The hunt for a silver bullet

18 October 2005

The powers that be at Google must be wondering why they ever bought Pyra Labs, the people behind Blogger. The relentless expansionism of Google has made the company the number-one target for angry bloggers who want to know why their ego feed searches are full of splog entries.

Chris Pirillo came up with some suggestions to Google. Unfortunately, as with email spam, we are getting to the situation where the cure could be worse than the disease, and have little to no effect on splogs themselves (other than forcing them to alter strategies a little). People tend to forget that spammers present a moving target.

I can't help but see problems with most, if not all, of Pirillo's suggestions simply because spammers do adapt. I've paraphrased the suggestions for brevity:

1) Employ a blog spammer. Maybe Google already does. Oh, you mean knowingly employ a blog spammer. And if you do get one, how do you know you've got a good one? Or should that be bad one?

2) Probationary period: only allow people with a track record to create more blogs. Good plan, if it were not for the case that, apparently, the spammers have been using lots of accounts to create blogs, not a few accounts spawning lots of blogs.

3) Sponsor a blogger: you need a reference to create a blog. And if a new blog goes spammy, revoke both it and the referee account. This is something that will run and run, in court. As with Ebay, Blogger account hijacking will become the new sport for keen phishermen (and women). Why use up your real accounts when you can phish one out of Little Jimmy and his blog on Star Wars puppets?

4) Flag splogs from the toolbar. Apparently now done. But that does not deal with the problems caused by false flagging. Who is to stop sploggers from flagging. Don't forget, these guys are operating with a very large number of accounts.

5) Take every experience seriously. Can't argue with that one in principle but do we know that not taking it seriously is the problem? How long does it take to alter what is already a large code base? Google engineers might be sitting on their fat arses, maybe their engineering is undermined by crypto-blog spammers (see point 1), or maybe they're just a bit overwhelmed.

6) Track bad neighbourhoods (ie link farms) and penalise Blogspot sites that start linking to them. Interesting but we could start seeing a new trend in blogbowling as well as Googlebowling. Also, it already seems that sploggers are building rings around their bad neighbourhoods and, if this were to become policy, the sploggers would simply make sure there was an insulating layer between new Blogspot splogs and the link farms they are really trying to support. For a splogger there is little point in linking straight from a new Blogspot splog to a link farm anyway, as the new splog would have sod all in the way of PageRank.

7) Reward flaggers. Flagging is it's own reward and maybe should stay that way. How many T-shirts do sploggers need (see point 4)?

8) Audit randomly with a "how's it going" question once in a while. Actually, that's not a bad one. But sploggers tend to work by probabilities rather than saying: "Oh dang! They've changed the script, knocking out 10 per cent of my bots' attempts. Better hang up the old splogging boots." Nope, they just try and get more accounts, or write scripts to account for the change.

9) Get the AdWords team to help flag 'hot' keywords. And then do what exactly?

10) No more dashes in blog names, cos sploggers like dashes. Well, there are plenty more characters in the ASCII set I'm sure they'd like to try.


Okay, so... you didn't like my suggestions... write *10* of your own. ;)

I'll see what I can do. Which is probably not a lot.