First, this has happened before, several times. Multiple accounts are pretty much a design feature fo K5/Scoop. I've got them, I'm pretty sure Rusty and Inoshiro do as well ;-) They're a topic directly addressed in the moderation system background docs. There's not much that can be directly done to prevent this. If this incident is shattering your faith in K5's moderation system, this is probably a good thing. The system's designed to be useful, not perfect. By me. Intentionally. Your expectations exceed the designer's by a long shot.
Second, some perspective. This isn't bringing K5 to a screeching halt. Hell, the site's own design is doing it more harm than this Archer clown (though we're addressing that as well).
As far as fixes. The social solutions are storngly encouraged: moderate more comments. There are limits to this, so while it's encouraged, I don't expect to see major changes in behavior. If trusted users decide that retaliating on the Archer family of accounts is something they want to do...well, Scoop/K5 doesn't handle that sort of abuse very well either. And moderation is transparent -- it's possible to see who's moderated specific comments. Collaboration or chained accounts tend to show up pretty rapidly, Archer was pretty careless -- two accounts with similar names, another pointing to the same website, and strongly suspicious activities with two additional accounts.
On the administrative side, there appears to have been abuse and intent to abuse with these accounts, I'd recommend simply killing them to Rusty and Inoshiro. Play with the system. Have fun with it. Don't abuse it. You lose your rights.
The other "fix" I'd suggest is that a set of useful system queries start being generated such that this type of activity becomes more readily evident. Fact is that eyeballing activity by these accounts is largely sufficient to establish a connection and malfeasance. While there are all sorts of high-end statistical tools devoted to detecting these kinds of patterns, simple frequency analysis will get you a long ways. Correlations will take you a few steps further. Cluster analysis beyond that. I've got to start building an analytic suite up against my Scoop installation.
Point is that any really earth-shattering attacks against K5/Scoop are going to be really blindingly obvious. The approach is also going to be pretty straightforward: close down the accounts involved, if there are new accounts being created to effect the attack, disable new account creation for the duration, or until a better authentication system can be implemented. Anything more subtle is likely to be lost in the noise.
No, I don't see metamoderation as necessary. Analysis of existing K5 moderation data is going to be far more effective in producing information and documentation of abuse. Email to admins over potentially abusive activities is probably as good a notification system as any, though something more detailed might be ginned up over time.
Rating affinity groups -- explicitly or implicitly specified -- might be of interest. In general though I think this is only going to be spectacularly effective where moderation is fairly rich -- articles and some top-level posts. Not leafnode comments in editorial comments.
Trust metrics, or other ways of determining weightings for users' moderations may also be effective. Though typically moderation distribution is fairly thin. Hmmm...actually better for top-level comments than I'd have thought, 8, 9, 10 mods here. Thread comments tend toward 2 to 4 moderations. In most cases, weighting isn't going to make much of a difference, particularly if we assume that the majority of users will have default weights. Currently the number of "trusted" users tends about 1% of all registered users, and "untrusted" is even lower -- the rest of y'all 98% of everyone out there is average. That's an awful lot of design work for a 1% edge condition. Better to have effective policing tools, you ask me.
While there have been some complaints about the K5 moderation system, they appear to be edge phenomena. By and large, most comments are rated about right. Far more so than at a site that's been an inspiration and friend (if only in how not to do things.
Or do you disagree?
Karsten M. Self
SCO -- backgrounder on Caldera/SCO vs IBM
Support the EFF!!
There is no K5 cabal.