Welcome to another exciting edition of Technology Policy Institute’s Research Roundup, or TPIRR (possibly the most exciting acronym to come out of our nation’s capital since CHOMP: Consumers Have Options for Molar Protection Act). In this edition we highlight the importance of how reputation affects participants in the gig economy, as demonstrated by Amazon’s Mechanical Turk. As it turns out, good reviews mean something; who knew? We also bring you research on large molecule biologics and pharmaceutical intellectual property, Online Dispute Resolution (ODR), and a primer on cyber-security at the network level.
The gig economy is transforming the work landscape, allowing more people to make more money on their own schedules. Simple tasks can be delegated to strangers with the appropriate skills, maximizing total productivity. But questions of trust emerge when work comes through an automated system sans human interaction. Regular jobs involve interviews, which allow employee and employer to theoretically vet one another. Online labor markets try to address this problem through reputation and rating mechanisms. Passengers and drivers rate each other on Uber, for example, as do renters and rentees on AirBnb. This paper asks how well these systems work using Amazon’s Mechanical Turk as its testing grounds.
Amazon’s Mechanical Turk is a system where an ‘employer’ asks for a task to be accomplished, and an ‘employee’ may accept a task at a given rate. However, the employer may decline to pay employees upon completion of the task and keep the product of the work, and employees have no recourse to seek compensation for their works or rate their employer. As a result, third party sites have emerged to fill the gap and provide a forum to rate employers. The study focuses on one such site, Turkopticon. Employees contribute to the site voluntarily, and as the authors point out, the only incentive to post to the forum is altruism, as giving an employer a good review makes it more likely that other employees will then learn about the opportunity and ‘steal’ work. In spite of this, the site is well maintained and populated, and thus was able to serve as the primary data source for the study.
The authors first tested to see if the reputations on the site correlated with wage. One RA randomly selected tasks from employers with good reputations, bad reputations, or no reputations, and sent them to another RA so they could be blindly evaluated. The authors found that employers with good reputations had 40% higher “effective wages” (the amount they are actually paid per task on average) than those with bad reputations, meaning employers with bad reputations were more likely to withhold payment either by refusing to pay employees, posting tasks as ‘trial tasks,’ or giving an intentionally short deadline for the task.
Next, authors determined whether or not these reputations had any impact on employee task selection. They created 36 employers endowed with either 8-12 good ratings, 8-12 bad ratings, or no ratings. They posted the same task for each employer at different times of the day to account for temporal fixed effects, and then measured how many responses they received as well as the quality of those responses. Despite some potential data hiccups (a group of Mechanical Turk users realized that the fake employers all originated from the same creator and exposed the deception on a Mechanical Turk sub-Reddit; the authors then clarified that the accounts were in the name of science), data showed a strong trend towards both more people wanting to work for better reviewed employers, as well as better work being done (employees spent more time on the task for better reviewed employers), compared to both employers with bad reviews and employers with no reviews.
Alan Benson, Aaron J. Sojourner, Akhmed Umyarov In two experiments, we examine the effects of employer reputation in an online labor market (Amazon Mechanical Turk) in which employers may decline to pay workers while keeping their work product. First, in an audit study of employers by a blinded worker, we find that working only for good employers yields 40% higher wages. Second, in an experiment that varied reputation, we find that good-reputation employers attract work of the same quality but at twice the rate as bad-reputation employers. This is the first clean, field evidence on the value of employer reputation. It can serve as collateral against opportunism in the absence of contract enforcement.
- Nicholson Price II and Arti K. Rai
As finding breakthrough small-molecule drugs gets harder, drug companies are increasingly turning to “large molecule” biologics. Although biologics represent many of the most promising new therapies for previously intractable diseases, they are extremely expensive. Moreover, the pathway for generic-type competition set up by Congress in 2010 is unlikely to yield significant cost savings.
In this Article, we provide a fresh diagnosis of, and prescription for, this major public policy problem. We argue that the key cause is pervasive trade secrecy in the complex area of biologics manufacturing. Under the current regime, this trade secrecy, combined with certain features of FDA regulation, not only creates high barriers to entry of indefinite duration but also undermines efforts to advance fundamental knowledge.
In sharp contrast, offering incentives for information disclosure to originator manufacturers would leverage the existing interaction of trade secrecy and the regulatory state in a positive direction. Although trade secrecy, particularly in complex areas like biologics manufacturing, often involves tacit knowledge that is difficult to codify and thus transfer, in this case regulatory requirements that originator manufacturers submit manufacturing details have already codified the relevant tacit knowledge. Incentivizing disclosure of these regulatory submissions would not only spur competition but it would provide a rich source of information upon which additional research, including fundamental research into the science of manufacturing, could build.
In addition to provide fresh diagnosis and prescription in the specific area of biologics, the Article contributes to more general scholarship on trade secrecy and tacit knowledge. Prior scholarship has neglected the extent to which regulation can turn tacit knowledge not only into codified knowledge but into precisely the type of codified knowledge that is most likely to be useful and accurate. The Article also draws a link to the literature on adaptive regulation, arguing that greater regulatory flexibility is necessary and that more fundamental knowledge should spur flexibility.
E-commerce is overshadowing face-to-face (F2F) transactions in business-to-consumer (B2C) commerce. This benefits consumers in providing more buying options, but may leave them with no remedies when purchases go awry. This chapter therefore discusses how online dispute resolution (ODR) systems may expand and equalize remedy systems in B2C exchanges. Part II of the chapter discusses the need for expanded ODR to provide consumers with access to remedies regarding online purchases. Part III explains how ODR systems are unfolding on international and domestic fronts in B2C exchanges. Part IV then highlights their strengths and weaknesses and proposes ideas for how ODR systems can be improved to offer consumers efficient and fair means for accessing e-commerce remedies. The chapter concludes with Part V, an invitation to continue the development of such ODR systems in an effort to foster revived corporate responsibility and empower all consumers regardless of their resources, power, or status.
Trey Herr and Heather West
Internet Security Governance covers the policy challenges that arise from building and governing security in the Internet’s architecture and key protocols. It includes the discussion of defensively oriented technical and legal topics that cross national boundaries and/or involve security of the underlying protocols and hardware which make up the Internet. Also discussed are international agreements, like the Wassenaar Arrangement, and the process of drafting, approving, and promulgating security standards for implementation across the Internet.