Time-Critical Crowdsourcing for Social Mobilization and Crowd-Solving

My good friend Riley Crane just co-authored a very interesting study entitled “Time-Critical Social Mobilization” in the peer-reviewed journal Science. Riley spearheaded the team at MIT that won the DARPA Red Balloon competition last year. His team found the locations of all 10 weather balloons hidden around the continental US in under 9 hours. While we were already discussing alternative approaches to crowdsourcing for social impact before the competition, the approach he designed to win the competition certainly gave us a whole lot more to talk about given the work I’d been doing on crowd sourcing crisis information and near real-time crisis mapping.

Crowd-solving non-trivial problems in quasi real-time poses two important challenges. A very large number of participants is typically required couple with extremely fast execution. Another common challenge is the need for some sort of search process. “For example, search may be conducted by members of the mobilized community for survivors after a natural disaster.” Recruiting large numbers of participants, however, requires that individuals be motivated to actually conduct the search and participate in the information diffusion. Clearly, “providing appropriate incentives is a key challenge in social mobilization.”

This explains the rationale behind DARPA decision to launch their Red Balloon Challenge: “to explore the roles the Internet and social networking play in the timely communication, wide-area team-building, and urgent mobilization required to solve broad-scope, time-critical problems.” So 10 red weather balloons were discretely placed at different locations in the continental US. A senior analyst at the National Geospatial-Intelligence Agency is said to have characterized the challenge is impossible for conventional intelligence-gathering methods. Riley’s team found all 10 balloons in 8 hours and 36 minutes. How did they do it?

Some 36 hours before the start of the challenge, the team at MIT had already recruited over 4,000 participants using a “recursive incentive mechanism.” They used the $40,000 prize money that would be awarded by the winners of the challenge as a “financial incentive structure rewarding not only the people who correctly located the balloons but also those connecting the finder [back to the MIT team].” If Riley and colleagues won:

we would allocate $4000 in prize money to each of the 10 balloons. We promised $2000 per balloon to the first person to send in the cor- rect balloon coordinates. We promised $1000 to the person who invited that balloon finder onto the team, $500 to whoever invited the in- viter, $250 to whoever invited that person, and so on. The underlying structure of the “recursive incentive” was that whenever a person received prize money for any reason, the person who in- vited them would also receive money equal to half that awarded to their invitee

In other words, the reward offers by Team MIT “scales with the size of the entire recruitment tree (because larger trees are more likely to succeed), rather than depending solely on the immediate recruited friends.” What is stunning about Riley et al.’s approach is that their “attrition rate” was almost half the rate of other comparable social network experiments. In other words, participants in the MIT recruitment tree were about twice as likely to “play the game” so-to-speak rather than give up. In addition, the number recruited by each individual followed a power law distribution, which suggests a possible tipping point dynamic.

In conclusion, the mechanism devised by the winning team “simultaneously provides incentives for participation and for recruiting more individuals to the cause.” So what insights does this study provide vis-a-vis live crisis mapping initiatives that are volunteer-based, like those spearheaded by the Standby Volunteer Task Force (SBTF) and the Humanitarian OpenStreetMap (HOT) communities? While these networks don’t have any funding to pay volunteers (this would go against the spirit of volunteerism in any case), I think a number of insights can nevertheless be drawn.

In the volunteer sector, the “currency of exchange” is credit. That is, the knowledge and acknowledgement that I participated in the Libya Crisis Map to support the UN’s humanitarian operations, for example. I recently introduced SBTF “deployment badges” to serve in part the public acknowledgment incentive. SBTF volunteers can now add badges for deployments there were engaged in, e.g., “Sudan 2011″; “New Zealand 2011″, etc.

What about using a recursive credit mechanism? For example, it would be ideal if volunteers could find out how a given report they worked on was ultimately used by a humanitarian colleague monitoring a live map. Using the Red Balloon analogy, the person who finds the balloon should be able to reward all those in her “recruitment tree” or in our case “SBTF network”. Lets say Helena works for the UN and used the Libya Crisis Map whilst in Tripoli. She finds an important report on the map and shares this with her colleagues on the Tunisian border who decide to take some kind of action as a result. Now lets say this report came from a tweet that Chrissy in the Media Monitoring Team found while volunteering on the deployment. She shared the tweet with Jess in the GPS Team who found the coordinates for the location referred to in that tweet. Melissa then added this to the live map being monitored by the UN. Wouldn’t be be ideal if each could be sent an email letting them know about Helena’s response? I realize this isn’t trivial to implement but what would have to be in place to make something like this actually happen? Any thoughts?

On the recruitment side, we haven’t really done anything explicitly to incentivize current volunteers to recruit additional volunteers. Could we incentivize this beyond giving credit? Perhaps we could design a game-like point system? Or a fun ranking system with different titles assigned according to the number of volunteers recruited? Another thought would be to simply ask existing volunteers to recruit one or two additional volunteers every year. We currently have about 700 volunteers in the SBTF, so this might be one way to increase substantially in size.

I’m not sure what type of mechanism we could devise to simultaneously provide incentives for participation and recruitment. Perhaps those incentives already exist, in the sense that the SBTF response to international crises, which perhaps serves as a sufficient draw. I’d love to hear what iRevolution readers think, especially if you have good ideas that we could realistically implement!

12 responses to “Time-Critical Crowdsourcing for Social Mobilization and Crowd-Solving

  1. Pingback: Time-Critical Social Mobilization « GEODATA POLICY

  2. I think it might be interesting to use the pyramid scheme to incentivize training as well. Sort of a buddy system for those starting out as well. Similar to a Stackoverflow points system, meaning people get badges for answer questions. This already exists in OpenStreetMap (http://help.openstreetmap.org/), but not specifically to CrisisMapping it could help the already existing volunteers be more effective.

    I think the concern about going against the spirit of volunteerism is interesting. Where it is true individuals don’t get paid for their time digitizing, geolocating, translating, sometimes these activities do lead to paid jobs, do they not? I’m not suggesting directly paying people, I think funding lacks for that and also potentially increases the changes of gaming the system. Though maybe there is something that can be done beyond “digital candy.” Perhaps partnership with an online bookseller or something similar. Though would that change the entire dynamic of the teams already working.

    • Thanks Kate, very good points, eg, volunteerism leading to jobs. I’ve written several letters of recommendation for outstanding SBTF volunteers, for example. Interesting thought re online bookseller!

  3. This is a great topic and definitely very important. I think it would be helpful if incentives were aligned to individual needs and interests instead of a broad-based approach. It might be helpful to directly engage in conversation with the SBTF volunteers, through a survey perhaps, and directly ask them what they are looking for. Questions like why they joined in the first place, what motivates them, how they expected to benefit/ be rewarded, are their expectations being met, etc. The answers might not be similar and identifying these different individual expectations and priorities might help design an array of incentives catering to different kinds of volunteer needs. Tedious process but worth a try maybe?

  4. Hi Patrick, after reshuffling my email inbox I discovered I hadn’t yet read this one. The challenge fascinated me in how the team achieved the miraculous time so the article was quite enlightening!
    Certainly a groundbreaking example for social media utilisation but i guess that depends on how they spread the message not so much the reward model
    My immediate thought was more to how such a system could be utilised for catching criminals, identifying situations etc where rewards are posted.
    Some security concerns, but a model might be devised.
    Cheers Dave

  5. Pingback: Some Thoughts on Real-Time Awareness for Tech@State | iRevolution

  6. Pingback: Six Degrees of Separation: Implications for Verifying Social Media | iRevolution

  7. Pingback: Verily: Crowdsourcing Evidence During Disasters | iRevolution

  8. “recursive incentive mechanism” and enrollment for benefits on debit cards – a great future :)

  9. Pingback: Veri.ly, la verifica dei social network per i disastri naturali | Piazza Digitale

  10. Pingback: Crowdsourcing Critical Thinking to Verify Social Media During Crises | iRevolution

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s