I’ve written before about the need for newspapers to go hyperlocal — allow readers to get ultra-targetted news about where they live from their neighbors. Topix, a company owned by newspaper companies, is offering a platform to do that. Topix founder Rich Skrenta’s blog post describes the new product and the thought process behind it.

The site is essentially a series of thousands of local blogs where volunteer editors promote, submit and edit stories. Editors become the equivalent of a newspaper wire editor by selecting from among stories Topix has crawled throughout the Web and posting them to their town’s front page.

Editors can also create stories from scratch. While the New York Times claims that it has “all the news that’s fits to print”, the reality is that major newspapers only print the news that fits. Stories that may have a lot of importance to a specific neighborhood, like what store is going into that vacant space on the main drag, rarely make it to print.

Readers can comment on stories.

In the absence of volunteer editors, an automated process called “roboblogger” fills in the gaps with news stories from Topix’s web crawling.

The new Topix is one of the more credible local offerings I’ve seen. The new site is clean and easy to grasp. The Topix tagline “Your town. Your news. Your take.” clearly sums up the goals.

There are some things I’d like to see:

Open up editing to everyone

You have to apply and be approved to become an editor. This involves facing a big blank box that says “Please describe why you would like to become a Topix editor. Include any relevant experience and qualifications.” I put in one line of text and was approved in about an hour, so at least right now the “review” isn’t very thorough. Even so, the Wikipedia model where anyone can participate seems more appropriate for this environment than a top-down approach.

Reduced clutter in the results

One of the problems with roboblogger is that it’s based on entity extraction. In areas like Washington, D.C., New York and San Francisco, which are home to multinationals, government agencies and nonprofits, entity extraction turns up a lot of false positives.

My Arlington, Va., page had a story titled “Identity Theft Driven By Dramatic Spikes In Threats”. It’s not a local story by most definitions, but because one of the companies quoted is based in Arlington, it shows up. In theory, moderators will be able to kill stories, but with the volume of data the newswires spew, this is an impossible game of whack-a-mole.

More local

Manhattan is treated as one place. People in Chelsea probably care a lot less about things that are happening on the Upper West Side. DC’s ZIP codes come back to the same location.

Recognize neighborhood names and let editors more precisely define what area a story is about. Use the knowledge of editors to improve the entity extraction. For example, Rosslyn, Clarendon and Ballston are sub-segments of Arlington. A story that makes reference to those sub-segments is likely to be more local than a story that references just Arlington.

Leverage local bloggers

There are thousands of bloggers out there who write about what’s happening in their areas. I regularly read Metrocurean and DC Foodies. Relevant posts from local bloggers should be included, with the ability for local bloggers to become editors and automatically publish their stories into the system.