The 2013 paper 'eScience in the Social Cloud' proposed SoCPR, a platform that recruits volunteer computing participants through a social network by coupling a Facebook front end with BOINC (Berkeley Open Infrastructure for Network Computing) project servers, and introduced three gamified social incentive roles — Project Champion, Social Anchor, and Compute Magnate. Each incentive was evaluated only in isolation, via simulations on a 6,000-person social network dataset, and the paper's projection that recruiting about 1% of a large social network's users would double or triple the volunteer computing participant base was an unvalidated extrapolation. No later work has studied the three mechanisms operating simultaneously, and no publicly released or user-evaluated production deployment has validated the social-recruitment scale-up conjecture — the group's 2012 follow-up built and institutionally deployed a 'first' production version implementing all three roles, but it was never released publicly or evaluated with users. The problem — determine the joint effect of the three incentive mechanisms and test whether socially-networked, gamified recruitment actually multiplies volunteer participation at scale — therefore remains open.

Background

Volunteer computing is a form of distributed computing in which members of the public donate the idle processing power of their personal computers to scientific projects; BOINC (Berkeley Open Infrastructure for Network Computing) is the open-source middleware platform that hosts most such projects. Volunteer computing supplies large computing capacity cheaply, but it suffers from a chronic participation bottleneck: only a small fraction of potential volunteers ever join, and drop-out rates are high, a limitation repeatedly identified in surveys of the field. The 'social cloud' idea addresses such bottlenecks by exploiting pre-existing trust relationships in online social networks — a social cloud is a computing platform whose resources are shared among friends or contacts in a social network. In this line of work, the 2011–2013 papers of Bubendorfer, Chard, John, and Thaufeeg described two prototypes. One, SoCPR (Social Cloud for Public eResearch), recruits volunteer computing participants through Facebook: it integrates Facebook as a social front end with BOINC project servers and BOINC clients, suggests projects matched to user interests, and publishes gamified milestones to users' Facebook walls. To motivate participation and recruitment, SoCPR defines three social incentive roles, each computed from live BOINC credit data (BOINC's accounting of each user's contributed computation) and localized to each user's personal social circle: a Project Champion is the top contributor to a specific project by project credits (encouraging users to focus on smaller projects, where becoming champion is easier); a Social Anchor is the user with the highest social score, defined as Ssu=∑fSvufSs_{u} = \sum_{f} Sv_{uf} summed over the user's friends ff, where a friend's social value is Svu=1/(nf+1)Sv_{u} = 1/(n_{f}+1) with nfn_{f} the number of that friend's connections already on the platform, so recruitment of socially valuable, sparsely connected users is rewarded; and a Compute Magnate is the user with the highest compute score, an aggregate of friends' rolling 30-day BOINC credits, rewarding users who induce their friends to contribute more compute. The prototype was only a development version entering user testing, and its evaluation consisted of three separate simulations on a social network dataset of 6,000 individuals with 29,888 relationships, each studying one incentive in isolation: the Social Anchor incentive raised simulated membership from an initial state of 15 members to about 3,000 members after one simulated year, against about 500 in the without-incentive control (with a Poisson-distributed coercion factor — the number of friends each member attempts to recruit — of mean 3, and the initial membership of 15 set to match the ratio of the total BOINC user base, 2.2 million, to the Facebook user base, 800 million, i.e. 0.25 percent of the 6,000-person dataset); the Project Champion incentive shifted contribution from an initial 60/25/10/4/1 percent distribution across five projects toward equal allocation; and the Compute Magnate incentive raised total contribution from a 12 percent baseline (the observed active contribution rate in BOINC) to roughly 55 percent at a Poisson disinterest mean of 5 (the disinterest factor models users' reluctance to keep contributing), and lower at higher disinterest. On this basis the authors conjectured, by simple extrapolation of publicly cited user statistics, that recruiting 0.5 percent of Facebook's active users would yield a volunteer pool comparable to the entire then-existing BOINC user base, and that recruiting 1 percent could double or triple the number of BOINC volunteers. The combined effect of all three incentives operating simultaneously was not studied — the paper evaluated each incentive in a separate simulation and described the 6,000-person model as 'a suitable model on which to base our analysis' despite its small scale relative to Facebook's 800 million users — and the scale-up projection was presented in the conclusion as a forward-looking expectation pending a production version of SoCPR, not as a validated result. (The same paper also described a second prototype, SoCC (Social Collaborative Cloud), for collaborative virtual-machine resource sharing, whose evaluation was limited to a 3-node local-area cluster; that engineering-validation gap concerns a different system and is separable from the incentives-and-scale question considered here.)

Problem Statement

Determine whether socially-networked, gamified recruitment can multiply volunteer computing participation — the open question left by SoCPR. Specifically: (1) what is the joint effect of the three incentive mechanisms (Project Champion, Social Anchor, Compute Magnate) operating simultaneously, when each has only ever been evaluated in isolation — including whether their objectives interact additively or antagonistically (Project Champion pushes contributors toward small projects, Social Anchor rewards recruiting sparsely connected users, Compute Magnate rewards inducing friends to contribute more compute); and (2) does the scale-up conjecture hold — would recruiting on the order of one percent of a large social network's user base double or triple the volunteer computing participant base, as the source authors conjectured by extrapolating then-current user statistics, or do real-world social dynamics, churn, and incentive saturation refute it? In (2), the participant base is to be measured as sustained active participation — recruitment together with retention and net contributed computation (BOINC credits) — rather than one-time registrations: under a registration-counting reading, the source's own arithmetic (one percent of a roughly 800-million-user Facebook base against a roughly 2.2-million-user BOINC base, of whose volunteers only 12 percent actively contribute at any one time) would make the conjecture true by construction, so the conjecture is empirically testable only for sustained active contribution. Both confirming and refuting answers are admissible.

References

  1. Kris Bubendorfer, Kyle Chard, Koshy John, Ashfag M. Thaufeeg, 'eScience in the Social Cloud', Future Generation Computer Systems 29(8):2143–2156, 2013. DOI 10.1016/j.future.2013.04.003, https://doi.org/10.1016/j.future.2013.04.003
  2. Koshy John, Kris Bubendorfer, Kyle Chard, 'A Social Cloud for Public eResearch', 2011 IEEE Seventh International Conference on eScience, pp. 363–370, 2011. DOI 10.1109/eScience.2011.57, https://doi.org/10.1109/eScience.2011.57
  3. Ashfag M. Thaufeeg, Kris Bubendorfer, Kyle Chard, 'Collaborative eResearch in a Social Cloud', 2011 IEEE Seventh International Conference on eScience, pp. 224–231, 2011. DOI 10.1109/eScience.2011.39, https://doi.org/10.1109/eScience.2011.39
  4. Koshy John, 'The Social Cloud for Public eResearch', MSc thesis, Victoria University of Wellington. DOI 10.26686/wgtn.16998664, https://doi.org/10.26686/wgtn.16998664
  5. Ashfag M. Thaufeeg, 'Towards a Social Cloud Framework for Collaborative eResearch', MSc thesis, Victoria University of Wellington. DOI 10.26686/wgtn.17000572, https://doi.org/10.26686/wgtn.17000572
  6. Ryan Chard, Kris Bubendorfer, Kyle Chard, 'Experiences in the design and implementation of a Social Cloud for Volunteer Computing', 2012 IEEE 8th International Conference on e-Science, pp. 1–8, 2012. DOI 10.1109/eScience.2012.6404452, https://doi.org/10.1109/eScience.2012.6404452
  7. Kyle Chard, Simon Caton, Omer Rana, Kris Bubendorfer, 'Social Clouds: A Retrospective', IEEE Cloud Computing 2(6):30–40, 2015. DOI 10.1109/MCC.2015.127, https://doi.org/10.1109/MCC.2015.127
  8. Simon Caton, Christian Haas, Kyle Chard, Kris Bubendorfer, Omer F. Rana, 'A Social Compute Cloud: Allocating and Sharing Infrastructure Resources via Social Networks', IEEE Transactions on Services Computing 7(3):359–372, 2014. DOI 10.1109/TSC.2014.2303091, https://doi.org/10.1109/TSC.2014.2303091
  9. Tessema M. Mengistu, Dunren Che, 'Survey and Taxonomy of Volunteer Computing', ACM Computing Surveys 52(3):1–35, 2019. DOI 10.1145/3320073, https://doi.org/10.1145/3320073
  10. David P. Anderson, 'BOINC: A Platform for Volunteer Computing', Journal of Grid Computing 18(1):99–122, 2020. DOI 10.1007/s10723-019-09497-9, https://doi.org/10.1007/s10723-019-09497-9
  11. Nabil Kadache, Rachid Seghir, 'A New Social Volunteer Computing Environment With Task-Adapted Scheduling Policy (TASP)', International Journal of Grid and High Performance Computing 13(2):39–55, 2021. DOI 10.4018/IJGHPC.2021040103, https://doi.org/10.4018/IJGHPC.2021040103
  12. Alimohammad Shahri, Mahmood Hosseini, Raian Ali, Fabiano Dalpiaz, 'Gamification for Volunteer Cloud Computing', 2014 IEEE/ACM 7th International Conference on Utility and Cloud Computing, pp. 616–617, 2014. DOI 10.1109/UCC.2014.97, https://doi.org/10.1109/UCC.2014.97
  13. Mario García-Valdez, Juan Julian Merelo Guervós, Lucero Lara, Pablo García-Sánchez, 'Increasing Performance via Gamification in a Volunteer-Based Evolutionary Computation System', Information Processing and Management of Uncertainty in Knowledge-Based Systems. Applications (IPMU 2018), CCIS, pp. 342–353, 2018. DOI 10.1007/978-3-319-91479-4_29, https://doi.org/10.1007/978-3-319-91479-4_29
  14. Ognjen Scekic, Christoph Dorn, Schahram Dustdar, 'Simulation-Based Modeling and Evaluation of Incentive Schemes in Crowdsourcing Environments', On the Move to Meaningful Internet Systems: OTM 2013 Conferences, LNCS, pp. 167–184, 2013. DOI 10.1007/978-3-642-41030-7_11, https://doi.org/10.1007/978-3-642-41030-7_11
  15. Nane Kratzke, 'Volunteer Down: How COVID-19 Created the Largest Idling Supercomputer on Earth', Future Internet 12(6):98, 2020. DOI 10.3390/fi12060098, https://doi.org/10.3390/fi12060098