But will it scale…?
I’ve had a theory for a while now that the Internet does not scale. Naturally, that doesn’t mean that the Net *itself* - its physical infrastructure - does not scale, because obviously we had a few thousand people using it 20 years ago and some hundreds of millions using it now. But that Internet *applications* do not scale, particularly those relating to community. Over and over again in the history of the Net we see that the same technologies that are wonderful when new and not widely used become unusable later on: there is a threshold above which they are overrun with abuse and bad behavior. Email and blogs are good examples: email is nearly broken by spam, and blogs above a certain threshold of readership (the number I’ve seen suggested is 10,000) also have terrible spam problems. The blog I use for net.wars every Friday (www.pelicancrossing.net/netwars) gets probably 100 spam messages a week (and, sadly, hardly any real ones). Usenet was another. Moderated fora can, of course, keep the abuse from being visible to the public, but at the expense of wearing out the moderator. The same is true, for different reasons, of IT projects - small, local projects can be more carefully defined and more likely to succeed than giant ones - build bottom up instead of top down, but this is completely contrary to the way government works in the UK. (And that in fact may be the most fundamental difference between UK and US government - the US actually does build from the bottom up - local legislatures and town halls have their own fundraising abilities, states have states’ rights that cannot be abrogated by federal government, and indeed the whole structure of the electoral college was about people voting for electors they knew and trusted instead of candidates they didn’t; whereas the UK began with a king and then replaced him with an elected Parliamentary dictatorship.)
This piece and comments talk about this in relation to Second Life; Linden Labs apparently said it thinks the world can scale infinitely. This seems doubtful - though at least the inherent limitations of each server processor mean that you can’t assemble too many people in a single place at a time. (Hard to plan a revolution within the confines of SL!) The game worlds all are grappling with the problem of abuse in the sense of farming and other exploits that allow people to cheat (buy a level 70 warrier account on eBay, pay someone real money for their magic sword, etc).
As UK government seeks to build really large systems, this is a principle it seems to me worth bearing in mind.
wg
September 6th, 2007 at 3:14 pm
Hmm. It’s certainly the type of against-the-grain thinking that makes me, well, think. Online role playing games scaled okay–but on the other hand, they have dedicated IT staff policing the environments. Cheating is not an IT issue, there, it’s a behavioural issue. This weblog has received less than 2,000 spam comments in 9 months, and most of them have been caught by Akismet–maybe Pelican Crossing should try it. (But don’t get me started on the abuse of our wiki…)
Email had a long head start in trying to figure out how to deal with abuse, but I would argue that they’ve come a long way. My two web-based email accounts perform scarily well in terms of keeping spam out of my inbox. The corporate account I use lets about 1 in a week–just to keep me on my toes, I think.
What (I think) they all have in common is dedicated resources that have accepted responsibility for dealing with the issue. I think some scalability issues may come down to that. Your opinion?
September 10th, 2007 at 1:34 pm
Your email may be working well, but a lot of people’s isn’t. And in any case I would argue that a system that is requiring people to throw away (one way or another) 92 percent of the email traffic traversing the network is in fact broken.
(And if I were getting that little spam through I’d be worrying about false positives…)
I think you’re right that it comes down to dedicated resources (moderators, spam filters, etc). But that means that every IT project has to allow for the potential for abuse that comes with scale, particular systems that are going to involve public interaction. There are a lot more people spraying crap around the Internet than graffiti in real life, and the consequences are more severe than the place just not looking “nice”.
As for game worlds, yes, there are people around to monitor behavior. But griefing is still a big issue. All very high school. (Gail Williams, the conference manager of the WELL, once said to me that she thought a lot of online behavior was working out unresolved “high school angst”.)
wg