There seems to be some confusion.

Spotted on the Rails mailing list:

Re: RailsConf in London
So? Is Jim Weirich coming or not?
YES: http://europe.railsconf.org/articles/2006/05/03/announcing-railsconf-europe
NO: http://skillsmatter.com/menu/255

The answer is YES, I will be there. I was a late addition to the list and my name might not have made it into the announcement from Skills Matter.

Tim Bray wishes for individual page feeds.

Page Feeds

In his blog entry, Wikipedia, Tim Bray confesses that he has taken an interest in editing a few Wikipidea entries, but that its just impossible for him to keep it up.

He says:

I don’t have time to go check back every day or even every week, and that’s what a conscientious article minder ought to do. I totally need, for each article, a feed I can subscribe to that will summarize changes.

I have no control over Wikipedia, but I would like to point out that Ruse wiki does indeed suppply individual page RSS feeds. So, if you are interested in monitoring a particular page, you can put that page’s RSS feed into your feed reader and be notified of every single change.

For example, if you wish to monitor the home page of the Ruby garden, you would use the URL: http://wiki.rubygarden.org/Ruby/feeds/rss/HomePage.

If you use feed autodiscovery, each page should present three different feeds, one to the page (as noted above), one for changes to the wiki, and a feed to capture changes to all the wikis supported by that instance of Ruse. The ability to stay on top of changes to a wiki is an important tool in the fight agaist wiki spam.

One interesting side effect of individual page feeds is the use of message pages. On RubyGarden, I created a page named JimWeirich/Messages and subscribed to that page’s feed. Now, if you want to drop me a message on the wiki itself, just edit that page and I will be notified of the change.

The new Wiki is up for 26 hours and we got our first spam attempt.

Ruse is Now Running RubyGarden

Chad and I quietly moved Ruse into production to support the RubyGarden wiki Wednesday morning. Spam immediately stopped with the new wiki software, mainly because the spam bots don’t know how to update Ruse pages (yet).

I figured that we would get a grace period before the spam starts. Well, here we are … about 26 hours after deployment. Here’s a look at the logs:

Notice the two red postings. They were made by an anonymous user and triggered the spam filters, so they went directly into the tarpit. Non-spamming users won’t ever see that spam.

So, it looks like it is working. It will be interesting to see just how effective Ruse turns out to be.

Oh, by the way. I just checked the logs and we have served over 11,000 wiki pages so far in the day and a half that we have been running.

Today’s topic is Spam … and what we are doing about it.

Spam by the Numbers

Anyone who visits the RubyGarden wiki regularly has probably run into wiki spam. You know what I mean, defaced pages with hundreds of links to questionable web sites. All done with the goal of increasing Google page rank.

Just to give you an idea of the magnitude of this problem, make a guess on how many time during the past 7 days someone tried to deface the RubyGarden wiki with spam.

Got a number? Its probably too low.

According to the logs, we had 18,139 attacks against our wiki. In just seven days! Over the past few weeks we have been averaging between 17,000 and 20,000 attacks in a 7 day period.

That’s a lot of spam.

Fortunately most of the attacks went directly into the wiki tarpit where only other spammers saw the results. Only about 250 attacks made it to the real wiki where they needed to be cleaned by hand.

(For those who aren’t familiar with a wiki tarpit, it is a shawdow wiki behind the real wiki where spammers are directed. The spammers spend all their time updating a virtual wiki that no one, except other spammers, will ever see. The goal is to have the spammers waste their time instead of ours.)

Now the tarpit isn’t perfect. Sometimes legitimate users get sent to the tarpit instead of the real wiki. If you ever went to RubyGarden and saw spam on almost every page, you were probably in the tarpit.

But, cleaning up 250 spams instead of 18000? That is a pretty good success story.

But We Need Something Better

As good as the tarpit approach is, we still need something better. The UseMod wiki software we are using makes it painful to clean up spam. The average page needs about four clicks to despam, with a lot of hard to automate decision making in the process. See this demo for a look at what I do to clean up a UseMod wiki page. Go ahead, click now. I’ll wait for you.

That’s a lot of work. Despamming several hundred posts can take hours.

Ruse

Ruse is new wiki with built-in anti-spam features. It supports UseMod style markup, so all of the RubyGarden pages can be easily migrated into it. It has an integrated tarpit that makes despamming a page a single button press. In fact, Ruse can move all of an author’s pending posts into the tarpit with a single click. Ruse can mark edits as spam based on either content (e.g. linking to a known spam site) or IP address (coming from a known spammer).

Best of all, Ruse makes it easy to distribute the job of detecting and marking spam across the regular contributers to the wiki.

Watch this demo to see Ruse in action.

The Beta

Chad and I have setup a mirror of the RubyGarden wiki at http://rubygarden.org:3000/Ruby. This is a trial run of the software before we commit to using it. Feel free to check it out, kick the tires and beat on it a bit. Shoot, if you have a secret desire to be a spam writter, go for it, just to see what happens.

You can post anonymously, or sign up for a guest account. After a certain number of spam-free postings, guest accounts are upgraded to full member accounts.

Oh, the account “spammer” with password “spammer” is already setup if you to see how the wiki reacts to spammers.

But remember, this is a beta trial and content of the wiki will be reset before it goes “live” for real.

Documentation is a bit skimpy right now, but we are working on that too.

Enjoy.