The benefits and challenges of online player tracking

Abstract

In their article, Newall and Swanton [1] make many important observations about the significance of online tracking for academic research and in the development of safer gambling measures. On the whole, I agree with many of their arguments. Online tracking data provide a valuable source of insights into actual gambling behaviour and associated harms [2]. Such information, whether used retrospectively or in real-time, can be used to inform safer gambling practices [3, 4]. These include the application of play limits; the provision of player information; self-exclusion policies; and the development of harm indicators/risk-detection algorithms [5, 6]. It is observed that, although some single operators have developed technology in these areas, there are fundamental challenges. On the demand side, people can use multiple operators and accounts or set non-binding play limits, whereas on the supply side, there may be difficulties gaining consistency across multiple operators or where some gambling is offered offline [2]. For these reasons, the authors discuss the benefits of a ‘single-customer view’ in which the access point for all gambling might be via an independent organisation that provides monitoring and aggregate safer gambling functionality [1].

Although such a model is being discussed in some countries (e.g. United Kingdom) [6] and is partially implemented in some Scandinavian countries [7], there are several practical challenges. First, the consolidation of large amounts of data in a single place, presumably in combination with a know-your-customer system, raises concerns around player privacy. Not only could such information be accessed by third parties for other purposes (e.g. credit agencies/loan approvers) with detrimental consequences for customers, it creates a target or ‘honey-pot’ for cyber attackers. A second issue is the ability to apply tracking in countries with State or Province based laws. For example, in Australia, laws relating to online gambling are applied federally, whereas all other forms of land-based gambling are subject to State laws. Each jurisdiction has multiple sectors (casinos, hotels and clubs), and within these, there are separate operators (e.g. chains or hotels or club syndicates). A third practical issue is the danger of leakage effects in the form of an outflow of higher risk customers to unregulated sites elsewhere using either fiat off-ramps (e.g. Paypal) or direct transfers of value using decentralised blockchain-based wallets. Such sites may have few safer gambling provisions.

Moreover, on the conceptual and design-side, there is the issue of the specifications of aggregated limit-setting and monitoring. How does one determine appropriate levels of expenditure? A commonly suggested solution are ‘affordability checks’ [8], but establishing wealth is a complex process (e.g. cashflow vs asset worth) and could arguably be seen as a form of discrimination based on social class. Other authors propose so-called ‘low risk’ gambling limits [9, 10] that could be used to alert operators to higher than ‘healthy’ levels of expenditure, but determining what is risky or affordable (particularly when there may be semi-professional and high-value sports bettors) may again pose difficulties.

In my view, a stronger approach may be to look for genuine indicators of harm rather than focus on static gambling limits. Current literature indicates that the incidence of these reliably differs between problem and non-problem gamblers. Examples include: declined deposits, reversed withdrawals, frequent in session top-ups or gambling at unusual hours [11-15]. Importantly such work suggests the importance of real-time changes in behaviour. Rather than focusing so much on how long or how much people spend (which raises the concerns above), one looks for statistically unusual spikes or patterns, an accumulation of risk indicators, which occur across time. Operators may, therefore, be required to provide dynamic tools that measure and monitor real-time risk and to provide evidence of their success in reducing harm. However, such insights require greater technical proficiency and potentially greater operator-government collaboration. Such systems could still potentially include a State-sanctioned ‘default’ limit, but require customers to opt-in to change the amount while also completing a risk assessment.

As Newall and Swanton [1] observe, such work can often only be pursued through industry collaboration. Several models are possible here. Industry might be asked to provide data and evidence to researchers, which is then reported to a regulatory body (e.g. Nower and Glynn) [8]. Alternatively, the industry might conduct its own research, which is independently reviewed. A third option is for data to be made available to academic research with measures in place to ensure appropriate independence of analysis and reporting [9]. The role of the regulator would be to ensure that the risk protections applied by industry meet certain evidence standards and are subject to periodic review.

I have undertaken research into online player tracking, which has received industry support (Kindred). My consulting role has been confined to conducting data requests, statistical analysis and reporting. Such industry collaborations I believe to be central to advancing knowledge in this area to gain access to relevant data, product differences and the best areas for regulatory reform. My commentary on this article is not commercially influenced by this work and based on my direct observations and knowledge of the academic literature.