This Wikipedia article presents a summary of the 2004 book of the same name by New York Journalist James Surowiecki. Initially it explains Surowiecki's ideas about how crowd intelligence works and his musings on why crowd intelligence sometimes fails, but the part I find most interesting and most relevant to my essay topic on mobile crowd-sourcing is the section dealing with the use of crowd intelligence to make predictions about the future.

Previous articles I've read about crowd-sourcing describe applications that seek to provide more insight about present conditions (for example, noise-mapping, road condition monitoring and invasive plant tracking) but this article states that Surowiecki's book focuses entirely on the predictive applications of crowd-sourcing. The betting market and the stock market are described as arenas where crowd intelligence is used to make predictions. The article goes on to state "A number of Web-based quasi-prediction marketplace companies have sprung up to offer predictions primarily on sporting events and stock markets but also other topics. Those companies include Piqqem, Cake Financial, Covestor, Predictify and the Motley Fool". Whilst it isn't actually stated, the suggestion is that these companies all use some form of crowd-sourcing to make their predictions.

The accurate prediction of future events is, in my opinion, an area of incredible potential. If mobile devices can effectively facilitate crowd-sourcing for predictions, and those predictions are proven highly accurate, one can imagine it having a massive influence on government, business, military and environmental sustainability. This is an application of crowd-sourcing I'm keen to explore further.

The Wisdom of Crowds (2012). In Wikipedia, The Free Encyclopedia. Retrieved 1 April, from http://www.icepredict.com/rsrc/files/wisdomofcrowds.pdf

This article discusses the usefulness of Twitter as a tool for research. Researcher Pablo Garaizar suggests that monitoring large volumes of tweets and identifying trends in what users are saying - a technique called data mining - can produce valuable results for researchers conducting socio-behavioural research. He describes (in agonising detail) the web application iScience Maps, which his team developed to expedite the process of data mining Twitter, before moving on to describe a study he conducted to test the effectiveness of the technique.

Garaizar describes his attempt to replicate the findings of a study run in 1993 which apparently demonstrated that the names Alexander, Charles and Kenneth are associated with ambition, intelligence and creativity, whilst the names Otis, Tyrone and Wilbur are not. Garaizar poses "If these names' having the connotation of a personality characteristic really holds, this likely should be apparent when Twitter is mined, because attributions to persons, such as 'Charles is an intelligent guy,' frequently appear in text-based message services like Twitter." So, using iScience Maps, he searched Twitter for one name at a time, in combination with the terms 'ambitious', 'intelligent' and 'creative'. His findings supported the findings of the 1993 study: Alexander, Charles and Kenneth were indeed found in conjunction with the positive terms multiple times, while the other three names were never found in conjunction with those terms.

Bizarre as Garaizar's example may be, the fact that tools like iScience Maps can provide insight into human psychology is incredible. The ability to examine the communications of thousands, potentially millions, of users and easily identify trends (both temporal and geographical) is something that will likely revolutionise research in the social sciences. One can imagine marketing agencies finding this type of insight into the minds of the masses very insightful too. In my opinion, mining Twitter is crowd-sourcing at its very purest - it's just a shame the word limit on this assessment prevents me from exploring that. My essay for module two probably will.

http://www.csee.usf.edu/~labrador/Share/PapersToRead/GameTheory/Incentive%20participation.pdf

(I'm doing mobile phone crowd-sourcing)

Beyond the technological challenges that face participatory sensing (a term that describes crowd-sourcing reliant on the 'sensors' in smartphones rather than thoughtfully created content) is a more fundamental human dilemma: how to motivate the required number of people to participate. Regardless of the application, participatory sensing (and any type of crowd-sourcing for that matter) requires large numbers of participants in order to deliver quality data, so providing an incentive for those people and minimising the dropout rate is crucial.

Authors Lee and Hoh explain that participatory sensing applications often suffer from low participation rates because participants soon lose interest when they're not rewarded for their efforts. To address this, they propose a new incentive mechanism called Reverse Auction based Dynamic Price (RADP) which "focuses on minimising and stabilising incentive cost while maintaining adequate number of participants by preventing users from dropping out of participatory sensing applications".

For readers lacking a degree in economics, the article becomes rather less accessible as it presents a series of complicated formulas, but essentially the idea behind RADP is simple: participants can offer to sell their sensing data to service providers at a price they choose (called their bid), and the service provider selects multiple participants who have offered the lowest bids. Of course, that means the higher bidders are excluded, but they are given virtual credit just for their participation, which aims to maintain a constant number of active bidders.

The second half of the article presents the findings of a study that compared their model to existing ones, with the conclusion that it reduces incentive cost by more than 60% and maintains the desired number of participants.

This model is universal enough to be applied to all types of participatory sensing and crowd-sourcing applications, including the weather and noisemap apps used as case studies in my chosen article. Indeed, the authors of that article acknowledged that an incentive model is a question that remains to be answered regarding the viability of their vision of mobile crowd-sourcing, so it would be interesting to apply the RADP model to that application.

Lee, J. and Hoh, B. (2010) Sell your experiences: A market mechanism based incentive for participatory sensing. Pervasive Computing and Communications (PerCom), 2010 IEEE International Conference on, pages 60-68: USA

http://techcrunch.com/2011/04/12/txteagle-raises-8-5-million/

(I'm doing mobile phone crowd-sourcing)

While most people would associate mobile crowd-sourcing with the developed world (after all, participants need to afford a mobile and the costs that go with it), in this article TechCrunch author Jon Evans highlights the work of Boston-based TxtEagle - a company that crowd-sources data from thousands of participants in developing countries.

TxtEagle is founded on the idea that while developing countries don't yet have a high penetration of smartphones, they do have more than 2 billion people who still use "the plain old GSM phone". Of course, these phones don't have capacity for the more advanced types of mobile crowd-sourcing like the Twitter-based examples in my chosen article, but they do allow participants to complete surveys via a protocol called USSD, which is similar to SMS, but free.

As Evans explains, the business model is simple: advertisers and government researchers hire TxtEagle to survey masses of people; TxtEagle then forwards the survey to their participants' GSM phones, and pays them upon completion. But instead of paying cash, TxtEagle pays them in phone credit, which is bought from the service providers in bulk. As Evans explains, "in the prepaid world, i.e. most of the planet, mobile airtime is becoming a currency as desirable as, and nearly as convertible as, old-fashioned cash".

This is an incentive model that contrasts with the reverse auction model proposed by Lee and Hoh (Lee and Hoh 2009) and one that could be seen as a better fit for mobile crowd-sourcing in developing nations. With the participants in this example earning so little, having to outbid one another as per the reverse auction model would likely lead to rather demeaning scenes by western standards, where the minimum bids would have to be incredibly low in order to be competitive. It is evidence that when we take crowd-sourcing beyond the western world, different incentive models are probably needed.

Evans, J. (2011) TxtEagle Raises $8.5 Million To Give 2.1 Billion a Voice Retrieved 20 March 2012 TechCrunch: USA

http://research.microsoft.com/pubs/78568/Nericell-Sensys2008.pdf

(I'm doing mobile phone crowd-sourcing)

Written by three researchers from Microsoft Research India, this article explores the idea of using smartphones to monitor road conditions such as potholes, bumps and chaotic traffic conditions. The authors speculate that systems that "piggyback" on smartphone technology could replace the need for traditional traffic monitoring infrastructure, which is expensive, especially for developing countries. The article presents an evaluation of one such system they developed called Nericell, based on experiments they conducted on the streets of Bangladore.

The article explains that Nericell differs to previous systems, which could only monitor speed and location information. Nericell instead uses a phone's microphone, GPS and accelerometer, "to glean rich information, e.g. the quality of the road or the noisiness of traffic".

Like my chosen article, this one states openly that providing incentives for participants is an important issue that is yet to be explored for this technology. Instead, the article focuses on the technical aspects of harnessing a phone's GPS, microphone and accelerometer to give an accurate depiction of the road conditions.

While this rather technical article provides the least insight of the five articles in terms of the cultural aspects of mobile crowd-sourcing, I chose it because it describes a use of crowd-sourcing which is a) arguably the most widely applicable in a global sense, and b) likely to be one of the best-supported of its kind. Both are owing to the ubiquity of roads and motorists.

Mohan, P., Padmanabhan, V. and Ramjee, R. (2008) Nericell: Rich Monitoring of Road and Traffic Conditions using Mobile Smartphones. SenSys '08 Proceedings of the 6th ACM conference on Embedded network sensor systems: India

http://www.nxtbook.com/nxtbooks/imagine/20110506_DMI/index.php?startid=10

(I'm doing mobile phone crowd-sourcing)

This article describes a suite of participatory sensing apps developed at UCLA that allow participants to upload data about their surroundings to a central database.

Written in a popularist style for a lay audience, and unashamedly promoting the UCLA apps, this article could be seen as evidence that research in the field of mobile crowd-sourcing, which in the past has been the stuff of academic papers, is beginning to manifest as something seductive and consumable.

One app described in the article is What's Invasive! which allows hikers in the Santa Monica Mountains Park to take photos of plants they identify as invasive, tag the photo with the name of the plant, and upload it to a server that displays the data gathered by all participants. Another example given is What's Noisy! which uses the phone's microphone to gather information about noise levels, in much the same way as the noisemap app described in my chosen article. These apps, as well as the others in the suite, are available for anyone to download from the Android Marketplace.

Interestingly, the article implies that the only motivation for users of these apps is an altruistic one - the furthering of science. In other words, the participants must have a shared motivation. In the case of What's Invasive! they require some scientific knowledge too, in order to correctly identify invasive plants. These requirements, which inevitably limit participation to a select group, are at odds with mobile crowd-sourcing examples in all other articles I've read, which favour quantity over quality and rarely require much specialised knowledge on behalf of the user.

Rather than conclude that the UCLA apps aren't proper examples of mobile crowd-sourcing, this article has led me to question what parameters need to be placed around issues of quantity and quality in mobile crowd-sourcing - something that could warrant further analysis in the next assignment.

Kim, K. (2011) Become a Citizen Scientist. Imagine Magazine by John Hopkins Center for Talented Youth pages 10-13: USA