Thanks for the Twitter search widget link!

Hi Zac,

Got any proof? I'm looking at the timestamp on the Post's article, and the timestamp says the Post got that article up 90 minutes after the crash happened. Your posting of this late on a Saturday night, your use of all-caps and your hard-hitting "...the Post dropped the ball. And it fell hard." line lead me to believe you were drunk or were drinking when you wrote this -- I don't have any proof of that either, but it seems making up shit is okay around here.

Joe

p.s. I know I know your point: Hey, Denver Media (aka the Post) could have let folk know that there was a crash (but little else more than "there was a crash") about 60 minutes earlier had we had a twitter widget on our homepage. Thanks for the tip.

Joe, I'm not trying to be a dick. And it's not just the Post I'm coming down on.

But it's a constant failure of media in breaking news situations. Locally, I heard about a business fire 10 blocks from my house through facebook, brightkite and twitter Sunday morning. The paper didn't post anything until today.

As for this particular incident:

Mike Williams tweeted the crash at 5:25 p.m. Central.

BreakingNewsOn had a post on twitter at 6:02 p.m. CST.

Ryan Sholin, and many others by at least 6:45 p.m. CST reported there still wasn't anything on the Post's site.

By 8:18 p.m. one person commented on the Denver story: "Three hours after this plane crash occurred, and you've got a total of five sentences posted, and some of it is grammatically suspect."

In the meantime, people on twitter, including Rob Scoble and others, had snuffed out people at the airport and people on other planes who were reporting to twitter about it.

One has to assume that anyone on the metro desk at any Denver media would have been listening to a scanner and would have heard the calls made to the fire department located right near the site.

I have a friend, a producer in San Francisco who heard about the accident through a producer in Vegas but couldn't find anything on the wires about it either.

There's some good conversation at friendfeed.

I'm not arguing that all the information on twitter was totally accurate. It wasn't. But there's when interest in a story swells up like this, there's got to be a better way to handle it than sitting on your hands.

Thanks for the links and info, Zac. In your timeline in the comments you mix up time zones and get the times wrong. The crash happened at about 6:20 p.m. Mountain, which is 7:20 p.m. Central... you see the problem...

Ryan's tweet (at 7:45 p.m. Mountain) came three minutes before the Post's article made it online. I don't know what happened in the Post's newsroom (or the Rocky's, or AP-Denver, or 9News', etc.). Sure, Denver news orgs could have been faster on the draw. As far as other points of light to be pocketed from this exercise go, here's what I've got:


Reporters should search twitter for sources when big news is breaking
Having a twitter-search widget tuned to your geographic place of interest and posted somewhere on your site (home page? news section?) may be useful


If you've got any more, do share.

Oops on the timezones.

I'm not sure if a simple keyword search is the best way to go. It's a start, but a search for 'Denver' or 'DIA' wouldn't have provided William's and others' tweets from the airport right away.

If I were to build something, I'd think more in terms of lots of keywords, with a weighted system behind them (higher priority given to local people using particular words, for example).

Here's a possible case:

On the article level, a module that pulls 1) possibly related local tweets based on the tags of a story, 2) tweets of people linking to the story, and 3) anyone on twitter talking about related tags to a story.

On the section level, it would be the same types of information, but get more complex, pulling based on tags of all stories in the section, etc.

I'd also think about breaking it out as a constant standalone feature of the site. Maybe map based? Pull from geotagged photos and video through various APIs out there (flickr, google, brightkite, youtube, lots of others).

You could get a little bit more advanced in the API, too. Williams did mention Denver in an earlier tweet before the crash. You could search all of twitter for certain general keywords and then follow up on those people for more specific keywords for a particular amount of time.

This stuff is all possible with the twitter API and some thoughtful server-side work and an elegant UI for users to filter the noise.

It's not the greatest API out there, but it's certainly good enough to pull off something more sophisticated and useful for a news organization.

I don't know. I'm mostly thinking aloud here.