UGC in Uttarakhand, and useful tech tools


A grab bag of a post today, organized around no particular theme.

Item 1: Any tips for verifying user-generated content?

I’m throwing this out there because I’m not sure what to do. The past few days have seen an outpouring (terrible pun not intended) of posts on social media about the ongoing natural disasters in Uttarakhand. There are great opportunities here for social media. In the past two days we’ve created an interactive map of user-generated pics and  put out a call to help connect missing family members.  But then I had an incident like this morning. A user sent in a terrible – but very newsy – video of a car being washed into a river by a flash flood.  I published it, but emailed him to ask him about the provenance.

No reply, but six hours later the same user submitted two more videos, also very dramatic, also clearly shot on mobile phones, purporting to be taken at sites within about 46 kilometres of each other. Now here’s my question: did this guy actually witness three buses/cars washed away into oceans or gorges in a single day?  (And if so, god, worst day ever).  And if he didn’t, where are these videos actually from?

No response from the user.  The videos are fantastic footage, if they were really taken by him, at the sites in question, in the past day or two. I searched for his name on YouTube – no hits. I looked for “Bus accident – xxlocation” in Google News and Google Images. No matching hits.  (Normally, Google’s “search by image” is an invaluable tool for figuring out if users have whacked pictures off the Internet and are passing them off as their own – if you don’t know how to use it, learn!)

Beyond all that, what solutions can people recommend? What are other folks doing, if anyone is using UGC to report on Uttarakhand?

UPDATE: Spoke with the user on Twitter and email, he confirmed the videos had been shot  by him and a few of his friends, on June 17th.  I posted one on Facebook.  Sure enough, within minutes another FB user mentioned that he’d seen the same video on YouTube a year ago, and provided a link.  The incident actually took place in Bolivia.  Any tips for how to handle it now?

Item 2: Free tools that are worth some time

Expecting every piece of digital journalism to be “Snow Fall” is like expecting every news article to be the Watergate investigation.  Here are some WYSIWIG tools to produce interesting things in a hurry.

1. Google Maps – surprisingly customizable. The only downside is that they don’t provide embed code for personalized maps (or do they, and am I missing it?)  We used this one to create an interactive map of north Indian floods.

2. Timeline.js – I haven’t used this one as much as I’d like, but it creates lovely timelines in a hurry.  If you’ve got a coder handy, a few simple hacks and you can do a lot – change the orientation from horizontal to vertical, alter fonts, etc.

3.  Storify – This one actually has a bit of a learning curve, or at least it did for me.  Nothing beats Storify for pulling together social media content fast.  The carding is beautiful. The embedding is painless and incredibly attractive. My only gripe is the awkward “read next page” blue link that they throw into the middle of a story. It’s unattractive and I doubt most users click past it, no matter how fascinating the content. Storify is great for sites like ours because it gathers our primary content in one place. We use it to create social news stories, where we gather social reactions to questions we put out in response to trending news topics.  They offer paid accounts with all kinds of promised upgrades, but I haven’t used those.

4.  Meograph – I don’t just list it because I know the founder.  An attractive tool for pulling together a lot of content into a single almost encyclopedic video, including Maps/data/etc. I love how clickable things remain within a Meograph vid (a bonus for Storify too) but playback on Meographs can be slow, or at least it was on the couple I’ve tried.  They also have paid options.

Pic by Walter Siegmund.

This story was written by robots

AINot this story, the one you’re currently reading, but this one, the one you’re about to click on: Robot story.

Narrative Science, which just raised an undisclosed amount of funding (partly from the CIA), uses artificial intelligence tools to produce business intelligence reports.  In the words of Peter Kafka, they turn “structured data sets into prose.”

As a high school student I remember learning how to create form letters in Microsoft Word, where we plugged in data values into empty fields.  I originally thought that NS’s tools must be a souped-up version of that.

It turns out that their tool, Quill, is quite a bit more complex.  From their description: “Quill applies complex and sophisticated artificial intelligence algorithms that extract the key facts and interesting insights from the data and transforms them into stories.  The resulting content is as good or better than your best analyst, and is produced at a scale and speed only possible with technology.”  (Emphasis mine – read more about Quill at their link.)

The most interesting thing about Quill is not that it grabs facts, but that it calculates the relationships between them. (You can see that calculation at work in the Forbes story above, which is propelled by relationships like the following: “The company has been reaping profit in the past eight quarters, and for the last four, it has seen an average of 19% growth in profit year-over-year. The biggest boost for the company came in the most recent quarter, when profit jumped by 48%.”)

For the very nerdy (and possibly unemployed): NS’s full patent filings are here and here.  What I find most fascinating about these filings is the mathematization (yes, I invented that word) of basic news story structure. At Northwestern (where some of NS’ founders are from) we were taught to write a news story like an “inverted pyramid,” stacking the most relevant information up top. When combing facts for a common thread, we would often ask, “so what’s the angle?” NS’ technology uses the exact same language.

In one of Nieman Lab’s “Future of Journalism in 2013” reports, Miranda Mulligan said the rise of the robot is one of this year’s biggest journalism trends.  She wrote that enhanced robot tech will make journalists’ jobs easier (rather than replace journalists altogether). Mulligan quoted Siri founder Dag Kittlaus, who has said that technology only needs to be about 90-95% accurate in order to go from novelty to utility.

Of course, if you subscribe to a business intelligence service, 95% may be better than what you’re currently getting.

Earnings reports are the utter, necessary drudgery of the financial journalism world. If we lived in a castle, writing earnings reports would be about as exciting as emptying chamber pots. It’s also one of those tasks that even experienced humans – extremely skilled ones – frequently get wrong.  (Some of these mistakes can have serious repercussions if the mistaken company gets angry enough about the error.) To be honest, most of these reports are already so formulaic in their construction that they might as well have been written by robots.

There’s certainly a case to be made for automating tasks like writing earnings reports, freeing up reporters for putting together the more dazzling data stories that sometimes lie buried in earnings reports and company filings. Spotting a real story in an earnings report takes serious skill and a talent for numbers (we dedicated several hours to it in a business journalism course I took).

Nor is NS the only firm headed in this direction. Recently, I had lunch with the co-founder of a firm that’s experimenting with robotic news anchoring.  This isn’t artificial intelligence in the way that Quill is. They would put together a complex film database of a news anchor’s facial expressions, pronunciation and intonations, and then use this database to animate news scripts.  Again, what news anchor wouldn’t mind if their robotic avatar could take the 6 am newscast? News never sleeps, but the same is not true of people.

Any student of economics has heard the theory that technology powers real economic growth. Interestingly, the advent of technology has not allowed any of us more rest – if anything, we work longer hours and in faster-paced jobs than ever before. So perhaps what we journalists should really fear – if robotic journalism becomes the future – is not that we’ll be replaced, but that we’ll have to work even harder, faster and longer than we already do to match our robots’ inhuman pace.

Blog at