Whatever happened to TrackBack?

September 24, 2006 6:43 PM

Whatever happened to TrackBack?

It used to be all the rage, and it seems like less and less blogs are supporting/using it. How come?

posted by goodnewsfortheinsane to computers & internet (17 comments total)

MeFi stopped using it, I believe, because people started embedding mp3s in it. Perhaps other things could also have been embedded causing other issues?
posted by dobbs at 6:50 PM on September 24, 2006

Some possible reasons:

1) They get meshed in with comments on Wordpress, and possibly Movable Type as well (either that or they're placed under a seperate section for Trackbacks, but look just like comments). This can be incredibly annoying, as often comment threads turn into conversations. Trackbacks would then be the equivalent of people randomly walking by the conversation, spewing vaguely related nonsense and generally making it hard to follow what the other person's saying.

2) Blog posts track back to the same blog. For example, if I write a new post and link some of my older blog posts, those older blog posts suddenly get trackbacks. More noise in the system I don't need.
posted by chrominance at 6:54 PM on September 24, 2006

SPAM
posted by Good Brain at 6:56 PM on September 24, 2006

H3rb4l v!4gr4
posted by bonaldi at 6:57 PM on September 24, 2006

In general it was because TrackBacks got overrun with spam on all blogs. I still use them, but I am very careful.

chrominance: For point 2, an easy work around to this scenario is to not use the full url when linking from one post to another in your blog entries. So for example, instead of www.example.com/post1/ just use /post1/. In other words, make your links relative to the root.
posted by maxpower at 6:58 PM on September 24, 2006 [1 favorite]

I don't even remember why we stopped supporting it. Good riddance.
posted by Count Ziggurat at 7:44 PM on September 24, 2006

It was really ended up as just a way for Spammy McSpamsters to ply their trade.
posted by bshort at 7:47 PM on September 24, 2006

Yeah, spam happened to TrackBack. I can say from personal experience that of the TBs that my various weblogs received, less than 5% of it was legitimate, and the remainder was spam. And by its very nature, it's hard to combat TB spam -- for one, there's no way to make a TB-sending site authenticate or otherwise prove its non-spamming ways, nor does the entire theory of TB make such a thing desireable. And since TB was intended to allow for more than just the "I linked to you, so I'll send you a TB", all the various methods of trying to validate a TB fall apart when subjected to the alternate uses of TBs...

In the end, it wasn't worth the processor time or antispam efforts for the few times when TB gave me info I didn't have from some other method, so I turned it off on all my weblog servers, and haven't looked back once.
posted by delfuego at 8:00 PM on September 24, 2006

Just for background, TrackBack is both a technical protocol which lets an application send a notification to another web page, as well as a user-facing feature that lets your blog show when someone has linked to it. This feature was described by the MIT Tech Review in 2002 as "turn[ing] static Web pages into active hubs of related information".

TrackBack definitely has a flaw in not having authentication as part of the original specification. Just like email, it was designed for a smaller community (there were only a few thousand blogs back then) and thus it faces some of the same issues when scaling to millions of users. Baen and Mena, when creating TrackBack, may not have anticipated it (1) being deployed on tens of millions of web pages and (2) that Google would later change the web enough that links would have an explicit economic value.

Since originally being deployed for blogs, TrackBack has been adopted for an incredibly broad range of uses -- PRWeb even enabled TrackBacks on their press releases, and CNET allows TrackBacks on its stories, as do a number of other news services.

Going forward, TrackBack as a format has some technical requirements. As mentioned above, there should be some kind of authentication, and various parties have requested clarifications on internationalization, or some effort towards signing TrackBacks or allowing TrackBack messages ("pings") to include more data in an XML format.

To that end, the TrackBack protocol has been submitted as an internet standard, and a number of prominent experts with experience in creating web standards signed onto the effort to come up with a new and improved version.
posted by anildash at 8:02 PM on September 24, 2006

dobbs has it, specifically midi's of Rod Stewart
posted by Mitheral at 8:05 PM on September 24, 2006

There was no authentication, no handshake to check if a site really was talking about your content, and it operated on the trust that people would use it to create links between different sites talking about the same thing.

Its openness was its undoing, because spammers soon figured out two things:

1. It worked by merely sending a simple POST request to a known URL.

2. Once sent, a link to any URL you specify in the trackback would appear on whatever site accepted trackback (including those with very high pageranks).

Those two things meant that anyone wanting to spam high pagerank sites to get their site ranked higher at google could build a bot that scoured weblogs, discovered the trackback address for each, then pummeled them all with their links to their content.

The whole point of trackback spam isn't to drive human traffic, it's merely to trick search bots into going to spammers' sites.
posted by mathowie at 8:07 PM on September 24, 2006

Actually, in regards to mefi, when someone started sending trackback spams with images of goatse in them, I turned them off.
posted by mathowie at 8:08 PM on September 24, 2006

One inventive way of combatting trackback spam I've seen is where you request a unique token on a blog that you need to send along with the trackback for it to be accepted. The token was randomly generated via javascript (which locked out many of the spam scripts, but also anyone w/o js enabled) and was only valid for 15 minutes. Seems like a "good enough" solution for most blogs, which were targetted by the mass spammer scripts but aren't high traffic enough for a spammer to write a specific js-enabled script.

Anyways, it was powered by a wordpress plugin, but I don't remember the author or name off hand.
posted by rsanheim at 7:06 AM on September 25, 2006

It's a bad solution to a bad problem. There are much easier ways to track incoming links to a page than exposing the equivalent of a free-for-all mailbox.
posted by nixerman at 7:13 AM on September 25, 2006

Nixer, it's probably worth mentioning that TrackBack wasn't merely supposed to be incoming links, it's supposed to be an expression of a deliberate act of linking between sites. So even if nobody ever clicks on the link, the author has made an explicit connection.
posted by anildash at 9:33 AM on September 25, 2006

Yeah, to amplify what Anil mentioned in a way that elucidates what I meant by "the theory" of TrackBack -- TrackBack was never ever intended to be a mere indicator that site A linked to site B. Sure, that's one intended use, but an example other use is to allow site A to let site B know that a similar topic is being discussed without site A necessarily linking to site B in some way. (Envision me writing a long post about how badly Verizon's Bluetooth implementation sucks, and then sending a TrackBack to a wiki page about cellphones and bluetooth.) The problem here is that if site B wants to implement some filter against TB spam, the low-hanging-fruit way of doing it might be to check the sites which send TBs for links back to site B... but this means that site B would be checking for adherence to a requirement that doesn't exist.

That's all just one example of the problem that spam brought to the world of TB. I, for one, would love to see a v2 TB implementation that solves some of this... but it's by no means an easy thing to solve, since either the semantic idea of TrackBack has to change, or entire clouds of trust and distrust have to develop to guide the methods of accepting and rejecting pings.

Oh, and rsanheim, that JS token method of TB isn't really a solution to TB spam, since the back-and-forth moves that entire conversation out of the TB world. There's nothing in the TB spec that allows for a token request and reply which precedes the actual ping, and thus, there was never a way for something like that to make it into the mainstream as a way to "fix" TrackBack. And you can rest assured that it's this very fact that made it "good enough" -- because once something like that made it into the mainstream, every TB spam script would implement it.
posted by delfuego at 9:45 AM on September 25, 2006

Trackback is Dead: Are Comments Dead Too? - Tom Coates. I turned trackback off on my old blog, it wasn't worth it.
posted by randomination at 1:58 AM on September 26, 2006