Learning lessons from ReliefWeb Labs
As another ReliefWeb Labs project - ReliefWeb Lite - is 'decommissioned', this post explains why that can happen, even with successful projects, and the benefits of abandoning prototypes that don't quite fit.
Since 1996, ReliefWeb has seen a lot of changes and for the past six years, Labs has been the place for trying out new ideas. It is also used to inform the humanitarian community about other services, such as the Humanitarian Data Exchange (HDX) and the Events Service.
For any technical changes to ReliefWeb, the priority is to keep the site usable and useful. A large part of usability is performance. For all users, the time it takes for ReliefWeb servers to answer a request is important, but for those on low-bandwidth connections, the size of the response can be decisive.
m.reliefweb.it, the ReliefWeb mobile version, launched in 2013 as the first Labs project, stripped down both the amount of information offered and its presentation for very light and fast access to ReliefWeb content.
The ReliefWeb API makes this possible. It separates the content - all the reports, jobs, training, etc - from how they are displayed at https://reliefweb.int, allowing much wider use. The mobile apps also rely on this as does RedHum and some displays on the OCHA corporate site and HR.info site.
When the mobile site was launched, smartphones were mainly used for accessing the web on-the-go. Now, the distinction is no longer between low-bandwidth mobile sites and high bandwidth desktops, but serving a site appropriate to a visitor's bandwidth is still very complicated. (The W3C, the body which develops standards for the web, tried to tackle this but has given up - for now.)
ReliefWeb Lite - also built on the API and featured in Labs since December 2017 - was an experiment with a new, responsive design, improved accessibility, offline capabilities and app-like behaviour. Building the site from scratch allowed these improvements, but we ended up with two similar ‘lightweight’ sites and chose to keep the one that was easier to maintain.
Fortunately, the innovations were easy to adapt to the mobile site, which now:
has a responsive design
remembers and displays previously visited pages when there's no connection
is more accessible for users with disabilities
can be installed as a Progressive Web App on recent Android and iOS smartphones
These innovations will in turn be used to make reliefweb.int responsive and more accessible.
In preparing this early retirement for RW Lite, we've taken the opportunity to update the descriptions of other labs projects that have been decommissioned. While we'd like each project to be a success, we hope it's useful for others to see what we've tried and why we decided not to continue with different ideas.
So what have we learned?
Stand-alone sites must be easy to maintain.
Not all formats are the same.
Most humanitarian information is still produced in pdf format, which is not without its problems but is easy to archive. We've recently added a way to display interactive content but an attempt to cover video content showed that the level of interest in humanitarian videos did not keep upward momentum.
The API is an important humanitarian resource.
The mobile site, RW Lite, Redhum, the Headlines and Jobs apps are all built wholly on top of the ReliefWeb API, which also provides most of the content for the Crises app and supplies information for unocha.org and humanitarianresponse.info. The API is fully documented and open for use by other humanitarian organizations. If you have questions about how to use it, please get in touch.
We want to share what we've learned.
As we consider content, presentation and technology for ReliefWeb, we realize these questions also apply to the wider humanitarian community. We're curious to see how long the pdf continues as the 'standard' reporting format for humanitarian organizations, and what might come next. We're experimenting with different formats to share the information directly with those who can use it. And while this post has tried to avoid the technical details, we'd be happy to share more information if there's interest. Again, please let us know.