⇓ More from ICTworks

Lessons From Failure: ICT4D Twitter Chat Synopsis

By Michael Downey on December 16, 2009

While many practitioners and pundits in the ICT4D community agree that the field has had more than its share of failure – with some even claiming an unbroken string of failures – little work has gone into understanding why these failures happened, and even less work sharing failure with others.

ICTworks attempted to change the nature of the community’s discussion of failures last Friday with “Fail Day”, the second edition of the now-monthly #ICT4D Twitter Chat.

By all measures, the event was a success, with nearly thirty participants — including researchers, practitioners, and interested people from industry and government — not to mention an even larger audience of observers. The event’s guiding questions were simple: What are the major types of failure in the ICT4D field, why are we faced with these patterns, and what can be done to change the course?

Unsurprisingly, at the end of 90 minutes, there were arguably more new questions than answers. (A full transcript of the event is available.) However, a few important themes arose:

1. Improper or missing understanding of users.

Technologists, like all people, have a tendency to dream big. As a result, systems are often over-engineered in the minds of the designer, without taking the time to base the design on real data points from potential users (or even from other projects). Designed in a vacuum, these projects are nearly guaranteed to fail — users in the developing world often end up finding very little in common with an application developer in the United States (for example).

The human-computer interaction field is partly to blame for this pattern, as we haven’t done a good job providing low-cost, meaningful ways to better understand users in the global south. We certainly haven’t promoted use of simple techniques such as use of personas, interviews, and surveys in the early phases of ICT4D projects. Some of us, including me, are working to change this, but it will not be an overnight process.

2. Failure to focus on real problems and needs.

A failure to understand real needs is somewhat related to misunderstanding of users. After all, the better technologists are at understanding users, the better they can understand their needs. At OpenMRS, we have a mantra that “care will lead the way”. This means that every bit of functionality can be traced to actual, real-life needs of the clinics and health care professionals we serve.

Unfortunately, this type of understanding can only be achieved by spending time first-hand in the environment where an ICT4D solution will be used, or, if that’s not possible, by involving people from that environment directly in the design process. Eliminating time spent on “imagined” problems not only makes the technology development process faster, it also increases the likelihood the product will be well-received.

3. Expectation gap between implementers and donors.

Building on the previous two points, the chat’s participants made it clear that implementers don’t always do a good job of educating donors and other stakeholders about the realities “on the ground”. (Perhaps this is because these “doers” don’t fully understand them, themselves!) This misunderstanding inevitably leads to faulty expectations not only of the project, but also the processes of designing and implementing. Technologists must become better at “speaking donor”, and also must help the donors learn a little about how to “speak tech”.

This communication gap has plagued the informatics world in the global north since its beginnings, so it’s not surprising to see it at play in the global south in international development projects. However, I believe the remedy may be the same in both situations – increased cross-training of people with solid technical backgrounds in concepts like interpersonal communication, project management, monitoring and evaluation.


While failures — particularly in ICT4D — will never be eliminated, focusing on these factors earlier in projects can help reduce of the impact of these failures, and help us “fail early and often”, iteratively improving project implementations instead of failing late in the game, wasting more donor funds and invaluable time.

What do you think? Do you agree with these ideas? Were some overlooked? Voice your opinion in the comments.

Other FailDay Chat Synopsis

Keep current – subscribe to ICTworks updates via RSS, Email, or Twitter


Filed Under: Management
More About: , , , ,

Written by
Michael Downey is a graduate student and researcher in human-computer interaction at the Indiana University School of Informatics, and a member of the OpenMRS project team. His research is focused on adoption of technology in the global south and usability of open-source software.
Stay Current with ICTworksGet Regular Updates via Email

2 Comments to “Lessons From Failure: ICT4D Twitter Chat Synopsis”

  1. downeym says:

    As a postscript to the ideas above, anyone interested in learning more about would be well-served to read the short article “Encountering Development Ethnographically” by Nithya Sambasivan et al. in the November/December 2009 edition of Interactions Magazine:

    Sambasivan, N., Rangaswamy, N., Toyama, K., Nardi, B. (2009). Encountering Development Ethnographically. Interactions, 16(6), 20-23.

  2. Great Summary! Sorry I couldn’t make it on the day.

    You say: “help us “fail early and often”, iteratively improving project implementations instead of failing late in the game” – this really resonates with me.

    The best model for software (or any technology I’d argue) engineering is using iterative, agile methodologies. This matches well with the idea of (sadly not often practiced) participatory international development.

    While it’s easy to think about this on the project scale (i.e. a specific application or local intervention) the thinking can be lost on the programme or organisational scale.

    A few more thoughts on the subject at a project scale: