Web based election monitoring in Liberia: a failure

Usahidi LiberiaFor the past three days I have been following the coverage of the Liberia elections on liberia2011.ushahidi.com (the elections were held on 11 October). Unfortunately, I’m far from impressed. To be clear: this is not the fault of Ushahidi: After all, Ushahidi is just the technical platform and it is the responsibility of others to feed the system information, but it shows the limitations of crowdsourcing information.

In total, only 23 reports were submitted to the web platform on election day for all of Liberia. Many of these reports were of dubious news value such as “Not many voters left at Nancy Doe Market” or “Voters have already started arriving @E J Good Ridge High School waiting to cast their vote”. I would say that only seven of the 23 reports had any news value at all – but that might be a matter of personal opinion.

Misleading or fake reports?

The bigger issue is that a large number of reports were automatically posted on 11 October at 00:00 by the Elections Coordinating Committee (see an example here). All of them were marked as “verified” and included lines like: “Did the polling place open on time (08. 00 am): Yes” and “Did the counting start after polling closed: Yes”.

Obviously these reports are wrong: either, they really were published before the polls opened, in which case they are completely fabrications, or the posts were backdated, which is a serious mistake. In either case, it is confusing, hurts the credibility of the whole monitoring exercise and might even give rise to allegations of manipulation. For a project like this, that is a disaster. (Update: See response from Ushahidi in the comments section)

Finally, I noticed at least one report that was shown in a completely wrong location on the map, which ain’t great for a mapping project.

Where are the results?

On October 12, only two reports were posted to the platform. This shows how thin the network of contributors really is. While the results of many polling stations had already been posted on the doors of the local police stations, none of this information made it onto the web platform. Obviously, there were not enough monitors in the field to report that information.

No infrastructure, no crowd, no crowdsourcing

I had been very curious to see, how well the Ushahidi platform would work in a country with as limited an infrastructure as Liberia. Unfortunately the answer is: it doesn’t work.

The success of any crowdsourcing initiative depends on the size of the crowd. As I mentioned before, many Liberians don’t have mobile phones and even those who have one, frequently don’t have credit on the phone or the electricity to charge it, or they are living in one of the many areas which have no mobile phone reception. Of the remaining people, I doubt that many were even aware of the monitoring initiative.

Internet access is even rarer and for many people the concept of a web based monitoring platform must be something terribly abstract and not very relevant to their lives. All of this limits the size of the crowd almost exclusively to the nine partner organizations that were supposed to feed information to the platform. Some of these organizations, like UNMIL, would certainly have been able to contribute something of value. But in the end they didn’t – UNMIL for example did contribute a single report.

Last but not least, the low quality of maps of Liberia certainly posed an additional challenge for Ushahidi. Many villages, and even towns, simply cannot be found on Google Maps and most places, and even districts, can be spelled three or four different ways.

Potential versus reality

Don’t get me wrong: the potential for crowdsourced tools like Ushahidi is enormous. But in order to fulfill that potential, we have to take a critical look at what the problems are. Any database is only as good as the information that you put into it and in Liberia neither the quality nor the quantity were good enough. Admittedly, my perception would probably be slightly different if the polling station reports I mentioned above had been published after the polls closed and not before they even opened. But these things are important and those reports were even published as “verified”. However, none of that shouldn’t stop us from trying to do it better in 2017.

For the time being, let’s enjoy that everything been peaceful and hope that everything will remain calm.


  • Here are a few clarifications regarding your post from the Ushahidi Liberia team:

    -The majority of the reports on this site are contributed by Ushahidi Liberia’s partner organizations; each organization has their own protocols and metrics for report verification and publication

    -It has been Ushahidi Liberia’s intention from the beginning not to crowdsource information regarding the electoral process. Given the potential for unreliable information and volatility from the crowd in Liberia, we have intentionally sought information from partner organizations – a crowdseeding approach rather than crowdsourcing. This is a different approach than is often taken in other deployments of the Ushahidi platform. Clarification about this is now provided here on the elections instance; your blog post reminded us that this was not apparent on the instance, and we thank you for clarifying the assumption that the instance was crowdsourcing

    -In many cases, reports’ quality and contents reflect how much training reporters have received from their parent organizations on sending information to the platform. Sometimes the Ushahidi Liberia team is invited to conduct trainings for reporters, at other times they are not; it is the choice of the partner organization with which our team works

    -As for the number of reports shared on the map during election day, we received an additional 126 messages that are currently being verified and approved on the backend, so the numbers of those published do not reflect the number received by our partners.

    -Ushahidi Liberia’s Tech Lead oversimplified the date/time when he bulk uploaded the ECC polling data. The data is accurate but in the upload our Lead did not list the exact time each was uploaded at the ECC data center; we apologize for this mistake and the corrections can now be found on the map. All former ECC polls reports were deleted and replaced with the same reports and the exact times they were entered into the ECC database.

    -We would be curious to know which report you have found with the wrong location. The mapping process for many messages is a manual process, leaving room for human error; we will be happy to correct the location when you provide us with the details

    -Regarding results, the elections instance never claimed that it would post this data. What is posted on this instance is the result of our partner organizations’ priorities and scope; none of these partners intended to post results to the map and the information is otherwise unavailable at this time. This website provides the only existing preliminary results, and is admittedly unofficial: http://liberiamediacenter.smagmedia.com.lr/LMC/. After the NEC’s prelims announcement today, the LMC website will display this NEC data alongside LMC’s own. In addition, the Ushahidi platform is designed to display discrete data points, it is not well suited to displaying summations, averages or other forms of numerical analysis. Thus it does not lend itself to displaying vote results. When it comes to monitors in the field, there are thousands of monitors gathering results; if anything, it is simply too early to determine the results.

    -We agree with many of the challenges you have listed for using the Ushahidi platform in a setting like Liberia. Please refer to the following blog posts we’ve written regarding these problems:
    — Liberianizing the Platform: http://blog.ushahidi.com/index.php/2010/07/30/lib
    — Getting better data on Google Maps: http://blog.ushahidi.com/index.php/2010/09/03/cho
    — Lessons Learned by end of 2010: http://blog.ushahidi.com/index.php/2010/12/07/lib
    — Launching the elections instance with new features: http://blog.ushahidi.com/index.php/2011/02/14/gea
    — Patrick Meier on wrong assumptions regarding technology use in places like Liberia: http://blog.ushahidi.com/index.php/2011/06/28/wro
    — Meeting with community crime watch groups to improve use of platform: http://blog.ushahidi.com/index.php/2011/08/18/whe
    — Trying to make SMS gateway work amid Liberia’s limitations: http://blog.ushahidi.com/index.php/2011/09/14/sms
    — Lessons from recent local trainings: http://blog.ushahidi.com/index.php/2011/10/03/old

    As you suggest, the challenges of implementing this tool in Liberia are significant and not to be underestimated or ignored.

    -As for UNMIL’s blank group page, that is because they wanted for us to create a link to their Facebook and website pages for them but have not been officially sharing data with them. We have since moved their links to the “election info” page on the instance and have removed their group page.

    I hope that these explanations clarify some of your questions and concerns. We would be happy to continue this conversation with you in person if you’d like.

    • Thank you very much for your response, Kate. I really appreciate it and your response has managed to correct a few things that I had misunderstood.

      I’m particularly glad to hear that the 00:00-issue was a technical issue and nothing worse!

      I think the main problem is that by setting up an Ushahidi site for Liberia, you were creating expectations that the quality of the information on the site would be similar to what people have seen in other contexts. In a way that is a lopsided compliment. However, I don’t think that elections in a country with as many technological and infrastructure problems as Liberia, can be realistically monitored with this particular tool which is not to say it can’t be the right tool in other contexts. As I said above: Let’s see whether it will work better in 2017.

      I’ll get back to you by email regarding the wrong location.

  • Pingback: Addressing concerns about Liberia’s election instance – The Ushahidi Blog()

  • Interesting and valid. The confusion around crowdsourcing report during an election and systematic election monitoring has driven me wild. More on that here http://mobileactive.org/q-ian-schuler-election-mo… and specifically, why citizen reporting is not election monitoring: http://mobileactive.org/cutting-through-hype-why-….

    Unfortunately, the good folks at Ushahidi and in the media have not differentiated and made clear that confusing one with the other is not wise and can even be dangerous. I hope that your post is moving the needle in a more clear and precise differentiation of the two methods of incidence reporting in elections, and the utility and shortcomings of either.


    Katrin, MobileActive.org

  • Interesting post on the use of technology in the 2011 elections in Liberia http://mobileactive.org/technology-2011-liberian-