficult part.

The problem with the model that they came up with was that you had to score every single ball without interruption, and if you stopped, the continuity would be lost. So they needed volunteers who would continue if someone lost their connection. How would this be done if they were in different places? This is where they created a master-slave relationship. When I run this software, I become the master. Then you run it sitting wherever you are and it will throw up a bunch of existing matches that are currently running and you can become a slave of one of them or you can become a master and start a new score. The second volunteer would first become a slave by connecting to the master and everything that the master does is simply rebroadcast to the slave. You could also communicate to all the people there, connected in the private network. So you say, okay, I'm going to give up in this over, can someone take over? Once you give up, you log out and the next slave in line can input a command and become a master. This was soon discontinued because it was too much to handle. Volunteers were difficult to come by.

The pressure was on us, and primarily on KS Rao. He convinced Vishal Misra and got him to change the program, to make it into something that had to work in a robust manner. Vishal was one of the guys from rec.sport.cricket and he took the initiative to change the programming completely to make it work and be able to put out a live scorecard.

So we had a scorecard-generating engine but we had to find someone who could do the scoring, and that's when we got to this chap at the University of Louisiana. We needed money and a few of us put some in, much of it by Professor Rao, who had his own exclusive dish connection. We are talking about the pre-laptop era. He would take a desktop into his room, sit there and score.

Then he and I started alternating. Operating the software was the difficult part as it needed a steady, high-quality internet connection. For some matches he would feed me through a private conversation, mostly through an IRC private network, a messenger-ish conversation. He would capture the ball-by-ball info and convey it and I would do the scoring, sitting in my office. I had a reasonably free lab with no one bothering me.

Vishal would do some matches as well, as he knew the program in and out. So between us we were covering all the matches, and that's how we managed live scoring around that time.

At the end of '96 I moved to Chennai. Once I was there I had ESPN Star Sports, so I could score comfortably. We had people in South Africa doing South Africa matches and people in Australia doing Australia slowly. Travis probably started doing matches around that time. UK also had people covering.

With ESPN Star in India it was a major boon for us because they were getting rights to all the non-India matches as well, so at a low-enough cost we could actually cover cricket. Finding volunteers in India was relatively easy. Our problem was internet connections. In '97 we had very low-quality internet, even until about 2002.

This program, Dougie, was doing only the scoring. Then Vishal made some enhancements and so did Travis. Commentary was enabled by one of them.

We probably had proper ball-by-ball commentary in '97. I remember scoring and doing commentary for that very boring draw in the Test between India and Sri Lanka in Sri Lanka where they scored some 900 runs or so.

The next major goal for Cricinfo was to somehow cover every international match live. In '98 we started doing that. Maybe there was just the odd match that we weren't getting coverage for.

Partnering a Rolling Stone
It was one of the Sharjah tournaments. It was a series for which no broadcaster in Britain took the feed, because England had sent a second-string team.

I had no direct interaction with Mick Jagger. Simon was his contact point. At that time my music interests were such that I didn't even know who this guy was. When he figured out that there was going to be no video coverage whatsoever, he probably called one of his assistants to ask who is producing this, and he called Mark Mascarenhas [of WorldTel, who had the rights] and said, I want video, do something. So Mark said, look, I will give you access to the signal and you pay me for it and you do whatever you want with it.

The problem was figuring out how to broadcast it. So then he tracked Simon down and talked to him. Simon must have been pretty amazed because he was a fan. After that he and Simon became really good friends.

So Simon got the feed and we tried to figure out what to do with it. That's when RBN [Real Broadcast Networks] was doing a lot of streaming and they said they would charge us to stream it. We had no mechanism for charging people, but Mick said that whatever it was, he would pay for it all, because he wanted to watch it. From then on, every Sharjah series for quite a while, we had streaming on RBN. The whole thing would be put out on some pretty low-quality stream, some 56kbps or 128kbps.

You have to imagine the kind of desperation people were feeling - they would take just the audio feed in some cases. There was no way of inserting advertisements. We constructed a page with some ads, but no advertiser was willing to come on board because how many people would watch it anyway? At best the capability was for 200-300 simultaneous users watching, not more than that.

A company is born
The company was incorporated in July-September '97. Simon and I were the first two directors. I had to go to the UK to sign the papers. I was coming to India anyway, so I came via UK and signed the papers.

In the '96 World Cup we had the servers in the Oregon Graduate Institute [OGI]. Someone working at Sun Microsystems said that Sun would be willing to give Cricinfo a couple of servers during the World Cup. So we had two machines that had to be housed somewhere. This guy at OGA provided space and bandwidth, with little realisation what was going to happen to them. The Cricinfo servers occupied 96% of the total bandwidth available to that computer department, and all their email and communication came to a standstill because people were just hammering the servers.

So they said they would somehow maintain it till the end of the World Cup but after that we had to find somewhere new to host the server. We were looking around but there was no way of figuring out any service provider in the US who would provide it for free. This is where Mike Whitaker and Jeff Green come in. They both worked at Cricinfo for a long time, even after the Wisden-Cricinfo merger.

Jeff used to build computer boxes to sell. Mike was working for an ISP in the UK. His boss had a 2mbps line coming to his house, given gratis by the company as part of the perks of working there. He agreed to host the Cricinfo server in his house. So we quickly bought a machine from Jeff, who assembled it. A group of us pitched in. So we bought a machine with various people giving $5, $10. We only needed $1000. He built a fairly good machine and moved it to Mike's boss' home.

Thereafter, until post Wisden-Cricinfo, every server was assembled by Jeff and Mike. They would choose the hardware, they would tune it, everything was done by them. They considered each machine their baby.

Money, it's a hit
In '97 we started getting advertisement deals, primarily from India and some from UK. For the '98 Australia tour of India, ANZ Grindlays gave us money. Some connection was there with Australia. They wanted to track NRIs or something. We had them paying us Rs 100,000 for that one series sponsorship.

There was a period in '97 when ICC gave us some money, before we got any advertising money. David Richards had moved to UK to head ICC as their chief executive. When Simon moved in '95 to UK, he had built up a relationship with Richards, who used to browse scores on Cricinfo. When I subsequently talked to Richards, he said Cricinfo was one way for him to get in touch with the cricket world and know what was happening, because ICC had no communication room or information flowing into it.

That was the time where Simon was proposing that ICC take over Cricinfo and that all of us work as employees. Richards actually liked the idea but he had minimal funding and he was honest with us about this. He said, I survive on money given to me by ACB and ECB, but let me see what I can do, and he did give us some money, though on what basis I was unsure, in terms of what we were supposed to deliver. It was around £7000-8000, which came in dribbles every month. That allowed Simon to take some salary and start spending more time on Cricinfo.

Within a few months we started generating money on our own, enough to build an office and hire people. Their salaries were not very great amounts - Rs 4000 even. By the beginning of 1999, we had a fantastic deal with Intel for Rs 12 lakhs per year. It pretty much took care of all of our expenses. Handling the India office was not difficult, but UK took a longer time to build an office. Simon was still working from home. In terms of proper money, '99 was a big year for us.

The 1998 Champions Trophy
You had three people: Simon, David Richards and Jagmohan Dalmiya. On one side Simon had great respect for David Richards and he had this idea about ICC owning Cricinfo.

The video thing that he did with Mick Jagger really gave him a kick. Whenever he would go to meet people to talk about Cricinfo, he would say, just imagine a day when all first-class matches are streamed live, so you could watch Railways v Uttar Pradesh - why he chose those two teams we didn't know, but he always would say Railways v Uttar Pradesh in his British accent.

In '98 there was an executive committee meeting of ICC in Dubai. Richards got us to meet with every board's chief executive. At that point AC Muttiah was the Indian board head. Dalmiya was head of ICC. We talked to almost all of them, presenting the idea that ICC should somehow end up owning this great Cricinfo and that the revenue generated should be shared. Obviously it was a non-starter because people like the Australian Cricket Board were looking at the net seriously at that time and had their own plans.

The meeting in Dubai happened before the Wills Championship in Dhaka. We ended up covering that tournament on behalf of ICC, so again Richards was trying to push us to make it into some sort of showcase to say, look, this is what we can do. But no one was sympathetic. The most he could do was to get South Africa, headed by Ali Bacher, and Zimbabwe to be with us. These two were very sympathetic. They immediately agreed for Cricinfo to run the official websites of their boards. Questions of if they would generate income and whether or not it would be shared was not clear.

Much of the money was still generated from India. Indian advertisers were willing to advertise. We were showing those as examples, saying, look, there will be a lot of money and we can do interesting things. But Dalmiya showed no interest. He said that as much as there would be some pluses to owning Cricinfo's profits, what if it starts making losses? Then I'll be losing money and all my constituent boards will be losing money. So I don't want it.

The one last thing David Richards did was getting us all to the '98 ICC Knockout to cover it. Our travel and stay was all funded by ICC, otherwise there was no way we could have gone there. So we had two people or one covering the matches - Travis mostly, and one back-up scorer.

We were still absolutely thin or nearly zero on editorial capability, or people who could write match reports, but we did take some journalists. Among the people we hired around that time was Partab Ramchand. Anand Vasu probably came in around '99. This '98 coverage involved mostly people who were volunteers, who just flew in for this. Travis was not an employee, so he just came in and did the scorecards. Rohan Chandran was one of the other volunteers who did the match reports, and Simon and I were there.

In Dhaka, the ICC gave us excellent space, a good internet connection, and we could do ball-by-ball sitting in the stadium officially. In South Africa too, the South African cricket board had already given us access and we were doing live coverage.

In February 1999, five of us met in UK, Simon and I who were the directors, and three others who showed a lot of interest - Peter Griffiths, Alex Balfour and Dave Liverman. We discussed the future of Cricinfo and each one had their own opinions. Dave was a professor in Canada and he said that his contribution could be only as an advisor, not as an employee. Alex and Pete were both ready to join the company if we could raise money in co-operation. I was committed. The only issue now was to find the funding.