Weblogs.com

» see weblogs.com as it lived, on the wayback machine

~ the goodbye ~

I have never typed this address into a browser and neither did almost anyone else who relied on it every day. Weblogs.com was a checkbox in your publishing settings, and usually not even that, because most blog software shipped with it already filled in. Its entire job was to be contacted by a program on your behalf, at a moment when you were doing something else. There are graves on this site for products people loved. This is a grave for a piece of plumbing that a whole era stood on top of without ever looking down.

UserLand Software put it up in late 1999, and at the start it was not a ping server at all. It was a crawler you registered your site with, and it went out and checked whether anything had changed, and it published a list of the weblogs that had updated most recently. In 1999 that list was itself a destination. There were few enough blogs that a page showing which ones had just posted was a reasonable way to read the web.

In October 2001 the whole thing was turned inside out. Instead of Weblogs.com going to check on you, your software would tell it: an XML-RPC call named weblogUpdates.ping, fired the moment you hit publish. The server took that ping and appended your URL and a timestamp to changes.xml. That file, refreshed constantly and free for anyone to fetch, became the single feed that told the outside world what had just happened on the blogs.

Everything downstream ran on it. Technorati read changes.xml, and so did Feedster and every other blog search product of the period, because crawling millions of sites on the chance one had updated was not affordable and this file gave you the answer for free. Dave Winer sold the ping server, along with audio.weblogs.com, to VeriSign in October 2005 for $2.3 million. VeriSign handed it off to a private investor group in May 2009. Then it just kept answering, ownerless in any way visible from outside, until it stopped.

~ the chronicle ~

~ what it was for ~

The problem it solved is one the web has never solved well since. A crawler has no way to know that a page changed except by fetching it again, which is enormously wasteful when most pages have not changed. The ping inverts that. The publisher, who knows exactly when something changed, spends one cheap call announcing it, and every interested party gets told for nothing.

It worked because it was small enough that nobody fought over it. One file and one method name, with no accounts to create. That kept the barrier low enough that every blogging tool of the era simply defaulted to it, and defaults are how infrastructure actually gets adopted. The design was also completely open to abuse, and by the mid-2000s the splog operators had noticed, which is a large part of why blog search never got good.

~ the day the blogs went dark ~

Weblogs.com had a second life nobody remembers as clearly, which is that it also gave free hosting to a lot of early bloggers. In June 2004 that side of it came down during a service transition, and the people whose writing lived there discovered it was gone with no warning and no export. Winer said publicly it would take a couple of weeks to sort out and it came back faster than that, with Rogers Cadenhead doing much of the work to get the archives standing again.

The episode became one of the founding arguments about hosting your own writing, and it is worth being fair about it, because the hosting was free and it did come back. What the episode demonstrated was not malice but arithmetic. A free service maintained as a side project has exactly one point of failure, and everyone using it had quietly accepted that risk without ever being asked to think about it.

~ the next chapter ~

It went out the way plumbing goes out. The last capture the Internet Archive holds with a working response is dated 13 April 2019, and by May somebody had opened a ticket on WordPress.org asking for Weblogs.com to be pulled from Ping-o-matic because it no longer answered. No announcement, and no owner left to write one. The service whose only purpose was to tell the world when something had changed managed to change permanently without telling anyone.

By then it had been outlived by nothing much. Technorati had stopped being a blog search engine years earlier and Feedster was long gone, and the reason to know which blogs had just updated had been absorbed into feeds and then into timelines that decide for you. What remains is the idea itself, which keeps getting rebuilt: WebSub, and the webhook sitting in every modern integration, are the same trade Weblogs.com made in October 2001. Let the thing that knows do the telling.

@}-->--

              )
             (_)
            .-'-.
            |   |
            |   |
            |   |
            |   |
          __|   |__   .-.
       .-'  |   |  `-:   :
      :     `---'     :-'
       `-._       _.-'
           '""""""

« previous  ·  random  ·  next »