5 Improving methods of identifying harmful play
There is a need to continue to improve the methods used to identify harmful play across all sectors of the industry.
Through the use of data, through observation and interaction by well-trained staff, through effective messaging which prompts self-reflection, or in other ways.
Much recent effort has been put into the development of algorithms designed to evaluate player behaviour. In particular, research into machine play (commissioned by the Responsible Gambling Trust in 2014-15) addressed the issue of whether it is possible to identify harmful play on machines with sufficient confidence to seek to mitigate it through intervention. That research is currently being followed up in a variety of ways by different machine operators. At least one operator is applying it to over-the-counter business as well as to machine play.
So far this work has tended to concentrate on situations where data can be linked to individuals, principally through the use of loyalty cards or verified accounts. That limits its potential, given that take up of cards or accounts is far from universal. Nor do all those who possess loyalty cards or verified accounts always make use of them when gambling. Some may have more than one card. So we are keen to see further exploration of the possibility of applying algorithms effectively to anonymous data.
If particular algorithms prove to be successful in one company’s operations, it is likely they will be successful in those of another, allowing for different circumstances. This work should not therefore be regarded as a competitive exercise. We are encouraged that the main machine operators have established a working group to share intelligence, recognising the commercial sensitivities to which data sharing can give rise. We have also been pleased to see transparency on the part of betting operators about the approaches that they are developing.
Algorithms are not only relevant to betting premises. The Responsible Gambling Trust has recently commissioned research into the use of algorithms to detect potentially harmful play in remote gambling, taking account of existing research in this area. The new research is expected to be completed by December 2016. Remote gambling sites cannot observe players directly for signs of stress in the same way as shop-based staff. Conversely, remote gambling does not suffer from the same limitations on identifying players (because payment mechanisms cannot be anonymous), though remote players can still gamble on several different sites with different operators. We look forward to seeing the outcome of this work.
The algorithmic approach should continue to be pursued energetically, despite its possible limitations. But algorithms should only ever be part of a package of approaches to identify potentially harmful play so that staff can intervene appropriately. Operators need also to continue to experiment and innovate around other methods of detecting harm.
In this context it is helpful that the Responsible Gambling Trust has commissioned a study, working with the Bingo Association and Ipsos MORI, to identify problem gambling in licensed bingo premises (including machine play). This work is due to report in Spring 2016.
Success in developing effective approaches to identifying harmful play using existing data and information may influence our future view on the desirability of mandatory account-based play (Machines gambling research: advice to the Gambling Commission). Mandatory account-based play would raise a number of difficult issues. It does, however, offer other potential advantage in terms of effective self-exclusion, the inhibition of under-age gambling and measures against money-laundering.
Lead responsibility: The gambling industry, both collectively and at the level of individual businesses, with support from the Responsible Gambling Trust.
Indicative timescale: The usefulness or otherwise of algorithms, for the industry as a whole, ought to be well-established by the end of the strategy period, if not earlier. We would expect to see further progress on other methods of detecting potentially harmful play in the same timescale. A Responsible Gambling Trust funded independent evaluation of the Association of British Bookmakers Player Awareness System, which attempts to use algorithms to initiate earlier and targeted interactions with any customer identified as displaying behaviours associated with harmful gambling, is expected to begin in March 2016 and complete in August 2016.
Indicators of success: Well-established protocols across the industry to identify potentially harmful play so that appropriate responses can be made. These will include, but not be limited to, use of data and algorithms to help target interventions towards those who are more likely to be suffering harm. We would expect to see a number of trials and pilots completed and evaluated, with the learning put into practice across the industry.
Relevant to: Priority objectives I and V.