|
|
Home
| Databases
| WorldLII
| Search
| Feedback
Computers and Law: Journal for the Australian and New Zealand Societies for Computers and the Law |
Jeannie Paterson, Gabby Bush, and Tim Miller [1]
19th November 2020
Technical developments in data collection, storage and analytics have provided firms with new opportunities for reaching consumers.[2] Firms may now target adverts to individual consumers based on predicted preferences derived from their digital profile.[3] The predictive analytics that enable targeted advertising also open the possibility of differential or discriminatory pricing deployed against individual consumers. Firms are able to use algorithms to individualise price offerings to consumers, and to vary pricing strategies in a short period of time in response to information about consumers’ responses to the advertising sent to them.[4] There are a number of potential harms to consumers in differential pricing, including the impairing consumer choice, and also discrimination and inequitable or unfair treatment. Although many consumers are yielding personal data that fuels these processes, it is not clear that they understand the consequences of such uses. Further, the incidence of differential pricing is difficult to discover. Such conduct runs counter to the common emphasis in formulations of standards for ethics in the use of artificial intelligence, which emphasise transparency and opportunities to contest adverse decisions.[5]
This article considers the harms to consumers occasioned by differential pricing and legal strategies for promoting greater transparency, which also provides a pathway to contestability. It does this by particular reference to the recent revelation of differential pricing strategies used by dating app Tinder.[6]
Data driven differential pricing
Differential pricing allows firms to vary the price of goods or services for different individuals or groups of consumers.[7] Differential pricing is made possible by the large amounts of information gathered when consumers sign up for a service, such as name, gender, location, interests, as well as data derived from consumers’ browsing, purchase and download history.[8] Firms use this data to distinguish consumers between according to their predicted price sensitivity.[9] Firms can then offer products to consumers at prices that accord with these predictions and also steer consumers to more highly priced products.
Differential pricing can be difficult to identify,[10] precisely because each consumer will be given a different price, and will not see what others are offered.[11] An example in Australia has nonetheless come from the consumer advocacy group Choice.[12] In August 2020, Choice published a study finding that Tinder was charging different prices to different users for its premium service Tinder Plus.[13] Users over the age of 30 were charged double those under 30, and prices also varied greatly even within age brackets. Furthermore, prices differed as between consumers in regional areas as compared to urban areas, people who were queer or with non-binary gender identities. The pricing difference for the premium services suggests Tinder’s algorithm sets the pricing tailored to each individual user based upon predictions about how much they are willing to pay. It is likely that this differential pricing is a central part of a business model. Tinder offers no explanation for the prices quoted to users and hence consumers are unlikely to be aware of the differences. Indeed, it may be that remaining taboos surrounding online dating disincentivise consumers from disclosing their payment habits. This makes it easier for Tinder to offer algorithmically generated prices without explanation.
A recent survey by Choice found that generally consumers do not think differential pricing is fair, but this depends somewhat on the basis of the difference:
• ‘14% say it's fair to charge higher earners more than lower income earners’
• ‘13% say it's fair to charge people living in wealthier areas more than those in less wealthy areas’;
• the majority of consumers ‘believe these practices are unfair (74% and 77%, respectively)’; and
• ‘about a quarter of Australians (26%) would be happy to pay a different price based on the use of personal information if the pricing benefits them’.[14]
We suggest that there are a number of possible harms arising from differential pricing. Such conduct may decrease overall consumer welfare through consumers paying too much for a product.[15] Of course consumers can always choose not to buy a product or can hunt around for alternatives. The problem is that consumers may not realise that a range of possible prices are being made available for any particular product and this may distort their attitude to what is a good price for that product.
There are some ethical objections to differential pricing as well, arising from the unequal, and potentially unfair, treatment of consumers. As the Choice survey suggests, some people may not object to well-off consumers being charged more for certain goods and services, as may be the philosophy informing Tinder’s charging older consumer more than younger ones. This approach to differential pricing may then benefit consumers who are not as financially capable of paying, therefore prioritising the interests of groups such as students, retirees and those from lower socio-economic demographics. However, the selection of who can pay may more be inaccurate because attributes such as age and location are, at best, crude approximations of wealth.
Differential pricing also work against the interests of already disadvantaged consumers. For example, firms might use profiling to charge higher prices to low-income consumers either to discourage them as customers or as a form of insurance against credit and performance risks perceived to attach to specific groups of customers.[16] In addition, differential pricing might to discriminate against certain consumers on the basis of protected attributes such as age, gender or race, which is bias that can also be perpetuated by the algorithms.[17] The data the algorithm uses to learn is based on consumer preference. If people prefer faces of those with the same colour skin as them, the matching algorithms will learn to segregate users based on colour. Discrimination law prohibits discrimination based on protected attributes such as sex, race, disability, age, sexuality, gender identity, parental and carer responsibilities, and political or religious beliefs.[18] Choice quotes an anti-discrimination specialist who says that Tinder’s differential pricing that burdens older users would be unlawful age discrimination.[19] As a society we may not think it is fair for firms to treat consumers who are less well-off differently from those who are better off. Even where differential pricing is justified there is the problem of consumers being placed in the wrong pricing category for their own circumstances. However, because the differential pricing practice are so opaque, it will be difficult for consumers and regulators to contest cases involving unlawful discrimination or error, and for policymakers to assess the extent of the associated harm. Thus, the first step in allowing differential pricing strategies to be contestable is to seek greater transparency.
Responses to consumer protection in differential pricing
Firms might consider that it is in their interests to be transparent to consumers about their use of differential pricing.[20] This could be seen as a way of building trust with consumers and supporting a good reputation for fair dealing in the market. But we think these issues are too important to be left to the discretion of individual firms. We suggest more emphasis should be placed on legal mechanisms for promoting greater transparency in differential pricing, including through prohibitions on misleading conduct, privacy protection, and direct disclosure requirements.
Prohibitions on misleading conduct
One key response to the use of differential pricing is in the prohibition on misleading conduct in s 18 of the Australian Consumer Law.[21] For example, advertising a best price guarantee for a product may be misleading if different consumers are, for no good reason, presented with different prices but not the best price. Earlier this year, the Federal Court found travel site Trivago engaged in misleading conduct when its algorithm recommended hotel rooms based on who paid the most for ad clicks, not on the best price available.[22]
The prohibition on misleading conduct does not only apply to positive misrepresentations. Omissions and silence can also mislead. This will be the case if other conduct has created a misleading impression or, in the context, consumers have a reasonable expectation of disclosure of key information.[23] Thus, a website that implicitly creates an expectation of equal prices for all consumers may be misleading if it fails to disclose the reality of differential pricing. Tinder's privacy policy and terms of use do not mention that it uses personal information to inform the range of prices available to customers. It is not clear whether its website creates an expectation of equal pricing but the fact the Tinder Plus service does not distinguish between different categories of users might suggest this to be the case. Research has shown that the three primary concerns of users of online dating apps are ‘personal security, misrepresentation and recognition’, however these concerns directly relate to how their potential date and themselves are presented and recognised, not how the platform represents the service.[24] The concerns are internalised, allowing for misrepresentation on the platform to go unchecked.
Data rights and consent
A second method for improving the transparency of differential pricing is through strong requirements for consumers to have provided informed consent to such uses of their personal data. The ‘price’ for consumers obtaining access to digital services is commonly foregoing control of their data. In many contexts in Australia, consumers are not required in law expressly to consent to these practices. Under the Privacy Act 1988 (Cth) entities do not need to seek consent to collect personal information where it is ‘reasonably necessary for one or more of the entity’s functions or activities’.[25] The ACCC suggests that data collection about a consumer web browsing activities might be reasonably necessary for a digital platform to perform its advertising functions.[26] Similarly, firms providing a service may be entitled to use that information to tailor both the service and the price.
Firms collecting and repurposing data must have a privacy policy disclosing how personal information is collected, held and used.[27] However, the ACCC’s review of privacy statements on targeted advertising provided by digital platforms found them unclear and unhelpful.[28] We know from other circumstances that consumers struggle to read the disclosure statements associated with financial products,[29] are confused by contract boilerplate[30] and cannot hope to read the terms of online retailers to know their rights to return.[31] They are unlikely to do better with privacy policies.
The ACCC Digital Platforms Inquiry Report made recommendations for the reform of the Privacy Act 1988 (Cth) to provide more robust mechanisms for ensuring consumer consent to data collection and processing practices.[32] These reforms would provide consumers with the opportunity to refuse to consent to the use of their data for differential pricing and other forms of targeted marketing. But this right is hollow is consumers do not really understand the extent of the practices in question.[33] This raises the role of warnings.
Warnings
A third strategy for increasing transparency around differential pricing is through a requirement for firms to provide a warning notice when they use differential pricing.[34] The efficacy of mandatory disclosure as a central strategy of consumer protection has been challenged in a number ways.[35] In particular, disquiet with disclosure as a policy strategy has arisen insights from behavioural economics, which identify the limits on the capacity of consumers to act rationally, even when provided with sufficient information.[36] Nonetheless, we know that concise, targeted information is generally a more effective way of communicating with individuals than generalised high volume information.[37] Some consumers may be happy to accept differential pricing. Nonetheless, a mandatory stark warning when differential pricing is being used may be sufficient to prompt consumers actively to compare prices from different sources, or to seek out other consumers to compare prices, as well as reflecting on the value of enhancing privacy by refusing to share data for advertising purposes.
Conclusion
We may have differing views on online dating apps and the ethics of charging different prices to various categories of people. However, consumers should be able to make choices about whether they proceed with a transaction based on differential pricing and to contest algorithmically informed decisions that are based on inaccurate or discriminatory categorisations. For these rights to be effective there needs to be transparency around the pricing practices. Substantive fairness protections may also be needed. However, measures designed to promote transparency are a first step in informing and protecting online consumers.
[1] Centre for AI and Digital Ethics.
[2] See generally Australian Competition and Consumer Commission, Digital Platforms Inquiry (Final Report, June 2019) Ch 7.
[3] See Ryan Calo, ‘Digital Market Manipulation’ (2014) 82(4) George Washington Law Review 995, 1003-4; Stuart A Thompson, ‘These Ads Think They Know You’, The New York Times (30 April 2019) https://www.nytimes.com/interactive/2019/04/30/opinion/privacy-targeted-advertising.html?mtrref=www.google.com&gwh=BC5DDB91D5483424EFDB04440D2D8F14&gwt=pay.
[4] Frederik Zuiderveen Borgesius and Joost Poort, ‘Online Price Discrimination and EU Data Privacy Law’ (2017) 40(3) Journal of Consumer Policy 347, 351; Maurice E Stucke and Ariel Ezrachi, ‘How Digital Assistants Can Harm Our Economy, Privacy, and Democracy’ (2017) 32(3) Berkeley Technology Law Journal 1239, 1264.
[5] A Jobin, M Ienca, & E Vayena, ‘The global landscape of AI ethics guidelines’ (2019) 1(9) Nature Machine Intelligence 389.
[6] Additionally, note that Tinder has recently been criticised for its lack of response to sexual assault: https://www.abc.net.au/news/2020-10-12/tinder-dating-app-helps-sexual-predators-hide-four-corners/12722732?nw=0.
[7] Silvia Merler, ‘Big Data and First-Degree Price Discrimination’, Bruegel (Blog Post, 20 February 2017) <http://bruegel.org/2017/02/big-data-and-first-degree-price-discrimination> .
[8] Digital Platforms Inquiry (n 1) 517.
[9] Borgesius and Poort (n 3) 351.
[10] Digital Platforms Inquiry (n 1) 446.
[11] See also examples in Borgesius and Poort, (n 3) 350; See also Dana Mattioli, ‘On Orbitz, Mac Users Steered to Pricier Hotels’, The Wall Street Journal (23 August 2012) <https://www.wsj.com/articles/SB10001424052702304458604577488822667325882>.
[12] Erin Turner, ‘Op-ed: Tinder's secret pricing shows how companies use our data against us’ Choice (11 August 2020) <https://www.choice.com.au/electronics-and-technology/internet/using-online-services/articles/tinder-plus-op-ed?gclid=Cj0KCQiAnb79BRDgARIsAOVbhRrgCaFQE3sAH3HIJ6hs2E4JAMkeKn0yjmB0XT-vj5yQPb9DEY9_MfAaAh9SEALw_wcB>.
[13] Saimi Jeong, ‘Tinder charges older people more’ Choice (11 August 2020) <https://www.choice.com.au/electronics-and-technology/internet/using-online-services/articles/tinder-plus-costs-more-if-youre-older>.
[14] Saimi Jeong, ‘How Australians feel about different prices for the same goods’ Choice (5 November 2020) <https://www.choice.com.au/money/budget/consumer-pulse/articles/differential-pricing-how-australians-feel-about-it”. Also Phuong Nguyen and Lauren Solomon, Consumer Data and the Digital Economy: Emerging Issues in Data Collection, Use and Sharing (Report, 2018) 34; Justus Haucap, Werner Reinartz, and Nico Wiegand, ‘When Customers Are — and Aren’t — OK with Personalized Prices’ Harvard Business Review (31 May 2018) <https://hbr.org/2018/05/when-customers-are-and-arent-ok-with-personalized-prices>.
[15] See Borgesius and Poort, (n 3) 53; Gerhard Wagner and Horst Eidenmuller, ‘Down by Algorithms? Siphoning Rents, Exploiting Biases, and Shaping Preferences: Regulating the Dark Side of Personalised Transactions’ (2019) 86(2) University of Chicago Law Review 581, 587-8.
[16] Kevin Smith, ‘Organizations Petition State to Ban Job, Education Criteria When Setting Auto Insurance Rates’, Los Angeles Daily News (22 February 2019) <https://www.dailynews.com/2019/02/22/organizations-petition-state-to-ban-job-education-criteria-when-setting-auto-insurance-rates/>.
[17] See eg Julia Angwin, Ariana Tobin and Madeleine Varner, ‘Facebook (Still) Letting Housing Advertisers Exclude Users by Race’, ProPublica (21 November 2017) <https://www.propublica.org/article/facebook-advertising-discrimination-housing-race-sex-national-origin>.
[18] See, eg, Age Discrimination Act 2004 (Cth), Disability Discrimination Act 1992 (Cth), Sex Discrimination Act 1984 (Cth), Racial Discrimination Act 1975 (Cth).
[19] Jeong, (n12).
[20] But cf Tami Kim, Kate Barasz and Leslie K John, ‘Why Am I Seeing This Ad? The Effect of Ad Transparency on Ad Effectiveness’ (2019) 45(5) Journal of Consumer Research 906.
[21] Competition and Consumer Act 2010 (Cth) sch 2 s 18.
[22] See Trivago N.V. v Australian Competition and Consumer Commission [2020] FCAFC 185.
[23] Demagogue Pty Ltd v Ramensky [1992] FCA 557; (1992) 39 FCR 31, 32.
[24] Jennifer Gibbs, Nicole Ellison and Lai Chih-hui, ‘First Comes Love, Then Comes Google: An Investigation of Uncertainty Reduction Strategies and Self-Disclosure in Online Dating’ (2011) 38(1) Communication Research 70.
[25] Privacy Act 1988 (Cth) sch 1 (Australian Privacy Principles) s 3.1.
[26] Digital Platforms Inquiry (n 1) 438.
[27] Australian Privacy Principles (n 24) s 1.
[28] Digital Platforms Inquiry (n 1) 413.
[29] Australian Treasury, Financial System Inquiry (Final Report 2014) 193.
[30] Margaret Jane Radin, Boilerplate: The Fine Print, Vanishing Rights, and the Rule of Law (Princeton University Press, 2013).
[31] David Berreby, ‘Click to Agree With What? No One Reads Terms of Service, Studies Confirm’, The Guardian (4 March 2017) <https://www.theguardian.com/technology/2017/mar/03/terms-of-service-online-contracts-fine-print>.
[32] Digital Platforms Inquiry (n 1) 34–5. These suggested reforms would bring Australian privacy law closer to the regime for data protection in the General Data Protection Regulation: Regulation (EU) 2016/679 of 27 April 2016 on the Protection of Natural Persons with Regard to the Processing of Personal Data and on the Free Movement of Such Data and repealing Directive 95/46/EC (General Data Protection Regulation) [2016] OJ L 119/1.
[33] Dennis D. Hirsch, ‘From Individual Control to Social Protection: New Paradigms for Privacy Law in the Age of Predictive Analytics’ (2020) 79 Md. L. Rev. 439, 460-461.
[34] Wagner and Eidenmuller (n 14), 590; House of Lords (Select Committee on European Union), Online Platforms and the Digital Single Market, 2016, [291] <https://publications.parliament.uk/pa/ld201516/ldselect/ldeucom/129/129.pdf>.
[35] Geraint Howells, ‘The Potential and Limits of Consumer Empowerment by Information’ (2005) 32(3) Journal of Law and Society 349.
[36] See, eg, Russell Korobkin, ‘Bounded Rationality, Standard Form Contracts, and Unconscionability’ (2003) 70(4) University of Chicago Law Review 1203.
[37] See Anthony Duggan and Iain Ramsay, ‘Front-End Strategies for Improving Consumer Access to Justice’ in Michael Trebilcock, Anthony Duggan and Lorne Sossin (eds) Middle Income Access to Justice (University of Toronto Press, 2014) 95.
AustLII:
Copyright Policy
|
Disclaimers
|
Privacy Policy
|
Feedback
URL: http://www.austlii.edu.au/au/journals/ANZCompuLawJl/2021/13.html