RIP, EveryBlock

EveryBlock, a hyperlocal news start-up that used data to filter neighborhood news and spark discussion, has been shut down by its corporate overlord. Apparently, NBC News acquired it last year (which itself was news to me) but couldn’t find the business model to continue operating the site. That’s a common tale among hyperlocal news sites, but it still stings when one closes down.

It’s too bad — I think it had more going for it than many similarly themed sites — and its founder, Adrian Holovaty, seems shocked that the site has met its end. When he sold the site last year, he was proud of its success and confident in its future:

“EveryBlock users have used our service to accomplish amazing things in their neighborhoods: starting farmers markets, catching flashers, raising money for their community, finding/reporting lost pets…and generally getting to know their neighbors and forging community bonds. These days, something like this happens on the site nearly every day — which casual onlookers might not notice because of our long-tail, neighborhood-specific focus. EveryBlock has become a force for good, and it’s got a bright future.”

Sigh. I suppose it’s not particularly interesting that a start-up failed to locate a business strategy or that it didn’t “pivot” quickly enough to “disrupt” via its “MVP.” What is interesting about this case is that the site was a news-centric one that really challenged newsgathering tactics, asked questions about the use and display of public data and, in its small way, wrought lessons for the [cue horror-movie scream] Future of Journalism. It began, after all, as a recipient of a Knight Foundation grant.

Even more interesting is that it evolved so much over its short life (actually, wait, is six years long or short in technology?). When it began, it was just one news-tech guy’s realization that news should not be story-centric but instead should be gathered as structured data. He married the programmer’s philosophy of the separation of content and presentation with the journalist’s instincts for ever-better storytelling. Holovaty’s blog post from September 2006 is, in retrospect, both amusing and prescient. In it, he calls for parsing data and creating CMSes that support content types other than words, two notions that are laughably obvious six years later.

(On the flip side, also laughable are the mention of PDAs and the idea that tagging was “trendy.”)

Holovaty turned those 2006 idea germs into EveryBlock’s mapping and reporting functionality and, ultimately, he created a robust community around neighborhood news. The site put forth a notion of what the oft-dreaded Future of Journalism could be, or one version of it, anyway. It tried something new. It experimented. And the experiment did yield results; unfortunately, the conclusion was that this model might not be quite right.

In its sad and clearly hasty post today confirming the shutdown news, EveryBlock seems to acknowledge that it was a victim of the unforgiving pace of change in the online journalism industry:

“It’s no secret that the news industry is in the midst of a massive change. Within the world of neighborhood news there’s an exciting pace of innovation yet increasing challenges to building a profitable business. Though EveryBlock has been able to build an engaged community over the years, we’re faced with the decision to wrap things up.”

In short: “We tried. We’d like to keep trying, but trying doesn’t pay the bills.” And that’s too bad.

 

Read More

Narrative Science and the Future of StoryTelling

kris hammond narrative science

On Friday I had the good fortune to attend the Future of StoryTelling conference. Among the leaders and luminaries in attendance (whose names I will not drop here) was Dr. Kris Hammond, who is the CTO at Narrative Science, which has created an artificial intelligence product called Quill that transforms data into stories (the product generates a story every 28 seconds, per Hammond). I’ve written about Narrative Science before, and I argued in that post that Narrative Science “is not a threat, it’s a tool, and it fills a need.”

Now that I’ve met Dr. Hammond and heard him speak, I’m more a believer than ever that this is the future of journalism — and not just journalism, but all of media, education, healthcare, pharmaceutical, finance, on and on. Most folks at FoST seemed to be open to his message (it’s hard to disagree that translating big data into understandable stories probably is the future of storytelling, or at least part of it). But Hammond did admit that since the Wired story came out in which he was quoted as saying that in 15 years, 95 percent of news will be written by machines, most journos have approached him with pitchforks in hand.

I went in thinking that the two-year-old Narrative Science went hand-in-hand with Patch and Journatic in the automated-and-hyperlocal space, but I now think that Hammond’s goals, separate from these other companies, are grander and potentially more landscape-altering.

I know I sound like a fangurl, but I was truly that impressed with his vision for what his product can be, and what it will mean to the future of journalism. No, it can’t pick up the phone and call a source. It can’t interview a bystander. It can’t write a mood piece…yet. But they’re working on it.

With that, my top 10 quotes of the day from Dr. Hammond:

The first question we ask is not “What’s the data,” it’s “What’s the story?” Our first conversation with anyone doesn’t involve technology. Our first conversation starts, “What do you need to know, who needs to know it and how do they wanted it presented to them?”

Our journalists start with a story and drive back into the data, not drive forward into the data.

We have a machine that will look at a lot and bring it down to a little.

The technology affords a genuinely personal story.

It’s hard, as a business, to crack the nut of local. For example, Patch doesn’t have the data, but they’re the distribution channel. There’s what the technology affords and what the business affords…. We don’t want to be in the publication business.

Meta-journalists’ [his staff is one-third journalists and two-thirds programmers] job is to look at a situation, and map a constellation of possibilities. If we don’t understand it, we pull in domain experts.

The world of big data is a world that’s dying for good analysis. We will always have journalists and data analysts. What we’re doing is, we’re taking a skill set that we have tremendous respect for and expanding it into a whole new world.

The overall effort is to try to humanize the machine, but not to the point where it’s super-creepy. We will decide at some point that there’s data we have that we won’t use.

Bias at scale is a danger.

The government commitment to transparency falls short because only well-trained data journalists can make something of the data. I see our role as making it for everybody…. Let’s go beyond data transparency to insight transparency. It can’t be done at the data level, it can’t be done at the visualization level, it has to be done at the story level.

Read More

Making data-driven journalism “work in the best way possible”

Great column by Craig Silverman at Poynter.org that gives a clear-eyed assessment of the “inevitable” shift to data-driven journalism. Good stuff.

“Journatic’s approach — and the change it represents — is not going away. That means it’s important for journalism to find ethical, responsible and productive ways to integrate these approaches. To set benchmarks and guidelines for producing quality content using the kind of low-cost labor and mass production techniques that were long ago adopted in manufacturing. To find a better way forward.”

“You have to determine which stories can be written from afar, and which must be done by those with local knowledge. … The starting point is to establish policies, procedures, and standards to guide outsourced, mass production content operations [for] quality control.”

Read More

Why the Journatic controversy is a good thing

The Journatic fallout continues, and apparently the story has legs. On the heels of the controversy around it systematically faking bylines so its offshore labor could appear to be nearby to its clients (that is, local newspapers) and named “Jimmy” and “Ann,” one of its biggest clients, TribLocal, discovered plagiarism (from Patch, no less!) and suspended its use of Journatic indefinitely, saying:

“[Fake bylines and plagiarism] are the most egregious sins in journalism. We do not tolerate these acts at the Chicago Tribune under any circumstances, whether from a staff member or an outside supplier like Journatic.”

But Tribune Co. is actually also a Journatic investor, so that’s a bit of sticky wicket, innit?

Then one of Journatics’s high-ranking (and quite recently hired) editors, Mike Fourcher, quit, on the grounds that Journatic is attempting to “treat community news reporting the same way as data reporting”:

Inevitably, as you distribute reporting work to an increasingly remote team, you break traditional bonds of trust between writers and editors until they are implicitly discouraged from doing high quality work for the sake of increasing production efficiency and making more money.

Cutting through the noise, it sounds like he tried to argue for paying people more for better quality stuff, and Journatic’s owners balked.

As I have said, hyperlocal, algorithmic journalism at scale is such a tough area, and one that’s evolving all the time (actually, at a very quick rate, if you take the long view). But the Venn diagram of quality, quantity, turnaround time, local expertise, ease of assignment, keeping readers happy, keeping writers happy, keeping staff editors happy, data-mining technology costs, platform costs, actually making money — and, you know, not lying about any of it — it’s not an easy nut to crack, and that’s why no one’s done it yet.

My dabblings in this area at now-defunct Seed certainly didn’t pan out as planned. But nonetheless I agree with Fourcher, the ex-Journatic guy, on this:

Journatic’s core premise is sound: most data and raw information can be managed much more efficiently outside the traditional newsroom; and, in order for major market community news to be commercially viable, it needs be conducted on a broader scale than ever before.

For Journatic’s part, it released a statement saying: “We are in the process of conducting a thorough review of our policies, software, technology and personnel. We are immediately and forcefully addressing the issues we find and making changes where necessary. Until we have completed our review we will decline any further comment.”

So all of this being said, now that TribLocal is back in the hands of “real” journalists, what will happen? Will the quality of coverage be so amazing that readers demand it continue? Will they even notice? Will the cost of paying writers who can write well in the first place be less than Journatic’s current model of paying editors to correct the writing of non-native English speakers, then selling that as a third party to TribLocal and others? Will the other papers who use Journatic’s service (the Chicago Sun-Times, the San Francisco Chronicle, the Houston Chronicle) also balk amid the controversy? Will there be a resurgence in hiring actual journalists to cover local news?

All remains to be seen, of course. But it’s exciting, because at the very least this kerfuffle has people (lots of them!) talking about this, and publicly instead of in back room deals and investments about which local readers are unaware. The Fourth Estate is actually weighing in on a controversy, doing their jobs — reporting on it, ruffling feathers, making waves. And ultimately that is a very good thing for us all.

Read More

Journatic and the future of local news

This week’s This American Life featured a segment on Journatic, a hyperlocal, scaled-content creator that’s apparently replacing local reporters in many markets. I’ve previously written about hyperlocal news and the value of using algorithms in news creation, so the story was of great interest to me.

My argument with hyperlocal is that no one has yet figured out how to do it right. It sounds to me like Journatic is finding some success, but it’s also failing in important ways. My defense of algorithms is mostly to do with the company Narrative Science, which as I said is “not a threat, it’s a tool, and it fills a need.” That need is basically the scut work of news reporting, and although the folks there are working on this very issue, for now, “It’s a tool that does a programmatic task, but not a contextual one, as well as a human.”

Journatic aims to solve the hyperlocal problem with the algorithmic solution. The company scrapes databases of all kinds, then uses that data to “report” on local bowling scores, trash pickup times, where the cheapest gas is, and who has died recently. The company does this by using algorithms to mine and sort public information, and there’s nothing necessarily wrong with that.

When it launched, Journatic-populated site BlockShopper was basically a real-estate listings site based on publicly available data. Using public records, it would “report,” for example, that “123 Main St. is in foreclosure.” But since then, the algorithms and tools have gotten smarter. Soon it was able to say a home was in foreclosure “by the bank” and also add that it “is up for auction on March 31.” The site is now so smart that it actually feels almost invasive. To wit:

The real estate information contained in the article is publicly available, from the names of the people involved in the transaction to the price paid to the location details. The fascinating thing, and what pushes it into a brave new frontier of journalism and privacy invasion, though, is that the information on the professions of the involved is also publicly available (probably via LinkedIn). Arguably, all the article is doing is presenting public data in a new format. The difference is access and availability. In the pre-Internet days, there was no way to know public information except to go to the city records office and look, and there was really no way to know about peoples’ professions except to know them or ask them. These tasks required interested and motivated parties (such as journalists), because actually going places and talking to people requires on-the-ground reporting (not to mention complicit consent). This is not the sort of work Journatic traffics in. That’s not a criticism, necessarily, just a fact: There used to be barriers to the information; now there aren’t; Journatic uses this lack of barriers plus its algorithms to surface the data.

 

Journatic aims to solve the hyperlocal problem with the algorithmic solution.

 

At first, the company didn’t do any (or much) writing or analysis. According to This American Life and its whistle-blower, though, the company now pays non-native-English-speakers in the Philippines between $.35 and $.40 a story to try to add a bit of context to the data. Thirty-five to forty cents! However shady this is, though, it is not necessarily unethical. It’s capitalistic, and it’s pretty shameful, and it feels wrong somehow, but it’s not unethical journalistically.

Where it does get unethical is when readers are misled, and that has apparently occurred. They force these writers in the Philippines to use fake bylines like “Amy Anderson,” “Jimmy Finkel” and any number of fake bylines with the last name “Andrews,” in order to Americanize them and dupe readers, according to the show. This is flat-out wrong, and I think Journatic knew it — they apparently reversed their stance on this after the story aired.

But ethics aside, and journalism in broader context here, Journatic’s founder, Brian Timpone, claims that the “single reporter model” doesn’t work anymore. The Chicago Tribune, one of Journatic’s customers, says that it’s gotten three times more content for a lot less money. These are serious issues for the future of the profession (along with the opportunity for privacy invasion and privacy mishandling that all this unfiltered data presents). It’s no doubt true that the Trib paid less money for more content versus hiring local reporters. But what is the quality of the work? I think we all know the answer. Shouldn’t that be a bigger factor than it is? If you’re just turning out junk, your brand gets diluted, and your readers soon abandon you altogether.

It’s easy to criticize, but it seems to me that Timpone is trying, as we all are, to devise a way forward. That’s admirable, in its way. It’s a little scary, and the desire for progress sometimes makes us color outside of the lines, and when that happens, places like This American Life need to be there as a regulator, as has just happened. We’re all still muddling our way through the ever-changing new online media landscape, and we will test theories and make mistakes and learn lessons, and with any luck we will end up with a better product, one that serves readers first, last and always. I hope someone is able to someday crack the code of good news done quickly at good quality for a good wage. Until then, we must keep trying.

Read More

Hyperlocal news in context

The New York Times quietly announced that it was ending its three-year-old experimentation with hyperlocal journalism, and on its heels, the Neiman Journalism Lab wrote a piece following up with some of the players and laying out five specific lessons learned. These lessons are:

1. It just doesn’t make sense for big media companies to pay their staffs to go hyperlocal.
2. Hard-hitting hyperlocal coverage benefits from some professional journalism.
3. Create a platform that makes it easy for people to participate in diverse ways.
4. Understand the power of email.
5. Don’t abandon experiments in “innovation land.”

Hyperlocal — which I define as small-time stories that are confined to a geographic area but which are incredibly important to that area — is a serious puzzle, one that no one has found a good solution for yet. When hyperlocal had buzz — probably back two or three years ago now, when it wasn’t yet clear that it was such a quagmire — the indicators were that as papers were shutting down left and right, folks needed an online equivalent for news and such. The argument was that print papers were closing down because their “business model” — which is to say, putting interesting and germane copy around relevant ads — was failing, but online news would somehow be the answer. For some unknown reason, but probably because it was much, much cheaper, the people running these new hyperlocal startups (or sometimes online versions of existing papers) also convinced themselves that, in addition to going online instead of print, they would also do it on the cheap: Instead of paying experienced beat reporters to do good ol’ writing about the day’s local news, a model that had worked forever, they would instead fired those people, “engage the community” and hire “citizen journalists.” In retrospect, that didn’t work so well.

According to the Times‘s Jim Schachter, it turns out it’s “impractical” for the New York Times, being a national and international news org, to turn to hyperlocal coverage. Well, yeah. Despite the fact that the Times itself has a well-read City Desk (read: hyperlocal news), it’s not their model to start an equivalent in Wherever Else, USA. They know New York City. They have a giant, well-recognized brand in which local places want to place their trust and their ads. It’s a good combination — it’s a business, in fact! — called “the news media.” It’s “scalable” in that it works (or used to) in almost every community across the world.

Another now-obvious hyperlocal lesson learned: Professional journalists are good at writing and editing. Non-journalism-skilled “citizens” aren’t necessarily good at that stuff. Extrapolating, when readers want information, they want it to be relevant and clear. They don’t want to have to work for it. When these waters are muddied — trying to parse what a non-journalist is trying to say among possibly irrelevant, definitely badly written prose — it is not a fun time. Readers’ response to this isn’t to get even more engaged and volunteer to be the citizen journalist, much as the bottom-liners at media companies wish they would. It’s not even that they get riled up and want to comment about the quality. What readers do is, they just stop reading.

The thing is, though: People care deeply about and do want to read about local news. They care about school board meetings and and city council decisions and high school sports scores and local heroes. They just don’t want to go to the board meetings or ball games themselves. They don’t care that much. Plus, they’re busy doing their own jobs.

So these are two key realizations of the hyperlocal business: You have to provide relevant and well-written copy that doesn’t ask anything of the readers other than to read. (And maybe, if it’s well-written and relevant enough, they might actually pay for the privilege and/or comment on a story.)

Another essential understanding is recognizing and respecting a corollary phenomenon: the rise of social media as a determiner of local relevancy. The Neiman article doesn’t touch on it at its own embarrassment, since engagement is this area is absolutely insane, off-the-charts, disruptive, phenomenal, revolutionary, whatever word or phrase that means a new paradigm has been created. That a new paradigm for social interaction occurred simultaneously yet oppositionally with the attempt at a new paradigm for local news is a coincidental but unfortunate event (unfortunate, that is, for these news sites).

My understanding of the fourth finding, “Understand the power of email” is that people like newsletters. But in my experience this is neither true nor relevant. However, the third lesson, the idea that technology must facilitate participation, is huge, and it’s another key point to answering the question that Schachter proposes: “How do you prompt communities into the act of covering themselves in a meaningful way?” I don’t have the answer (does anyone?), but I think social is playing no small part in this, too, and one need only see all the check-ins and status updates to see that people do like participating in the community around them (and, in turn, telling others about it in order to either humble-brag or exacerbate their followers’ FOMO or both).

I almost want to (badly) paraphrase Mean Girls by saying, “Stop trying to make hyperlocal happen.” But actually, I do think hyperlocal has a place in our evolving news, online and social ecosystem. Hell, I think algorithms have a place. It’s early days yet. But it seems to me that what successful products have in common, and what they all come to realize sooner or later, is that above all else, they must serve the reader (or user). It’s so incredibly obvious yet so often overlooked. And as soon as hyperlocal sites incorporate this truth into their businesses, the better the experience will be for readers and the online news industry at large.

Read More