By Invitation Only

Saturday 19th Feb 2011

We know spam. It’s everywhere. It’s miserable, nasty, shitty stuff. All sorts of filters and active processes are in use to get rid of it, but still it grows. Email is pretty much a joke now, looking at the messages I get to one of my oldest addresses, about 5% of them are real emails to me. The rest, a horror show of appalling grammar, worse spelling and the kind of offers that could only seem plausible to the criminally gullible or politicians. And just lately there’s a rash of it on Twitter — it seems every tweet I write (quite a lot I admit) seems to garner at least two spam @-replies. It boils my blood.

But I have an idea. It might work on more public networks, but mainly I’m thinking about it for the kind of private services like Twitter, i.e., the ones that aren’t utilities, with their public service obligations. It’s simple enough. Go invitation-only.

But aren’t there invitation-only sites out there already? And don’t they have spam? Yes, yes there are, and they do. That’s why my idea has a something else, a little thing that exists in real-world social networks, but is apparently absent in online ones, for various reasons:


Yup. When you invite someone to a service, you’re responsible for them, and to a lesser extent for those that they invite, and so on. You can invite more people as long as the people you’ve invited aren’t spammers. So say we’re dealing with a Twitter-style service. Imagine you’re an utter bastard and you’ve invited all your Charlene449 and your HotChick9221 types and they’re busily pumping out messages about Great Deals On iPads and Super Low Pharma Prices and inviting yet more spam accounts themselves and soon you’ve got a bloody great outbreak of spam. Oh dear. But some of those accounts are going to get reported for spam. Soon, there’ll be enough reports for automated systems to take action (or maybe you’d have human intervention, I don’t know). So far, so normal. But this is where my idea comes in.

If you invited a spammer, you don’t get any more invitations to give out for a while. If you invited more than one spammer, you don’t get any more invitations forever. What’s more, if you invited more than one spammer, none of your invitees get to invite anyone. If more than some proportion of your invitees are spammers, your account is closed, as are all those you invited. If you’re a spammer, your account is closed, as is that of everyone you invited.

And there’s more! I was thinking of strategies to get around it and get outbreaks of spam from time to time, and that would be to create accounts that themselves don’t spam or invite spammers, but invite accounts that do invite spammers — so you have your little node of disruption in your network, staying just far enough away from the damage to presumably stay undetected. Well, I’m thinking that would also be fairly easy to detect over time, if a branch of the network is persistently affected by spam, you’re going to look at ancestors of that branch, and check their behaviour. This would probably require some human intervention (but the alerts could be automated) but would work.

Now, ideas are cheap and implementations aren’t. Balancing the rules to get a system up and running would be quite a challenge, but this is something that game designers do for a living — this is effectively a combat system. Could something like this be imposed on an existing system? Would you want to? Something like this might be very good at keeping spam down, but it might also be prone to DOS attacks on other users. That’d would require human intervention to detect, arbitrate and resolve (which could get expensive), but for specific social networks, like Ffffound, Dribbble and the like, it might just work.

So, community types, would it work?