To Top

Researcher Discusses State Of Responsible Gambling Algorithms In Online Gaming

Dr. Sarah E. Nelson recently gave a presentation discussing the current state of algorithms promoting responsible gambling.

Computer Algorithm
Photo by Shutterstock/TippaPatt
Derek Helling Avatar
4 mins read

Data science, among the numerous disciplines, gets a lot of attention because of how novel its uses are and how pertinent data is to everyday life. In many facets of life, a human being’s very existence is inextricably linked to their data.

One of those facets is a person’s relationship with online casino operators. Data governs many aspects of that relationship, including to what extent people can play on those casino apps, if at all, financial transactions, and storage of identifying information.

The science of data has enabled some insight into whether people struggle with gambling-related behavioral pathologies and how casino operators respond to that situation. Those methods are still evolving, as Dr. Sarah E. Nelson recently discussed in a presentation for the International Center for Responsible Gaming.

In the presentation, Nelson examined the development of algorithms connected with responsible gambling and what they are actually measuring. Nelson also gave recommendations on how to shape the continued evolution of those algorithms to improve outcomes for everyone in the gaming industry.

Nelson draws on experience in discussing algorithms

Nelson gave her presentation on May 15, 2024, and entitled it “Responsible Gambling Algorithms: What Are We Really Measuring?” Nelson is director for research at the Division on Addiction, Cambridge Health Alliance.

Her numerous published works spanning two decades include papers on problem gambling from a number of perspectives. In her current role, Nelson researches the etiology of addiction.

In her presentation, Nelson presented her experiences with building and utilizing algorithms within the context of responsible gambling. That included laying bare how the process happens.

Nelson discusses weaknesses of early algorithms

The deployment of algorithms to identify potentially problematic gambling behavior in users and address that issue is becoming quite commonplace for US online casinos. While the development of such algorithms is a positive development, Nelson’s presentation points to the importance of understanding what the algorithms measure.

Nelson explained that when the gambling industry began building algorithms for this purpose, the algorithms weren’t really built to predict people’s risk for gambling problems. Rather, the algorithms assessed the presence or lack thereof of what were understood to be signs of gambling problems in existing recorded data.

In plain language, the algorithms indicated how frequently a behavior occurred within a data set. They struggled to connect those behaviors with documented gambling problems, however.

“Early on with player records and gambling algorithms we didn’t have access to assessments of gambling disorder so we used other things to stand in for it,” Nelson elaborated. “As a result, though, our algorithms might be trained to predict self-exclusion, not gambling problems necessarily.”

The algorithms have improved with effort and time. However, Nelson still has a major criticism of the existing algorithms.

Current state of responsible gambling algorithms

Nelson verbalized her primary concern with the performance of existing algorithms.

“Basically what we’re distinguishing with a lot of our algorithms is yes, whether somebody has a problem or not, but also simply whether they gamble a lot or not, and I would consider this a problem because I think we don’t need to institute and put in all these complex algorithms to tell us who’s gambling a lot,” Nelson commented. “We already know that. What’s important is understanding among fairly regular gamblers, what are the triggers or behaviors or markers that are distinguishing people who are really experiencing problems from those who are just regular gamblers? I think we are doing a much less better job of that.”

Nelson pointed to a common misconception among people who do not work in the sciences regularly; correlation does not equal causation. While frequent gambling can be a sign of a gambling problem, it is not inherent. Not all people who gamble frequently have an issue with compulsive gambling and not all people who struggle with a gambling-related problem play frequently.

As far as how to improve current algorithms, Nelson identified potential weaknesses with the data processing itself.

Addressing flaws in the data

Nelson stressed the need to ensure that the data itself is accurate and relevant to improve algorithms’ ability to identify potential problems.

“No matter how complex our models, they are only as valid as the data and the outcomes upon which they’re built,” Nelson stated.

Flaws in the data can include biases that are either intentionally or unintentionally “baked in” according to Nelson. For example, if the data only represents people from a certain geographic area, that could skew the algorithm and potentially make it less useful when applying it to people from a different area.

Nelson emphasized that, in her experience, there are few efforts to detect such biases in responsible gambling algorithms. Furthermore, she mentioned that among algorithms that have been tested for their accuracy in predicting risk for problem gambling, that accuracy is low. That especially applies when separating out people with a low or medium risk.

The great thing about identifying flaws with any algorithm is that doing so enables data scientists to try to address those weaknesses. Nelson shared some strategies for doing exactly that.

Nelson’s solutions for improving algorithms

Overall, Nelson called for the devotion of more resources to evaluate algorithms on the basis of their performance, not their intent. For example, she called for testing of the algorithms to identify coded biases.

Nelson also acknowledged that progress has been made in improving these algorithms and is ongoing.
“I want to make sure we still acknowledge all the work that has gone into the predictive modeling and the algorithms that we do have,” Nelson added. “We do have some really advanced models and advanced systems that can be used in really positive ways to help people at risk.”

True science has no laurels to rest on, as professionals like Nelson continue to believe that the products and services it creates can always improve. The results of that labor and research are fewer people struggling with problem gambling and better interventions. In that way, data science can save lives.

Derek Helling Avatar
Written by

Derek Helling is the assistant managing editor of PlayUSA. Helling focuses on breaking news, including finance, regulation, and technology in the gaming industry. Helling completed his journalism degree at the University of Iowa and resides in Chicago

View all posts by Derek Helling

Derek Helling is the assistant managing editor of PlayUSA. Helling focuses on breaking news, including finance, regulation, and technology in the gaming industry. Helling completed his journalism degree at the University of Iowa and resides in Chicago

Privacy Policy