http://www.hpl.hp.com/research/scl/ Recent papers from HP Labs' Social Computing Lab http://www.hpl.hp.com/research/scl/papers/predictions popularity, youtube, digg, attention, predicting future downloads. attention youtube popularity social media predictions online content We present a method for accurately predicting the long time popularity of online content from early measurements of user access. Using two content sharing portals, Youtube and Digg, we show that by modeling the accrual of views and votes on content offered by these services we can predict the long-term dynamics of individual submissions from initial data. In the case of Digg, measuring access to given stories during the first two hours allows us to forecast their popularity 30 days ahead with remarkable accuracy, while downloads of Youtube videos need to be followed for 10 days to attain the same performance. The differing time scales of the predictions are shown to be due to differences in how content is consumed on the two portals: Digg stories quickly become outdated, while Youtube videos are still found long after they are initially submitted to the portal. We show that predictions are more accurate for submissions for which attention decays quickly, whereas predictions for evergreen content will be prone to larger errors. Gabor Szabo and Bernardo A. Huberman 2008-11-03 15:27:00 http://www.hpl.hp.com/research/scl/papers/watercooler Visibility, attention, and recognition drive participation in internal corporate social media. watercooler blogs attention social media hp CSCW Blogs, wikis, and forums can break down geographic distances, workgroup boundaries, and organizational hierarchy in an organization. While these tools significantly lower the barriers to producing content, employees may perceive there to be little incentive to invest their own time in providing this content for public consumption. We found that increasing visibility often motivated employees to participate and contribute content. Employees were motivated by the opportunity for attention, and the ways in which social media tools enabled or hindered this opportunity influenced the way it was used. In this paper, we describe the design and use of the internal social media platforms at Hewlett-Packard and examine the ways that employees used these tools. Specifically, we explore ways in which designing for increased visibility and providing opportunities for recognition improve the ways that social media platforms can be used in organizations. To appear at CSCW 2008 Workshop on Enterprise 3.0. Michael J. Brzozowski and Sarita Yardi 2008-10-13 15:27:00 http://www.hpl.hp.com/research/scl/papers/blogging/ Participation in internal corporate blogs is both work-related and social, indicating a desire to connect with coworkers on multiple levels. blogs community temporal patterns hp CSCW Blogging at work has gained considerable interest in the knowledge management community. It is not clear, however, how much of work blogging is work-related versus social, or when work blogging takes place. In this poster, we present results from our examination of the temporal aspects of blogging within a large internal corporate blogging community. We compared our findings to similar analyses of employee email use and to college student Facebook use. We found that blog posting is temporally similar to email, while blog reading is more similar to Facebook messaging. Our results suggest that participation is both work-related and social, indicating a desire to connect to coworkers at multiple levels. To appear at CSCW 2008. Sarita Yardi, Scott Golder, and Michael J. Brzozowski 2008-10-13 15:15:00 http://www.hpl.hp.com/research/scl/papers/sncf/ User-generated social networking links can be as predictive as algorithmically identified "neighbors" in recommender systems. collaborative filtering social networks prediction recommender systems This paper demonstrates that "social network collaborative filtering" (SNCF), wherein user-selected like-minded alters are used to make predictions, can rival traditional user-to-user collaborative filtering (CF) in predictive accuracy. Using a unique data set from an online community where users rated items and also created social networking links specifically intended to represent like-minded “allies,” we use SNCF and traditional CF to predict ratings by networked users. We find that SNCF using generic "friend" alters is moderately worse than the better CF techniques, but outperforms benchmarks such as by-item or by-user average rating; generic friends often are not like-minded. However, SNCF using "ally" alters is competitive with CF. These results are significant because SNCF is tremendously more computationally efficient than traditional user-user CF and may be implemented in large-scale web commerce and social networking communities. It is notoriously difficult to distinguish the contributions of social influence (where allies influence users) and "social” selection (where users are simply effective at selecting like-minded people as their allies). Nonetheless, comparing similarity over time, we do show no evidence of strong social influence among allies or friends. Rong Zheng, Dennis M. Wilkinson and Foster Provost 2008-10-06 12:00:00 http://www.hpl.hp.com/research/scl/papers/crowd/crowd.pdf How to solve the digital commons dilemma. attention social networks reputation crowdsourcing The tragedy of the digital commons does not seem to prevent the copious voluntary production of content that one witnesses in the web. We show through an analysis of a massive data set from Youtube that the productivity exhibited in crowdsourcing exhibits a strong positive dependence on attention, measured by the number of downloads. Conversely, a lack of attention leads to a decrease in the number of videos uploaded and the consequent drop in productivity, which in many cases asymptotes to no uploads whatsoever. Moreover, we observed that uploaders compare themselves to others when having low productivity and to themselves when exceeding a personal threshold. Bernardo A. Huberman, Daniel M. Romero and Fang Wu 2008-09-11 12:00:00 http://www.hpl.hp.com/research/scl/papers/howopinions/wine.pdf How web discourse evolves. To appear in the Proceedings of the Workshop on Internet and Network Economics-2008 opinion formation social networks polarization crowdsourcing No aspect of the massive participation in content creation that the web enables is more evident than in the countless number of opinions, news and product reviews that are constantly posted on the Internet. Given their importance we have analyzed their temporal evo- lution in a number of scenarios. We have found that while ignorance of previous views leads to a uniform sampling of the range of opinions among a community, exposure of previous opinions to potential review- ers induces a trend following process which leads to the expression of increasingly extreme views. Moreover, when the expression of an opinion is costly and previous views are known, a selection bias softens the ex- treme views, as people exhibit a tendency to speak out differently from previous opinions. These findings are not only robust but also suggest simple procedures to extract given types of opinions from the population at large. Fang Wu and Bernardo A. Huberman 2008-09-11 12:00:00 http://www.hpl.hp.com/research/scl/papers/reputationExpt/reputation-expts-and-Prosper.pdf How people use reputation information. reputation incentive design experimental economics We evaluated how people use reputation in a laboratory market and in Prosper, an online microfinance business. We found people use information on past behavior to ostracize previous poor performance in both cases. The laboratory market did not show significant price discrimination, but people used their ability to not fulfill contracts to punish poor performers. Price discrimination was significantly correlated with reputation in Prosper. Thus we find people apply multiple strategies to deal with reputation. Kay-Yut Chen, Scott Golder, Tad Hogg and Cecilia Zenteno 2008-08-19 12:00:00 http://arxiv.org/abs/0707.4195 How people perform in an auction using simulated quantum information processing. quantum information incentive design experimental economics We describe human-subject laboratory experiments on probabilistic auctions based on previously proposed auction protocols involving the simulated manipulation and communication of quantum states. These auctions are probabilistic in determining which bidder wins, or having no winner, rather than always having the highest bidder win. Comparing two quantum protocols in the context of first-price sealed bid auctions, we find the one predicted to be superior by game theory also performs better experimentally. We also compare with a conventional first price auction, which gives higher performance. Thus to provide benefits, the quantum protocol requires more complex economic scenarios such as maintaining privacy of bids over a series of related auctions or involving allocative externalities. Kay-Yut Chen and Tad Hogg 2008-08-19 12:00:00 http://arxiv.org/abs/0704.0800 A privacy-preserving auction using quantum information processing. quantum information incentive design game theory We present a quantum auction protocol using superpositions to represent bids and distributed search to identify the winner(s). Measuring the final quantum state gives the auction outcome while simultaneously destroying the superposition. Thus non-winning bids are never revealed. Participants can use entanglement to arrange for correlations among their bids, with the assurance that this entanglement is not observable by others. The protocol is useful for information hiding applications, such as partnership bidding with allocative externality or concerns about revealing bidding preferences. The protocol applies to a variety of auction types, e.g., first or second price, and to auctions involving either a single item or arbitrary bundles of items (i.e., combinatorial auctions). We analyze the game-theoretical behavior of the quantum protocol for the simple case of a sealed-bid quantum, and show how a suitably designed adiabatic search reduces the possibilities for bidders to game the auction. This design illustrates how incentive rather that computational constraints affect quantum algorithm choices. Tad Hogg, Pavithra Harsha and Kay-Yut Chen 2008-08-19 12:00:00 http://www.hpl.hp.com/personal/Thomas_Sandholm/sandholm2008a.pdf Tradeoffs between using spot and reservation markets. tycoon incentive design resource allocation markets We propose, implement and evaluate three admission models for computational Grids. The models take the expected demand into account and offer a specific performance guarantee. The main issue addressed is how users and providers should make the tradeoff between a best effort (low guarantee) spot market and an admission controlled (high guarantee) reservation market. Using a realistically modeled high performance computing workload and utility models of user preferences, we run experiments highlighting the conditions under which different markets and admission models are efficient. The experimental results show that providers can make large efficiency gains if the admission model is chosen dynamically based on the current load, likewise we show that users have an opportunity to optimize their job performance by carefully picking the right market based on the state of the system, and the characteristics of the application to be run. Finally, we provide simple functional expressions that can guide both users and providers when making decisions about guarantee levels to request or offer. Thomas Sandholm, Kevin Lai, and Scott Clearwater 2008-06-06 12:00:00