BurningSmell

Michael Jackson falls, internet stumbles

Michael Jackson’s sad death on June 25th, 2009 provided a morbidly fascinating glimpse of the way major news events unfold across the internet.

To illustrate the point, I pulled together a visual timeline (click image thumbnail, or PDF, 5.3MB) relating to the passing of the Prince of Pop: the information in the timeline came from SEOmoz.org and the Reuters website.

In essence, it took 49 minutes from the time of the fateful 911 call from Jackson’s home at 7:21pm (all times quoted here are GMT) to the first sniff of the story on x17online.com, a small entertainment site. Ten minutes later, a much more prominent site called tmz.com had what was then a story about MJ’s cardiac arrest and hospitalisation.

TMZ’s RSS feed then pumped the news out to tens of thousands of newsreaders, and by 9:12pm the story was running at unconfirmed level on Wikipedia, making it the first major news, reference or social media site to carry information about Jackson’s collapse. (Wikipedia never sleeps, it seems: I remember when the destructive tsunami hit the Indian Ocean in December 2004, Wikipedia had the news on the front page before almost anyone else, driving home the advantages of many amateur eyes and ears over a few professional ones.)

A few minutes after Wikipedia started to run the story of Jackson’s cardiac arrest and hospitalisation, he was pronounced dead by the hospital and this sad final news was published on the TMZ website immediately afterwards.

At peak, Twitter hit 5,000 Michael Jackson-related tweets per minute.

The first squeak from Twitter came at 9:30pm, as @CNNbrk pushed out news of MJ’s hospitalisation (not his death announcement) to its two million followers. Digg had the story just after 10pm. Twitter was running about 2,000 mentions a minute of ‘Michael Jackson’ around 10:30pm, and its notoriously wobbly infrastructure started to go weak at the knees a few minutes later. At peak, Twitter hit 5,000 Michael Jackson-related tweets per minute.

Not that Twitter was the only internet landmark to topple. Google started to experience a trickle of michael jackson queries around 9pm (once again, remember we’re talking GMT here) and by 10pm the volume of queries was so high that the Googleplex decided that it was under some kind of automated attack: searchers started to get polite ‘go away’ messages instead of news results. Google had nothing on the Michael Jackson story on Google News until 10:40pm, and the SERPs produced nothing until nearly 11pm.

… [Google] decided that it was under some kind of automated attack: searchers started to get polite ‘go away’ messages instead of news results.

So, with Twitter wobbling and Google creaking, how did the rest of the internet do? The Wall Street Journal reports that the AOL, ABC, CBS and LA Times news websites were all overloaded: download times for pages nearly doubled, and site availability dropped from near 100% to 86% as web servers and content management systems stumbled.

As news of of Michael Jackson’s death sunk in, his fans went to work, putting up memorial sites. Within an hour of his death being confirmed, more than 500 MJ memorial groups popped up on Facebook, some boasting more than 50,000 members. (As of June 28, 2009, the Facebook group R.I.P. Michael Jackson (We Miss You) had nearly 1.3 million fans.)

The real-time web is wobbly

Events like this tragic one remind me of a root-level difference between internet and traditional newscasting technologies. With the internet, each additional reader, searcher or radio-listener costs the publisher that little bit more in CPU cycles, bandwidth and hard cash; with traditional broadcasting, there’s no penalty for increased demand.

Beyond that, it highlights the challenge of ‘the real-time web’. As SEOmoz points out, a site like Twitter is a microcosm of a new kind of real-time web, one in which indexing and searching become pointless unless these features operate in near-real time too.

As for Google, SEOmoz comments:

The events of Thursday demonstrated that Google is falling behind in the emerging real-time web. It was 3 hours and 17 minutes after TMZ first announced Michael Jackson had experienced cardiac arrest before it appeared as a auto completion suggestion on Google’s homepage. In the computer age that is a huge amount of time. It is 3 hours and 17 minutes during which consumers may choose to go somewhere other than Google to get the information they want.

While this may be an overly glum summary, it certainly highlights a problem that Google has been addressing for some time: the fact that it’s predominantly an indexer of an archive, a curator of content snapshots, a catalogue of the recent past, rather than a finger of the pulse of what’s happening out there at any instant. The much smaller world of Twitter, wobbles and all, is a lot closer to this ‘right here, right now’ model.

We’ll just have to wait and see. Meanwhile, Brewster has a listening date with Billie Jean.