Crowd Wisdom

I just posted the article about Ushahidi and its future challenges that was published in the Index on Censorship last month (‘Crowd Wisdom’ by Heather Ford in Index on Censorship December 2012, vol. 41, no. 4 33-39 doi: 10.1177/0306422012465800) . I wrote about Ushahidi’s emergence as a powerful tool used in countries around the world to document elections, disasters and food – among others – and the coming challenges as the majority of Ushahidi implementations remain ‘small data’ projects and as tools move towards automatic verification, something only possible with ‘Big Data’.

Why the muggle doesn’t like the term “bounded crowdsourcing”

Patrick Meier just wrote a post explaining why the term he coined, “bounded crowdsourcing” is ‘important for crisis mapping and beyond’. He likens “bounded crowdsourcing” to “snowball sampling”, where a few trusted individuals invite other individuals who they ‘fully trust and can vouch for… And so on and so forth at an exponential rate if desired’.

I like the idea of trusted networks of people working together (actually, it seems that this technique has been used for decades in the activism community) but I have some problems with the term that has been “coined”. I guess I will be called a “muggle” but I am willing to take the plunge because a) I have never been called a “muggle” and I would like to know what it feels like and b) the “crowdsourcing” term is one I feel is worthy of a duel.

Firstly, I don’t agree with the way that Meier likens “crowdsourcing” work like Ushahidi to statistical methods. I see why he’s trying to make the comparison (to prove crowdsourcing’s value, perhaps?) but I think that it is inaccurate and actually de-values the work involved in building an Ushahidi instance. Working on an Ushahidi deployment is not the same as answering a question through statistical methods. With statistical methods, a researcher (or group of researchers) tries to answer a question or test a hypothesis. ‘Do the majority of Hispanic Americans want Obama to win a second term?’ for example. Or ‘What do Kenyans think is the best place to go on holiday?’

But Ushahidi has never been about gaining a statistically significant understanding of a question or hypothesis. It has been designed as a way for a group of concerned citizens to provide a platform for people to report on what was happening to them or around them. Sure, in many cases, we can get a general feel about the mood of a place by looking at reports, but the lack of a single question (and the power differential between those asking and those being asked), the prevalence of unstructured reports and the skewed distribution of reporters towards those most likely to reply using the technology (or attempting to game the system) make the differences much greater than the similarities.

The other problem is that the term lacks a useful definition. Meier seems to suggest that the “bounded” part refers to the fact that the work is not completely open and is limited to a network of trusted individuals. More useful would be to understand under what conditions and for what types of work different levels of openness are useful, because no crowdsourcing project is entirely “unbounded”. Meier says that he ‘introduced the concept of bounded crowdsourcing to the field of crisis mapping in response to concerns over the reliability of crowd sourced information.’ But if this means that “crowdsourced” information is unreliable, then it would be useful to understand how and when it is unreliable.

If we take the very diverse types of work required of an Ushahidi deployment, we might say that they include the need to customize the design, build the channels (sms short codes, twitter hashtags, etc), designate the themes, advertise the map, curate the reports, verify the reports, find related media reports, among others. Once we’ve broken down the different types of work, we can then decide what level of openness is required for each of these job types. I certainly don’t want to restrict the advertising of my map to the world, so I want to keep that as “unbounded” as possible. I want to ensure that there are enough people with some “ownership” of the map to keep them supporting and talking about it, so I want to give them some jobs that keep them involved. Tagging reports as “verified” is probably a more sensitive activity because it requires a set of transparent rulesets and is one of the key ways that others come to trust the map or not. So I want to ensure that trusted people, or at least those over whom I have some recourse, do this type of work. I also want to get feedback on themes and hashtags to keep it close to the people, since in the end, a map is only as good as the network that supports it. Now if I have different levels of openness for different areas of work, is my project an example of “bounded” or “unbounded” crowdsourcing?

Although I am always in favor of adding new words to the English language, I feel that the term “unbounded crowdsourcing” is unhelpful in leading us towards any greater understanding of the nuances of online work like this. Actually, I’m always surprised at the use of the term “crowdsourcing” over “peer production” in the crisis mapping community since crowdsourcing implies monetary or commercial incentivized work rather than the non-monetary incentives that characterised peer production projects like Wikipedia (see an expanded definition + examples here). I can’t imagine anyone ever “coining” the term “unbounded peer production” (but I seem to be continually surprised, so I should completely discount it from happening) and I think that this is indicative of the problems with the term.

So, yes, if we’re talking about different ways of improving the reliability of information produced on the Ushahidi platform, I’m excited to learn more about using trusted networks. I just think that if a term is being coined, it should be one that advances our understanding of what the theory is here. Is it that: if you restrict the numbers of people who can take part in writing reports, you get a more reliable result? Where do you restrict? What kind of work should be open? What do we mean by open? Automatic acceptance of Twitter reports with a certain hashtag? Or an email address that you can use to request membership? Is there a certain number that you should limit a team to (as the Skype example suggests)?

This “muggle” thinks that the term doesn’t get us any further towards understanding these (really important) questions. The “muggle” will now squeeze her eyes shut and duck.

What is the next step for Ushahidi verification?

Cross-posted from

As Ushahidi ethnographer, my job is to do on-the-ground research on users’ experience with our technology in particular contexts. Something that we’ve been thinking about a great deal as we develop SwiftRiver is the process of verification, the ways in which technology and society work together to create useful, trustworthy and actionable information, as well as where the technology in particular contexts might be failing.

With over 20,000 installations of Ushahidi and Crowdmap since January, 2009, Ushahidi has been used in a number of different contexts – from earthquake support in Haiti, to reports of sexism in Egypt, to election monitoring in the Sudan. In each of these cases, a map is publicized and individuals are encouraged to send reports to it. The process of verifying information reported by the crowd has taken on a variety of different forms depending on the needs and affordances of the environment and the community supporting it.

The memo I just published on scribd introduces the concept of verification, how it has evolved at Ushahidi and in sample deployments, alternative ways of thinking about verification and some suggestions for further research. Its goal is to inform developers and designers as they develop the next generation of Ushahidi and SwiftRiver software to meet the needs of our users rather than prescribing what should be done.

Ushahidi support for verification has until now been limited to a fairly simple backend categorisation system by which administrators tag reports as “verified” or “unverified”. But this is proving unmanageable for large quantities of data and may not be the most effective way of portraying the nuanced levels of verification that can practically be achieved with crowdsourced data.

What research needs to be done to test verification alternatives so that Ushahidi and Crowdmap deployers are provided with due diligence tools that can advance trust in their deployments? Can we do this in a way that doesn’t add any new barriers to entry to those who need to have their voice heard on Ushahidi? How can we ensure that this solution is as close as possible to the needs, incentive systems and motivations of deployers and users? What is the next step for Ushahidi verification?

What should be remembered?

I’ve been thinking a lot about the disputes around Ushahidi’s role in humanitarian efforts and came round to thinking that we may be looking in the wrong place to discover the work that tools like Ushahidi’s Crowdmap are doing in the world. Whereas humanitarian organisations are asking (good) questions about whether Ushahidi’s tools help or hinder their efforts, another way to look at it might be to look from the perspective of the people making the maps and reports themselves. What work is Ushahidi doing for them? How do they see Ushahidi’s effectiveness? What social role does reporting play and how could we begin to measure effectiveness?

This morning I read a wonderful article by Tamar Ashuri from Ben-Gurion University for an upcoming edition of the journal New Media and Society entitled ‘(Web)sites of memory and the rise of moral mnemonic agents‘. Ashuri looked at how two websites set up by Israelis – one to monitor human rights of Palestinians at Israeli checkpoints; the other to collect testimonies of Israeli soldiers who served in the Occupied Territories – act as agents of collective memory. Ashuri argues that digital networked technologies is challenging the mechanisms that society employs to deny memories of immoral acts and how the online archives created by moral witnesses become a space of living memory and a sphere of moral engagement.

Ashuri explains that ‘collective memory is a social necessity; neither an individual nor a society can do without it.’ She quotes from Kansteiner (2002: 180) to describe how collective memory is different from history:

Collective memory is not history, though sometimes made from similar material […]. It can take hold of historically and socially remote events but it often privileges the interest of the contemporary. It is as much a result of conscious manipulation as unconscious absorption and it is always mediated.

Ashuri describes how Avishai Margalit distinguishes between “common memory” (a group of people who recall a certain episode that each of them experienced) and “shared memory” (which requires communication). Shared memory is is not just an aggregate of individual memories because it requires those who remember the episode to come together to create one version (or at least a few version) through an active presentation and retelling of a story that Margalit terms ‘a division of mnemonic labor’ (2002: 52). Margalit wrote that whereas in traditional society there was a direct line from the people to their priest, storyteller or shaman, shared memory in modern society ‘travels from person to person through institutions, such as archives, and through communal mnemonic devices, such as monuments and the names of streets’ (2002: 52). Ushuri posits a new term “joint memory” to describe a new type of memory that is a ‘compilation of personal histories made public for the public’ (2011: 4). She argues that digital networked technology is challenging the exclusive role of professional mnemonic agents designated by the church, state, monarchy etc.

Significantly, joint memory is not motivated by personal interests – the desire to tell an interesting story or reveal new information – but is driven by a social purpose: Witnesses who add their recollections to an accessible and shareable compilation of memories attempt to expose events that the default collective (such as the nation) denies or wishes to forget.

I don’t agree that such reporting is not motivated at all by personal interests but I do agree with the fact that the social/moral purpose of witnessing is really critical here. Ashuri builds on Margalit’s conception of a moral witness whose testimony is ‘essentially driven by a moral purpose. It reflects hope for the witness to be a social agent who, in testifying to his or her harsh experience, transforms (passive) addressees into active audiences’. She says that what is happening now is slightly different because the moral witness now performs memories of suffering experienced in a public space.

In my conceptualisation, the (moral) mnemonic agent is the one who recalls his or her memories regarding events in whcih others have suffered and by that act of witnessing renders this suffering visible and hence difficult to marginalize or deny. The moral aspect of this act, in my estimation, derives from the content of the mnemonic text (testimony about suffering inflicted by evil) and from its objective (calling on the audience to shed their observer garb and re-enact the experience of the harsh realities). (p5)

I think this is really useful way to think about how websites like Ushahidi as well as engagement on social media sites like Twitter are acting as platforms for this kind of performance and the communication of suffering, and how this is one way of looking at how collective narratives about the world are being wrested from those who traditionally controlled this in the Middle East. Whether it is reporting on human rights violations in Saudi Arabia, harassment of women in Egypt on Harassmap  or reports of arrests and casualties in Syria I think that looking at the maps through the lens of moral witnessing and Ashuri’s “joint memory” could be a wonderful entry point for re-thinking Ushahidi’s role and effectiveness in the world.