In the mid 1990s, I was asked by a large U.S. gambling company what I knew about internet gambling. “Not much”, I replied. I was given an assignment to find out as much as I could about the internet and whether “internet gambling” was/would be an opportunity or a threat.
For the next six months, I travelled the globe speaking to people about the technology, trying to understand the technical challenges and how things were likely to develop.
I presented my findings to my client and produced a summary presentation, which I gave at a conference. Someone from the GB Gaming Board (as it was then) asked me if I would present it at the upcoming Gambling Regulators European Forum meeting, which I did. I took the audience through the early days, how the internet was a U.S. Department of Defence communications network project (ARPANET) that was designed to be self-healing; if the network was damaged, the message would find a route through. My presentation explained the challenges that regulators would face in the future as the technology improved and internet gambling would inevitably become more mainstream.
The question I posed was how you regulate something that, due to the nature of the network, is very difficult to block access to. One regulator from a country’s Ministry of Justice proposed a solution. “It is very simple, Mr Tottenham. We will ban it!” Either I had not explained the problem very well or he had not been listening.
Fast forward 30 years. Picture a pub a few hundred metres from a national border. The border is porous with very few checks on passports; cars are just waved through. The pub on one side of the border is run by a landlord scrupulous about the rules his country imposes on him. He pours drinks in small glasses. He pours slowly, because the law requires him to take his time about it. Lines form at the bar. He will not serve anyone more than two drinks in a sitting and scrutinises every ID. He has to keep a register of the times the person visited his pub and if they have been there more than five times in the last month, they are refused a drink. Casual visitors, on the whole, put up with it.
On the other side of the border, you see a pub with a large billboard with the words, “Come drink with us, no restrictions, no lines”. This pub serves drinks in whatever glasses it likes, pours as fast as bar staff can, and has never once counted a customer’s drinks or the times they have been there. Nobody asks for an ID. Meanwhile, the landlord in the first pub, without quite meaning to, is running the best advertising campaign the pub on the other side of the border has ever had.
This is roughly the argument of a paper I am presenting at the EASG conference in Copenhagen this week, though the paper calls it “channelisation” rather than “the pub across the border”, because it was written for the gaming industry rather than for people who would simply like a pint without being measured for it. Channelisation is the share of gambling that stays within the licensed, supervised market. Every rule that makes the licensed market safer also makes it slightly less convenient — friction! And every gambler who finds that friction one step too much has somewhere else to go.
The comfortable assumption has always been that in the face of these regulations, gamblers simply gamble less. Without question, some do. But the evidence from six jurisdictions suggests a good number of them cross the border instead, to an unlicensed operator who has never thought about a deposit limit and has no intention of finding out more.
The Netherlands offers a case study and a policy analyst could not hope for a better one. In October 2024, the Kansspelautoriteit, the Dutch gambling regulator known as the KSA, introduced monthly deposit limits. Within months, channelisation fell below 50%. By the end of 2025, by some counting, the illegal market had overtaken the licensed one in revenue and the KSA’s own annual report conceded that the deposit limits might be helping this migration.
Sweden is quite candid about the impact of its regulations. Its regulator, Spelinspektionen, publishes channelisation figures every year, currently around 85%, and says what other regulators know but are not saying: “A high level of consumer protection within the licensing system may therefore lead to lower channelling.” A regulator has admitted, in its own published report, that its safeguards carry a cost.
Sweden’s self-exclusion register tells a related story. A meaningful share of the people who signed up to stop gambling later turn up gambling on unlicensed sites instead. The register worked exactly as intended; it simply excluded them from the one part of the market with any safeguards left, rather than from gambling itself.
I like to think that demand is rather like those balloons a party entertainer makes animals from. You squeeze them here (friction), they expand over there.
Italy took the most direct route of all. In 2019, it banned gambling advertising outright in every medium. Six years on, the unlicensed online market is estimated at roughly the same size as the legal one and AGCOM, Italy’s communications regulator, has noted that illegal gambling advertising simply moved to social media, which the ban does not reach. A total ban on advertising for licensed casinos, alongside no restriction at all on the unlicensed ones, has obviously not helped anyone except the unlicensed ones.
The United Kingdom has managed something rarer still: an unlicensed market with its own marketing department. Search engines and social media in Britain push advertisements for “Non GamStop casinos”, aimed squarely at people who have placed themselves on GamStop, the national self-exclusion scheme, by name. The invitation is very public, because it is the licensed operators who are bound by advertising rules, not the ones recruiting the self-excluded.
Germany contributes a statistic that ought to embarrass someone. Its regulator, the GGL, puts channelisation at 77%. H2 Gambling Capital, an independent industry analyst, puts the licensed share at somewhere between 36% and 60% for the same market in the same year. That is two credible sources unable to agree whether the illegal market is a rump, albeit a large rump, or a majority. Nobody can manage a market whose size is a matter of opinion.
Australia rounds out the tour with the most uncomfortable figure in my paper. Half of Australians gambling on offshore sites are simultaneously registered on BetStop, the national self-exclusion scheme. They are people the licensed market formally identified as needing protection. These are not curious newcomers who accidentally stumbled on a site.
Not everyone who ends up there meant to. Sweden’s own surveys find that a good many gamblers cannot say with any confidence whether the site they are using holds a domestic licence at all.
Put these six countries together and a pattern is hard to dismiss as coincidence. The gamblers who leave the licensed market are not a random sample of the gambling public. They tend to be the ones spending the most and worrying regulators the most, precisely because they have the strongest reasons to look for a platform that asks fewer questions.
None of this is an argument against safer gambling regulation and my paper is careful to say so. Some gamblers genuinely do gamble less because of these regulations. But the argument is narrower and I think harder to ignore: A regulator that reduces losses on the licensed market, while overlooking the fact that its highest-risk customers are migrating to a market with no protections at all, has relocated the harm to somewhere it can no longer see, not reduced it. That is a different — and smaller — achievement.
Public-health researchers have mostly left this territory alone for reasons that are understandable, even if the silence is not especially helpful. Most of the data on the size of the unlicensed market comes from industry sources with an obvious interest in making the number look large, which makes it awkward and unpalatable to cite.
Population-health frameworks are built to measure prevalence and harm, not market share, so channelisation sits slightly outside the discipline’s usual purview. And there is a genuinely uncomfortable admission buried in the argument: Some of the people most in need of protection are also the ones most determined to escape it. Saying so risks sounding like blame, when the more accurate reading is that this behaviour was entirely predictable and the regulator should have planned for it.
My suggested fixes are not particularly glamourous, which is rather the point.
First, an agreed method for measuring the size of the unlicensed market, so that regulators and researchers are at least arguing over the same numbers and trends can be measured.
Second, channelisation rate as a standard line in every regulator’s annual report, alongside problem-gambling prevalence and enforcement statistics. It is a KPI by which a regulator is measured. A policy that enlarges (and ignores) the unlicensed market cannot be declared as a success.
Third, new rules introduced as trials, with a stated goal, a way of measuring whether it was met, and an actual mechanism for scrapping the rule if it was not.
None of the six regulators discussed here can single-handedly shut down an offshore casino registered in Anjouan and that was never my ask. My ask is smaller and arguably harder to refuse: A regulator adding a new safeguard should be required to ask, and answer, how many of its most vulnerable customers that new safeguard is likely to send across the border. At the moment, most regulators can report a drop in losses on the licensed market as an unqualified win, without ever checking who moved and where they ended up.
Back at the pub on this side of the border, the landlord pours carefully and never misses an ID. Business is tough. What he knows is that for years, the pub across the border has been thanking him for every customer his assiduous following of the rules has sent across. I argue for keeping at least half an eye on who is crossing and finding out whether smaller glasses and faster pouring really do help those who should not be drinking.
