Chat. It's a continuous annoyance for me - there are almost as many different chat networks as I have people to chat with. To keep up with friends, colleagues, business contacts, family, etc I need to have a plethora of applications and browser tabs open and online. Often I simply refuse to register on yet another chat network, though this attitude do put restrictions on my social and professional network.

I'm planning to write three posts here, this first one as a historic overview and touching the current siloization problem, then one describing what features and characteristics I would like to see in a perfect chat system, and at last I will try to get an overview of what networks/applications there are out there and/or under construction, and which ones of those may look promising.

History

Plain Old Telephone

It may be a stretch to define the Plain Old Telephone as a "chat network" - it's voice-only, and one of the characteristics of a modern chat system is that one can deliver lines of text, instantly. Then again, traditionally "chat" was never about exchanging written messages. I'd even go so far as to say ... ordinary telephony is still the most significant chat network as of today!

I was born after Internet - but for all practical purposes, some fourty years ago the plain old telephone system was the only chat network of any significance. It was pretty much one network - one could connect with almost anyone, everywhere. Almost anyone not living in a cave would be connected to The network. The network had a very big downside, it was expensive. In some cities/towns in some countries there would be free local calls - but international calls and inter-continental calls would be very expensive and even ludicrously expensive. On the positive side, I've learned to always be quick with my phone calls.

Back in my childhood, mobile phones was suitcase-sized, something only the self-declared "important business people" would install in their cars, at a great expense. Today almost everyone carries around a POTS-connection in their pocket (I can count the people I know who refuse to take part in this network on one hand), so it is still the most established chat network.

The plain old telephone can give a relatively decent quality on speech, but (with the possible exception of "low latency") almost whatever metric one would use to measure quality, it's possible to find better chat services nowadays - and while it still usually costs a decent chunk of money to use the telephone, particularly at long-distance calls, the alternatives are almost always for free. The reason why the telephone network is still going strong is all due to the network effect!

Emails

Emails may not be recognized as a "chat system", the idea was more to replace the paper mail rather than the telephone chats - but while paper mail costs money and can take days to deliver, email is for free and usually takes a fraction of a minute, hence email conversations can have a lot more "chatty" style than paper mails. It's not abnormal with tens of short emails sent forth and back within a short time frame, that is a "chat" for all practical purposes.

If the POTS-network is both the eldest and most widely-distributed network, then email is probably on a good second position. The first emails was sent in the early 1970s, but it took some 25 years until the standard got adopted by the masses. There was competing standards out there, personally I got connected to Fidonet several years before I got my first working email address - but by today there is only one email standard and most people have a working email address.

There has been a decline; nowadays some people don't use email at all, and we've had a big centralization - only big actors like Google are able to cope with the spam problem. The original standard was written in a time where nobody bothered thinking about security; the network was still so small that every actor on the network could be trusted. Earlier one could assume the email would be delivered and read - today the reliability has become poor due to all the phishing and spamming. Clear-text emails is still the norm, even though we've had standards in place both for encryption and digital signatures since the early 90s. It's really trivial to send emails with a fake From-address. From a technical point of view, all those issues could be solved relatively easily, but getting people to agree on how to do it and to actually change such an old standard is very difficult.

SMS / mobile phone text messages

With mobile phones and particularly GSM getting traction, SMS has also become an established chat network.

SMS is not particularly good at anything - just to mention some few problems, the messages typically costs money to send, the standard has no protection against spoofing of the sender ID, it doesn't offer good protection against spamming, it's not even a particularly reliable standard when messages are passed across different telecom providers and country borders, and one is typically locked to a spesific phone and simcard when receiving and sending the messages, etc. Still, it's on a pretty strong third place behind plain old voice telephony and emails - again, the network effect is extremely important. The thing with SMS'es is that, as virtually everyone switched from land telephony to cellphones, they also got enrolled into the SMS chat network. SMS is a big thing due to the network effect!

BBS chat, unix talk and others

Not chat networks as such, but we had some precursor for line-based chats - it was typically character-based chats, possible only between users logged into the same machine. I remember it was such a thrill to be able to log into a BBS and have a real-time text chat with a stranger.

One such BBS chat system developed in August 1988 eventually got connected with other servers over the Internet, and became the Inter Relay Chat.

IRC

In November 1988, IRC had spread across the Internet and in the middle of 1989, there were some 40 servers worldwide. Relatively to the total Internet population it's peak was in the mid-90s, in terms of absolute number of users the peak was around 2003, with around a million users worldwide - later it declined and by today there are mostly nerds and old farts still using IRC.

IRC is a novel thing because it's indeed a public standard - just like the email. Anyone can create their own IRC server or IRC client, and there exists lots and lots of different software out there. As with the email, one problem with IRC is that it's difficult for established standards to evolve, and the standard does have serious flaws. One of the problems is that the standard itself offers no way to register ownership of usernames or channels, another problem is that anyone running a server efficiently has super user access over the whole network and hence can lay claim on usernames and channels. While there are lots of other reasons why the internet population moved on and abandoned IRC, those two combined has probably been the biggest reason for IRC to lose it's network effect. It was relatively quickly realized that one has to be restrictive on whom can connect to the IRC network. In 1990 there was the first fork in the network, eris.berkeley.edu allowed anyone to connect to the network and this caused a lot of problems. The very most of the IRC network ended up as EFNet (Eris-Free Network), the eris server was eventually shut down, and again there was only one IRC network.

In 1992 came the next network fork. Undernet was initially just meant to be a test network, but eventually it grew into an alternative IRC network and was the first network to introduce services for registering channels. Undernet never grew big enough to seriously undermine the network effect of EFNet, but still it did contribute to the fragmentation. Next was the DALNet fork in 1994, again DALNet was introducing new functionality, namely improved channel registration features, possibility to register usernames, allowing longer usernames and functionality to store messages for people who are offline. As with Undernet, DALNet never grew into something big, it was still EFNet that was the Inter Relay Chat Network.

Then came the Great Split in 1996, with EFNet and IRCNet splitting over disagreements on how the standard should evolve. Today most IRC clients has good support for being connected to multiple IRC networks simultaneously, but back then ... no, common folks had to chose one or the other. With the network effect disrupted, IRC lost it's competive edge.

ICQ and the age of chat siloization

In 1996 ICQ was born, and a new era was born - an era of chat siloization. Most of my friends at IRC registered with ICQ and started using that service.

I never understood it; people were abandoning the open, standardized network running on open-source software for a commercial centralized propretary service; people abandoning "group chats" (ICQ was only good for one-to-one-communication) and abandoning usernames to serially given user ID numbers. Anyway, apparently ICQ did something right - in 2001 there was around one million users of IRC, versus 100 millions of ICQ accounts.

Later came AIM, MSN, Facebook, Twitter, Skype, Slack and various other chat/instant messaging services, most of them being commercial centralized propretary silo services ... and the trend has continued, hence today we are in the situation where plain old telephony and SMS are the only really significant chat networks. I'm not going into the details in this post; the thing that was unique with ICQ is that it was the first significant commercial actor to the "online chat" market; it started the trend of commercialization and siloization.

Jabber and XMPP

Jabber.org was an exception; much like IRC it started out as open software and was eventually standardized as XMPP. Google Talk was based on XMPP and still today it probably offers some interoperability with XMPP-based clients. There are also various other commercial chat operators offering some kind of XMPP support. The X in XMPP stands for "Extensible", the point being that the standard should be allowed to evolve faster.

Particularly google talk has had a relatively big user base, but XMPP has (relatively spoken) never become as big as IRC.

There has been other attempts on creating a new and better standard for chat, but so far none of them have become notable - ref this xkcd strip:

Security

After Snowden it has become obvious that unless one wants to be snooped on by the US government it's pretty important to use strong end-to-end-encryption and avoid centralized communication services. I believe strong end-to-end encryption is pretty much important, and none of the solutions presented in this historical overview offers anything of that sort out of the box. This is a topic that I'll dive more into in the next posts.

If chat services (other than plain old telephony and SMS) is fragmentated, the situation is even more dire if one wants privacy. Petter Reinholdsen just wrote this blog entry about that.