Notes on the Watershed Post's implementation of Ushahidi's Crowdmap

[Guest post by Julia Reischel of the Watershed Post] Several days before Hurricane Sandy arrived in New York, our two-person staff at the Watershed Post began building our Hurricane Sandy page: www.watershedpost.com/sandy. We had learned from experience during Tropical Storm Irene in 2011 that early preparation and installation of web tools was indispensable in covering a disaster. During Irene, our live blog became the locus for disaster coverage about the Catskills nationwide. It attracted almost 100,000 reads in three weeks and was featured in media outlets like the New York Times and CNN. This fall, we decided to use a Ushahidi's hosted Crowdmap platform to organize our coverage of Sandy in the Catskills.

Here's a list of what worked and what didn't:

What worked:

--The setup: Aside from the email feature (see below), setting up our Crowdmap was easy and fast. This was of tantamount importance to us. We had one day to prepare for the storm, and we went from knowing nothing about Crowdmap to using it on our site in hours. --The categories: We divided our Crowdmap into seven categories of fast-moving data and information that need to be tracked during a Tropical Storm disaster. These are 1) places where relief is needed, 2) shelters, 3) places to get help, 4) stranded people, 5) road closures, 6) reopened roads, and 7) storm reports. --The design: We got many compliments on the cleanliness and simplicity of the Crowdmap from our readers. During Irene, we had used a messy set of Google Documents spreadsheets and GIS maps to categorize data on road closures and relief resources. Crowdmap gave us an elegant, visually-intuitive interface in which to store all of that disparate data. --The map: It was great to geotag the data and have it searchable in the map itself. The search function was robust and made it easy for us to find things. --Embeddability: Ushahidi's Crowdmap platform made it easy for us to embed the Crowdmap on our site on a special Sandy page: www.watershedpost.com/sandy. Most of our readers were unaware that the Crowdmap was being hosted on another platform, and our URL became the go-to place to find it. We distribute content for free and generateour revenue from ads, so third-party apps that embed seamlessly are very important to us.

What didn't work:

--Geocoding and addresses: While we love the crowdmap's geocoding feature, our contributors and even our trained volunteers had trouble correctly geocoding addresses. This is due in part to the confusion of having two separate "location" fields in the report form, one that geocodes on the map and a second that is meant for more general directions. Our users often put the address into the second field, and then the geocode would default to a non-specific location on the map, making it hard for editors to know that the location had been geocoded incorrectly. A suggestion: By removing the extra location field and forcing all users to geocode their addresses, we think this problem would be solved. --Bulk editing: We found that the lack of a bulk editing feature made it very difficult to change the fast-moving data on our map. For example: Our list of road closures had to be edited one-by-one when the roads re-opened. Having a batch edit feature that would allow us to select many roads at once and move them simultaneously from the "closed" to an "open roads" category would have speeded up delivery of this vital real-time information considerably. --Reports by Twitter: We found that in practice, taking reports via Twitter hashtags created problems. For example, when anyone retweeted the hashtag, a duplicate report was made, creating more work on the editing side. Also, reports submitted by Twitter were automatically assigned to the last category on our list, which required immediate editing to fix. --Submitting reports via email: We could not get this feature to work in the limited time that we had to set it up. --Site speed: The Crowdmap site is relatively slow -- it took a while for it to process data input. Since we had to input data one item at a time, that meant that updating the map took quite a lot of time.

Deployment Summary:

In general, we found that while Ushahidi Crowdmap provided an excellent way for us to organize data during a disaster, it was difficult for anyone other than a full-time, dedicated and trained member of our staff to use it. There were few user-generated reports submitted to our map, and of those that were submitted, many were off-topic or irrelevant. This is partially due to the fact that Sandy did not hit the Catskills as hard as Irene did last year, but it is also due to the fact that the Crowdmap interface did not reliably guide users into submitting useful, error-free information to us. We found that our role as a mediator between users and a system for organizing data was indispensable.

Background on the Watershed Post

The Watershed Post was founded in January 2010 by Julia Reischel and Lissa Harris as an experiment in sustainable rural online news. The experiment has been a resounding success: Today, we are the go-to source for news about the Catskills, with 25,000 readers a month. Our coverage of Tropical Storm Irene in 2011 was recognized by the New York Times, CNN, WNYC, Democracy Now, The Guardian, Chronogram, Hudson Valley Magazine, and the Columbia Journalism Review. We also received the 2011 “New Business of the Year” award from the Delaware County Chamber of Commerce, the 2012 “Entrepreneur of the Year” award from the Central Catskills Chamber of Commerce, and citations from the New York State Senate and Assembly for that work.