Remosi’s Rants

Here’s an interesting thing I’ve noticed: Most of the influential people in the IT industry have had backgrounds in either muds or IRC. The important thing is to get people involved in a community. This (IMHO) has big impacts for how people should be looking at getting people involved in projects and communities.

If you want to get involved in an open source project, the best thing to do is to find out what IRC channel(s) the authors hang out in and listen to what they have to say. Often you’ll hear about little things that need a bit of work here and there. People really like it when you answer all those commonly asked questions for them. Write up FAQ’s and Documentation if you can’t/won’t/don’t like to code.

So now onto Google. Well, they are at a clear disadvantage at the moment, MSN has the userbase mostly all tied up, everyone has a MSN account, mostly because Windows practically forces one apon you when you use the machine. So what should google be doing? They should be emphasising the users they do have. I’ve written a conference bot which lets multiple people join a public chat room and talk to each other. There have been quite a few people that have turned up and said “I added this conference bot coz I don’t know anyone else that has google talk”. They’ve then found a few people to talk to from chatting in the channel. I’ve met quite a few interesting people with all kinds of various interests from the bot. Surprisingly many are Chinese, I suspect because IRC being the only real public chat system around doesn’t support CJK languages at all well so they’ve not had anything at all like it.

It’s a pity that IRC is so overstretched and nothing has stepped up to take it’s place. I think it’s a very important part of the culture of the Internet.

Today I was thinking about IRC, and how it’s dead (prompted by a good friend of mine). IRC doesn’t know it’s dead yet, it’s wondering around still thinking it’s alive. But it’s dead inside. Where it matters. IRC is archaic. It predates ISO-10646 (Unicode) which is annoying. But the fact that it predates ISO-8859 (latin-#) is even worse. It’s bizarre case mapping rules, that are even incorrectly specified in the RFC however make it ripe for confusion! It’s centralised in a spanning tree of servers that can be easily attacked. It’s fragmented into little fiefdoms called “Networks”, and has more political intrigue than a byzantine brothel. IRC is extremely vunerable to denial of service attacks. Because of IRC’s centralised nature denial of service attacks on any part of IRC causes major disruption to the entire network. There is only one reason that IRC still exists today and that is because there is no other medium on the Internet that allows for realtime chat amongst multiple participants that is keyed off a name. Some Instant Messanging networks allow for multiuser chats, but the only way to join the chatroom is to be invited by a current participant, as opposed to IRC’s /join #channelname. The fact that users mistakenly confuse “IRC” the protocol and “mIRC” the de facto windows client, shows that it’s not IRC itself that they are attached to, but a well written client that provides a service that they cannot get from any other application. I’d replace IRC with a well written (preferably crossplatform) client that doesn’t use spanning trees, and doesn’t have the concept of a “network”. I’d model it more on the idea of a “URL” or “email address”. You can DoS any part of the email infrastructure. You can DoS peoples mail servers and stop them getting mail. You can DoS peoples mailing list servers and stop everyone on the list getting mail. There are a few innocent casualtys when this occurs, but nothing compared to the number you get on IRC networks. I’d throw in user registration for good measure. People that know me well can probably guess where I’m going with this. My proposal is to replace IRC with Jabber. It has all the requirements, it can support multiuser chat with people only knowing the username, support for newer standards such as UTF8, XHTML, etc, and has a good community behind it. So perhaps someone needs to write a good client that supports Jabber, some kind of good scripting language and is “channel” orientated, not roster orientated. Maybe even in Java. You could use Rhino for scripting.

Big Fractal Tangle: irc street corners The most interesting bots on IRC I think are the infobots, or the infobot alikes. What they do is sit in a channel and look for lines of the form “a is b”, when they see a line like this they save it into a database somewhere. Then, later, you can ask “what is a?” and the bot will reply “a is b”. By leaving it sitting quietly on a channel it can quickly learn a lot of valuable common information that the channel has, and can become a virtual wiki like system itself where anyone can add an “answer” to a question. I’ve been wondering (but never got around to trying), implementing an infobot that import/export’s to RDF. For instance asking “what is Timothy’s email address” the bot could look it up in a FOAF file and return the answer.

Bletch, Last night I decided I wanted to reorganise the display of /who output and /stats l in xchat to be more readable. A simple task yes? Well, no. Xchat allows for hooking the numerics directly before xchat gets it’s sticky paws on them, however when you do that, Xchat won’t recieve them and then it won’t update it’s nicklist for things like /away status. Sigh, ok, so how are things displayed? It uses the event “Server Text”, however “Server Text” doesn’t tell you what kind of server text you are seeing so you can’t hook just /who output. So I wrote a patch to XChat that made numerics be sent using a “Server Numeric” event, which provides which numeric it is. Still, it’s not very pretty, but it works.