Skip to Content

Instant Messaging: Past, Present and Future

A brief history...

Instant messaging has been around on the Internet for longer than the World Wide Web. In its earliest, purest (and, it's probably fair to say, crudest) form, it was possible to use the Unix command line tool write to output a message to another user's terminal, provided that they had previously typed mesg y (i.e. messaging yes), or indeed to directly echo or even cat the contents of a file to another terminal. Surprisingly, there was also a tool for real-time typing to the other terminal, which eventually settled on a split-screen approach. (Far more recently, this was,  one of the supposed "killer" features of Google Wave before its development was abandoned - yet it had existed in a simpler form many years before.) While the original write and talk utilities have been gradually improved so that they can talk to users on different servers - and, for example, provide security over the Secure Sockets Layer (SSL) - they were never a user-friendly tool for the non-technical user. They are still installed by default on some Unix/Linux distributions but are little used even by developers, given the huge variety of more modern, scalable technologies.

What was not provided by these early utilities was the ability to have anything but the crudest control over who could chat with you. Once you had typed mesg y, anybody on that server could contact you until you typed mesg n (i.e. messaging no). In addition, giving somebody else control over your terminal was a major security issue. Even so, the modern concept of contact lists (i.e. friends) and presence information (e.g. available, busy, offline etc) were also missing. You can still tell, although only by user name, who is logged into the same Unix/Linux server as you by typing simply who into the terminal.

The next great step in the history of instant messaging was Internet Relay Chat (IRC). This essentially provides command-line chat rooms, which can be made at least somewhat more user-friendly through graphical user interface (GUI) tools such as mIRC, as well as private messages to individuals. While it is not particularly obvious how to indicate presence, amongst the myriad other commands that are available, all of the functions that one would expect in modern instant messaging are available. It was later made available over SSL, which provides basic security from snooping. However, IRC remains susceptible to netsplits, takeover wars from hackers, denial-of-service attacks, and one is not automatically guaranteed a unique identifier or nick if it has already been used on that server by another user or if the server does not allow a nickserv (nickname registration server). Despite all these failings and its consequent decline in popularity, IRC remains popular with developer communities because of its relative simplicity, in addition to a certain retro chic.

The MSN era and beyond

Instant messaging came to the ordinary user through a myriad of mutually non-interoperable commercial protocols, each with their own Graphical User Interface (GUI) provided by the company in question and many spin-off open source replacements. The underlying technology behind these protocols was not published, but they effectively supplied what would now be called an Application Programming Interface (API), stating how developers could write tools that could communicate with their servers. One could not simply run one's own server because the underlying technology was proprietary. Many of these are still in use, for example AIM, Yahoo Instant Messaging etc, and perhaps the largest, Microsoft's MSN or later Windows Live Messenger, has only just been retired through a merger with Skype. (This has, as a side effect, removed the ability of MSN users to chat with users of Yahoo Instant Messaging, as this is not possible in Skype.) For most users, all that has changed since those days is the gradual migration to new tools such as Skype, which adds voice and video chat, and Facebook chat, which is merely convenient because of the critical mass of contacts who are already on Facebook. Similarly, Google Talk offers IM services to anybody who already has a Google account and uses GMail for web-based email. Both Facebook and Google Talk have later added audio and video chat. Together, these dominate the market because they are attached to the most widely used Internet services and are accessible to ordinary, non-technical users. In the case of Facebook and Google Talk, there is the added advantage of access via the Web without downloading any dedicated software.

Open Standards

Both Google Talk and Skype are particularly interesting because, unknown to the bulk of their users, they implement the open standard XMPP (also known as Jabber), although Skype is not fully compliant with the roster system that enables one to have contacts across different XMPP servers. The reason for this is, of course, that they only want Facebook users to chat with other Facebook users rather than enable chat with other XMPP users, which would naturally include competitors such as Google Talk that also implement the protocol. However, the competition for instant messaging does not seem to be as fierce as it was, and the competitors have formed agreements: Facebook chat is now integrated into Skype despite Facebook offering competition in audio and video chat tools from its own Web site. This may be because the free service is effectively a loss leader: it does not provide the commercial income directly, since the service is free. Instead, Skype market additional paid services such as providing Skype Out (calling landline or mobile telephones), providing users with external telephone numbers and voicemail services,  group video chat and so on; similarly, Facebook make their revenue through advertising on their site, which is attractive because of the free social networking tools, including instant messaging, audio and video chat. It appears to be in everybody's interest to cooperate to some degree.

Google stands out among the other commercial players in allowing its users to chat to other XMPP users who have accounts on different servers, either commercial, free or privately operated. One can talk to Google Talk users (or any other XMPP users) using a free account with or even run one's own server (as one could with IRC) using ejabberd or similar open source XMPP server software. However, audio and video chat is limited to users with Google accounts, providing the incentive to prefer their all-in-one, one-stop shop approach to Internet services, which is convenient for most users. The development of open source extensions to XMPP has been slow. It is still difficult to find XMPP servers that deploy Jingle, the extension for audio and video chat, which is considerably harder to do effectively than merely installing an XMPP server, which is the work of an hour or two.

While XMPP is the de facto standard for modern IM, both for open source and increasingly commercial services, it is not without criticism. It is verbose, relying on XML, which can be an issue where bandwidth is an issue. This is a small problem for IM services but a much larger one, for example, when audio and video streams are added: it does not support binary data streams natively. It is designed for a federation network run on a number of servers and its network vulnerability, while not as high as IRC, remains a structural issue. It uses massive unicasting and does not support multicasting, which is a minor efficiency issue in chat rooms but becomes much more of a problem for group audio and video streaming. It is possible to directly substitute a newer, although relatively little known protocol called PSYC, an interserver that supports XMPP and IRC natively, which alleviates most of these problems. It takes about an hour or two to set up the psyced server, about the same as a basic IRC or XMPP server. This does, however, retain the federation approach: in future iterations of the protocol, an entirely re-engineered Peer-to-Peer (P2P) approach is under development. Being an open source project of interest mostly to technical users, development has been relatively slow. This lows XMPP and IRC to interoperate seamlessly, in addition to enabling fine control over notifications to and from other systems, friendcasting, multicasting, news federation, interoperability with microblogging systems such as Twitter and so on, via programmable chatrooms.

Voice Over IP

Coming to the same market from a diametrically opposed perspective is the SIP standard for Voice Over Internet Protocol (VOIP), which began as an audio service and later developed both IM and video services in addition. This is widely used in the commercial sector: for example Vonage in the UK. There are open source varieties that can be deployed by anybody, albeit with some technical difficulty, such as Asterisk and FreeSwitch. These only cost where they connect to the Public Telephone Service Network (PTSN) that provides ordinary landline telephony, but they also enable low-cost, in-house management of telephone extensions, voicemail and related services, as well as making telephony available through computer terminals as well as telephones. One can manage distributed calling, effectively enabling call centres, using this free technology, which can be installed even on a home server. While most people would not have a particular reason to go to such effort, the entry costs to setting up complex systems have been radically reduced to the point where they would now be affordable for small organisations who can rely either on voluntary contributions of development effort or who can outsource the work cheaply.

Why is this important?

Technologies such as XMPP may not be of immediate interest to the average Internet user, either in the HE sector or more widely. However, they underlie so many of the Internet services that we may use on a daily basis that issues such as interoperability of services via open standards are worth knowing about, at the very least in order to gain an understanding of the relative difficulty of providing such services and the costs involved. Given that more and more reliance is being put on an increasingly small group of major providers of Internet services by vast numbers of ordinary users, the consequences for privacy and management of personal information are potentially immense. There is an intense debate going on about whether services taking a federated approach, relying on a network of servers, or a peer-to-peer approach, is the best way (or even a feasible way) to mitigate against these risks is relevant to many other technologies, of which instant messaging is only one: the most significant of these may be social networking. For most people, social networking is vastly more important than, for example, darknet services and/or file sharing, which currently account for the large bulk of peer-to-peer services in widespread use. Indeed, it is social networking, that typically gathers together a number of pre-existing technologies together for convenience with the core microblogging service, that best highlights the widely differing approaches to the future of the architecture of Internet services.

Dr. Radut | blog_post_tf