Intrigue and Instant Messaging

By Jim Dutton

That’s what I told my co-founders at Activerse, Inc. back in the winter of 1996. We had formed an Internet startup to develop a cool new virtual office environment using a Java graphical front-end to Pavel Curtis’ LambdaMOO. After building the prototype and showing it around a bit, we were disappointed to find that most people’s reaction to the idea went something like, "I like the pretty pictures, but I would never actually use that." They did, however, seem to like the fact that they could see when their co-workers were online and in the virtual building. Sort of a high-tech in-out board.

Inevitably, some practical soul suggested we pare our product down to just that little feature and make it work for the entire Internet. And despite my reservations about basing an entire company around something as trivial as a buddy list, that’s exactly what we did. Thus began a four-year adventure in a world of intrigue, irony, espionage, and whimsy, which eventually became known as Instant Messaging.

1997 – Innocence Abounds

We didn’t know it at the time, but soon learned of another group of developers who had been working on a similar idea for several months already. Four high school students in Tel Aviv, Israel had built a simple little application with a cutesy three-letter name, ICQ. Arik Vardi, along with Yair Goldfinger, Sefi Vigiser and Amnon Amir called their company Mirabilis, and they released the first beta of their buddy list product in November of 1996. To our profound disappointment, it had almost all the features we had envisioned for our own product, plus a few more we hadn’t even thought of. It did, however, have (and still has) a couple of serious drawbacks that we thought we could improve upon.

Nobody, we felt, outside of an Orwell novel would want to use a number, a simple integer, as her identifier or screen name. Related to that problem is the fact that ICQ relies on a single server farm. Unlike email or DNS or the WWW, ICQ’s flat namespace resides solely on servers owned by Mirabilis, which, contrary to the Latin meaning of their company name, is not all that wonderful. Worst of all, these little boys were giving away the software, and apparently had no intention of ever making any money with it. Clearly, we could do better than that.

So we went ahead with our development plans at Activerse. Our CTO, Gordon Mohr, designed a presence protocol, dubbed whodp with a nod to the Unix utility ‘who’ which provides similar functionality. We used a hierarchical namespace and URIs for user names, which allowed whodp servers to be distributed in the same way that web servers are. The buddy list product, which we called Ding!, was finished and released in beta form by mid-summer of 1997, only eight months after the Mirabilis boys had introduced their cute little buddy list to the world. Unfortunately for us, that wasn’t fast enough.

While we carefully crafted our Java server and client code (yes, a Java client, but that’s another unpleasant story), ICQ’s user base had grown faster than any other software application in history. By mid-year, they had millions of registered users. It wasn’t quite a household name yet, but a boatload of 15-year-old boys pretending to be 19-year-old girls knew about it and used it everyday. And we were happy to cede that market to them so we could concentrate on paying, corporate customers, a position that admittedly took some of the fun out of the endeavor and one that would lead us to our first encounter with the dark side of the IM force.

In October, we received a phone call from Martin Calsyn and Lisa Dusseault at Microsoft Corporation. Microsoft, they told us, had been working on a standard protocol definition for instant messaging since March and was just about ready to submit an Internet Draft to the IETF. They asked if we would be interested in sending a technical representative to review and endorse the draft before its submission. Gordon Mohr answered the summons to Redmond and, in the end, made significant revisions to the document and became one of the three co-authors by the time it was submitted as draft-calsyn-rvp-01.txt (expired September 1998). Microsoft’s Rendezvous Protocol, RVP, was the evil empire’s first public endorsement of instant messaging and its first attempt to derail Mirabilis’ speeding train of user registrations.

1998 – Exploding Eyeballs

Despite ICQ’s grass roots, network effect, popularity, it did not have the largest user base at the beginning of 1998. America Online had noticed the IM phenomenon and converted their proprietary AOL Instant Messenger, AIM, to work on the open Internet. Every AOL user was, by definition, a registered AIM user as well, and there were roughly 10 million AOL users by this time. The rate of growth, however, heavily favored the boys from Tel Aviv, and it became clear that the size and loyalty of ICQ’s online community would soon overshadow AOL’s, even though Mirabilis had spent almost nothing on marketing and advertising their product. It also became clear that Microsoft was being left behind in this rapidly emerging market, and in typical Microsoft fashion, they decided to cure the problem by swallowing the offending company.

By this time, Mirabilis had grown to about 100 employees, mostly in Tel Aviv with a few folks in New York where the server farm was located. It had no revenue, and no plans for any. It had not yet released version 1.0 of the product (it still hasn’t!) and it was run by a group of twentysomethings with no experience in the ways of the business world. How much would you pay for a company like that? Sometime in the first part of 1998, Microsoft made them an offer they felt could not be refused. Yossi Vardi, the tough, charismatic Israeli investor and father of Mirabilis’ CEO, proved to be a force that even Microsoft couldn’t overcome.

Vardi refused Microsoft’s overtures, which he considered insulting, and punctuated his decision by bringing the deal to the last company on Earth Bill Gates wanted to own ICQ. In June of 1998, a year and half after four high school students invented a new way for people to communicate on the Internet, AOL paid $287 million for ICQ and its 10 million eyeballs. Reportedly, each of the four founders received somewhere in the neighborhood of $80 million plus performance incentives. And this was a cash deal – no funny money stock options or warrants here. Microsoft acquired a small IM upstart called Flash Communications and immediately announced they were planning to release MSN Messenger, their first entry into the IM market. But it was too little, too late. By the end of the year, there were more subscribers of ICQ than of AOL, and ICQ is still the number one download at Download.com with over 100 million downloads to date and another million every week.

1999 – Clash of Titans

Now that the IM market was completely legit, consolidation and reorganization among the major players began to heat up. At Activerse, we hired an investment banker to field inquiries from companies like AT&T, Novell, and a Boston-based Internet conglomerate we had never heard of before called CMGI. In April of 1999, we became a wholly-owned subsidiary of CMGI, one of 30 some-odd companies fully or partially owned by the holding company and venture fund. With only the best intentions I’m sure, CMGI proceeded to make several unwise moves that would eventually kill the Ding! IM technology. First, Activerse was combined with three other unrelated companies to form CMGI Solutions in the Fall of ‘99. Next, CMGI acquired Tribal Voice, a west coast IM company and transferred all IM development to the new subsidiary. Finally, Tribal Voice was merged with yet another operating company unfortunately named CMGion. As Gordon Mohr said, "CMGI has more stomachs than a cow and it feels like we’ve been digested in all of them."

In the mean time, AOL has been battling mightily to control their dominant position in the IM market. When Microsoft finally released their MSN Messenger, which incidentally does not use their proposed RVP standard protocol or anything like it, the new product included the desirable ability to inter-operate with AIM. AOL immediately changed their servers so that the Microsoft IM client would not work. Microsoft retaliated by releasing a "fix" for the problem, and the two behemoths continued in this manner, slugging it out at the expense of users and to the delight of journalists for several weeks. Finally, AOL took advantage of a buffer overrun bug in their own client to ensure that MS clients could not see AIM buddies in their lists. The ironies in that very public exchange are too numerous to mention.

In November of 1999, Microsoft finally gave up and released a version of MSN Messenger that did not even attempt to inter-operate with AIM. AOL continued to vigorously block other companies, including AT&T, Tribal Voice, and Odigo, who tried to merge the networks. Interestingly, it is that central server architecture of ICQ and AIM, the flaw that we noticed back in 1996, that allows AOL to so tightly control access to their networks and thus thwart all attempts at inter-operability. Apparently, AOL still values those eyeballs very highly, despite the fact that they haven’t figured out yet how to monetize them.

2000 – The Feds Step In

Today, the IETF working group investigating IM and presence protocols has all but given up. Most of the original members of the group from Microsoft, Lotus, AT&T, Fujitsu, and Activerse have moved on to other pursuits. In June of this year, Tribal Voice, CMGI, Microsoft and others asked the FTC to examine AOL’s dominance of the IM market in light of their intended mega-merger with Time Warner. I personally doubt the government will find any relevant connection between the two things, but then the very fact that that little in-out board feature we began working on nearly four years ago is now being discussed in relation to a hundred-billion-dollar merger is unfathomable to me.

Hey guys. It’s only a buddy list.