Why All Local News Aggregators Seem Destined to Fail

I have yet to not be disappointed by a local news aggregator, and I consider myself a devoted connoisseur of their consistently unusable incoherence. Everyone wants these things to work – who doesn’t want to know what’s going on in their neighborhood? – so engineers continue to try, and fail, to create one that isn’t awful. The thing is, it’s not technologists’ fault.

Here’s a screenshot for my neighborhood, a suburb of Washington, DC, from web geotagging startup Fwix:

To be fair, local news doesn’t appear to be Fwix’s main play; even so, their algorithmic approach to it is illustrative of the shortcomings of this approach.

It’s a classic problem of signal and noise. Google News, the important bits of which are entirely automated, has an incredible amount of high-quality signal to play with. Despite the fallout in the media industry, hundreds of outlets are still throwing thousands of the world’s best journalists at the day’s top stories. It’s an embarrassment of riches.

But how can you figure out what stories are relevant in a town or neighborhood when the local media is so thin on the ground? And how do you even define “newsiness” or “relevance” when there’s hardly enough material to allow a user to narrow their interests?

Fixing the local news conundrum

I’m not convinced that local news aggregation can ever work. But if it did, here’s how it would happen.

1. Use humans - yes, humans - to figure out what the good local news sources are.

Google News may be an algorithm, but its supply of high-grade starting material comes from outlets vetted by human beings. Whoever’s in charge of this for local news aggregators always manages to appear to not know what the best sources are. Perhaps all of these startups underestimate the scale of this problem and the resources required to do it well.

2. Use more humans to pick that day’s top stories.

Another heresy, but how else can we guarantee that the day’s top story won’t be incomprehensible gibberish? Compare the Fwix results for my neighborhood to AOL’s human-curated Patch. Patch, at least, presents information I might actually want to read.

3. Use even more humans to engage with the humans that are making local news.

Who better than local bloggers will know what’s important in a given location? Old-fashioned relationship building would go a long way toward identifying who is knowledgeable in a city and, perhaps, getting them involved in highlighting the most important stories in their area. (A system like NPR’s Argo network seems optimal for this approach.)

4. Recognize that local news is just that – local.

Attempts to create “local news” sites that span the entire country are prima facie problematic. What makes local news interesting and relevant is the very thing about it that resists homogenization – its individual, local character. Yes, Starbucks is ubiquitous, but that’s because Starbucks managed to impose its character on the entire country. A news aggregator, on the other hand, has to reflect the character of every place it purports to cover.

In short, local news is not a problem that’s going to be cracked with an algorithm – or at least, it hasn’t been yet.