Transactional Privacy: An Auction to Pay You for Your Data
In our information economy, online networks have been utilized by marketing firms and large corporations to exploit user information in an effort to strategically control their respective markets. By knowing a users demographic and behavioral data firms can utilize algorithms to target individual users to increase their bottom lines. Thus, the economics of information sharing is an extremely important facet of networks, as understanding both the network structure and the information can create asymmetry and hurt consumers in the free market. Furthermore, the murky legislation surrounding these new information technologies makes it unclear to both firms and users what is legally within their reach to obtain or protect their privacy. Thus, Riederer et al. describes a system that would create transparency for the end-user and possibly allow them to be financially compensated for their data.
In order to use many of today’s popular web services (i.e. Facebook, Twitter, Instagram, etc…) users are required to consent to terms and conditions outlining how firms will utilize their personal information generated on the site. However, many users don’t take the time to read these ambiguous conditions, leaving them to consent to unknowingly sell valuable market research data. Riederer et al. instead, proposed a new mechanism for the handling of this information known as Transactional Privacy (TP). To explain, TP works by users opting into a website or service with clear guidelines that “certain” information will be sold on the open market to information aggregators. Here, these aggregators will utilize an auction system (described in the journal article as a first-price sealed-bid auction) to buy an individuals information. Thus, aggregators will place a bid b(i) at their maximum value for the information V(i), however there is a fixed price p that acts as a price floor that all aggregators must bid above. Auctioneers will utilize a complex algorithm to set the fixed price for bidders. This auction market intelligently preforms all required accounting and collecting mechanisms for the end-users and the aggregators. Finally, the winning aggregators pay their bid (bi) and a report is sent to the user with the information of who bought their data.
There are many benefits to using an open market auction to crowd-source user-data. For one, many of the current mechanisms put in place by companies to sell data to advertisers utilize a complex algorithm that has only “limited variables” that are “safe” to sell, thus these aggregators must reveal either how the algorithm works or collecting less relevant data based on legal guidelines. By selling “raw information” by permission from the end-user information aggregates will get more reliable and relevant data. The end-user will also have peace-of-mind knowing exactly what information they choose to sell, and they will be financially compensated for their privacy data (unlike in our current back-door data aggregation system). This should also create a “positive reinforcement effect” in which end-users come to value their personal information that will increase privacy in the macro-system. On the other hand, aggregators can expect better quality information without the legal burdens of anti-privacy advocates. Finally, because the auction only provides user information for a “short-time” no one aggregator can establish a monopoly on the entire market and strategically manipulate it.
In conclusion, as end-users increasingly disclose more private information on the internet, it is in the best interests of the end-user, information aggregates, and public policy officials to institute an auction-based platform to regulate and sell the flow of this sensitive information. This will provide more transparency to the user, increase utility to the aggregators, and increase privacy in the overall network. It is extremely vital that we adopt a system similar to Transactional Privacy because who controls our information ultimately controls the free market.
This article was originally published in Cornell Universities Networks Blog on September 19th, 2016. Source material for this discussion can be found at Columbia University department of computer science.