💬 Discussion
Two Greek thinkers examine an online casino tablet, two types of tokens and a legal ledger.

Ban Dual-Currency Casinos

A sweepstakes casino sells one currency that cannot be cashed out and gives away a second that can. Its slot-style games therefore count legally as a promotion, and gambling law does not apply. Since 2025 states such as New Jersey, Connecticut, California and New York have banned the model, while Louisiana's governor vetoed a ban on 12 June 2025 [1].

  • 3.6 billion dollars players in the states without a ban lose to sweepstakes casinos each year once prizes are paid out [4]
  • 7 states run licensed online casinos, so most players have no legal alternative within reach [2]

The proposal

Federal law bans online casino games that pay out prizes through a second, nominally free currency, and leaves ordinary retail prize promotions untouched.

The debate

Supporters say the sites run casino games without any of the safeguards a licence imposes: no real age checks, no betting limits and no help for problem gamblers. The operators' trade groups answer that these are free and low-cost social games that adults have enjoyed for years, and that a ban only pushes players toward illegal offshore sites. New York banned the sites in December 2025, and Maine followed in April 2026 with a ban of its own. Louisiana's governor vetoed a ban on 12 June 2025, arguing that the sites are already illegal and regulators already act against them.

  • “These unauthorized sweepstakes sites lack safeguards, such as age verification, betting limits, problem-gambling support, and oversight of payouts or fairness. This law closes a dangerous loophole and protects millions of New Yorkers from predatory, unregulated gambling schemes.”

    argues for: Joseph Addabbo, Democrat, New York State Senate Press release, 8 December 2025

  • “Instead, LD 2007 will ban free and low-cost social games that have been operating responsibly for over a decade, leaving the thousands of Maine adults who enjoy them without a safe, trusted outlet for this mainstream entertainment.”

    argues against: Sean Ostrow, Social Gaming Leadership Alliance Statement, 8 April 2026

The analysis

Ban Dual-Currency Casinos

69% of the weight is for, 31% against.

An argument's weight combines three questions: how big the effect is, how certain it is and how much the good at stake matters. The larger the slice, the more the argument weighs.

For

Against

On the current analysis, banning dual-currency casinos is a good proposal, and the clearest result in this area.

Very likely better

Very likely better: Under the assumptions used, the proposal brings more than it costs in more than 99 of 100 model runs.

 Overall effect: medium

How strongly the proposal works overall, in five steps from very small to very large, comparable across all debates. The ring is always full, even when the overall effect is small; this step says how large it is.

The analysis reaches the supporters' verdict, but mostly for a reason they mention less often: the money. The bulk of the weight is money changing hands: the losses that stay with households, and the same sum no longer reaching the operators. That money does more good in households under financial pressure than as income for a company that offers a few hours of a slot-style game in return. The safeguards supporters talk about count as well: fewer people with a gambling disorder is the third-largest item, just under a quarter the size of the money kept. The enjoyment the trade groups defend is real but small, and the move offshore adds little, because players would lose that money either way.

In detail

All arguments

Arguments for3 arguments

Money that stops leaving households

13points

Unlike the prediction market question, there is no cheaper channel waiting on the other side of this one. Most of the country has no legal online casino at all, so a ban does not move the spending — it mostly ends it. The money stays where it was.

Value 5 · Household budgetsImpact 4.8Plausibility 5.5
▸ Show reasoning & sources ▾ Hide reasoning & sources

Value

The stream is money staying in a household rather than reaching an operating company, priced at the middle of the scale like any other money. What the players keep is counted here at what it is worth to them, and what the operators no longer receive is counted separately, on the other side of this evaluation, because the two ends of a payment are not worth the same thing. The spending concentrates among people who can least afford it, which is why the dollar they keep counts above face value. The enjoyment they lose along with the spending is real and is counted as its own argument against this measure rather than netted away here. The value is the middle of the scale, because the stream is money and the position of the people holding it is priced in the Impact.

Impact

Players in the states that have not banned the model lose about 3.6 billion dollars a year to sweepstakes casinos once prizes are paid out — the analysts' 2026 base case of net revenue [4]. What moves is what is lost, not what is spent on coins: most coin purchases come back as prizes. Where the losses go when the model is closed decides the size. Seven states run licensed online casinos, so for most players there is no licensed alternative within reach; a fifth is assumed to move offshore, another fifth to sports betting or a licensed casino where one exists, and the remaining three fifths simply stops — about 2.1 billion dollars a year staying with households. That money counts at 1.3 rather than face value, because the losses concentrate among people under financial pressure. Losing a large share of one's spare money hurts more per dollar than losing a small one, and for the few heaviest players it would count higher still; at the average of this group, about 1,200 dollars a year against a household income near 58,000, the step is not reached and the standard figure is used. That gives a Normalised Impact of 4.84 on the scale of this evaluation. Maine authorised online casino games in January 2026 and had not launched them by October; a start within the five-year horizon is assumed, and with 0.4 percent of the population it does not change the shares used here [7][8]. The Impact is the largest in this debate and it is large for a structural reason: this is one of the few gambling measures with no cheaper channel on the other side of it.

▸ Show calculation ▾ Hide calculation
Net player losses to sweepstakes casinos net revenue after prize redemptions [4], at 1.15 dollars to the euro (setting, range 1.05 to 1.25); the analysts' base case already assumes operators leave California and the other states that have banned the model [4] 3.6 billion dollars a year, 2026 base case 3.1 billion euro-equivalent
× Share that simply stops Setting, range 30 to 80 percent: a fifth moves offshore and a fifth to sports betting or a licensed casino; online casino games are running in only seven states, with Maine authorised and not yet launched [2] 60 % 1.86 billion euro-equivalent
× What the money is worth to the households keeping it the losses concentrate among people under financial pressure; at about a thousand euro-equivalent a year against a household income near fifty thousand (about 1,200 against 58,000 dollars), the step to a higher figure for very large personal losses is not reached [6] 1.3 2.42 billion euro-equivalent
÷ Normalised Impact scale of this evaluation 500 million euro-equivalent a point 4.84
Score 4.84 Impact × 5 Value × 5.5 Plausibility ÷ 10 = 13 of 100

Plausibility

The market size is an industry estimate rather than a regulatory return, which is the main weakness: nobody files anything, so the 3.6 billion dollar figure is an analyst's base case, already revised down as states acted, with a low case of 2.8 and a high case of 4.55 billion [4]. The counterfactual is the current position in the states that have not banned the model. What is genuinely estimated is where the spending goes, and the assumption that three fifths of it stops rests on a structural fact rather than a measurement: online casino games are running in seven states and authorised in an eighth, Maine, which has not launched [7], so the substitute that exists for sports betting does not exist here. The confounder that would matter is offshore substitution, which is booked as its own argument against this measure rather than discounted here. Several states have already banned the model, so this could be measured and has not been. Reverse causation does not arise. The Plausibility is at the upper end of what a projection can carry: the structure is clear and the market figure behind it is an estimate rather than a return.

Evidence: Projection · plausibility at most 6 Evidence of effect: Definitional

The most addictive format, with no brakes

3points

Slot-style games produce more gambling disorder per player than any other product, which is why every licensed jurisdiction wraps them in exclusion registers, session limits and loss caps. A sweepstakes casino runs the same games with none of that, at any hour, from 18. Removing the product removes the exposure.

Value 9 · HealthImpact 0.8Plausibility 4
▸ Show reasoning & sources ▾ Hide reasoning & sources

Value

The stream is mental health: the compulsion itself and the depression, sleeplessness and suicide risk that travel with a gambling disorder. This site places it in the class it uses for life and health, one step below the top because it is a condition most people recover from. What is priced is the person's own state, not the money they lose, which is counted separately, and not the harm to the household around them, which nothing here measures well enough to price. That the games are marketed as free entertainment rather than as gambling does not change what happens to the person playing them. The value sits one step below the maximum: the stream is health, and health that can be regained.

Impact

Players in the states that have not banned the model lose about 3.6 billion dollars a year to sweepstakes casinos once prizes are paid out — the analysts' 2026 base case of net revenue, which already assumes operators leave California and the other states that have acted [4]. Spread across an estimated 12 million accounts that is a few hundred dollars each, but the losses are concentrated: about 3 million are set as regular players, in a range from 1.5 to 6 million, and it is among those that harm arises. No source counts them; the figure is constructed from the market size. In the national gambling survey, 17 percent of sports bettors and 24 percent of fantasy sports players report problematic gambling behaviour many times a year. Slot-style play is not a milder format, and 20 percent is used here, a transfer from other products rather than a measurement of this one [3]: 600,000 people. Removing the product outright is set to resolve or prevent the problem for a quarter of them, in a range from a tenth to nearly a half, since the rest move to another form of gambling: 150,000 people. The survey counts behaviour, not a diagnosis [3]. The 0.2 quality-adjusted years a year this site uses for a gambling disorder is therefore applied only to the clinical share, about 28 percent, from the 2.5 million adults with a severe problem against 5 to 8 million with a milder one [9]; the rest is set at a tenth of that. That averages 0.07 quality-adjusted years per person, in a range from 0.04 to 0.12: about 10,500 quality-adjusted years a year. The Impact is a sixth of the money kept in this debate and carries the heaviest weight; without it, the money arguments alone would still put the measure ahead.

▸ Show calculation ▾ Hide calculation
Net player losses to sweepstakes casinos net revenue after prize redemptions [4], at 1.15 dollars to the euro (setting, range 1.05 to 1.25); the analysts' base case already assumes operators leave California and the other states that have banned the model [4] 3.6 billion dollars a year, 2026 base case 3.1 billion euro-equivalent
= Regular players Setting, range 1.5 to 6 million: spending is heavily concentrated and most accounts are trivial of an estimated 12 million accounts 3 million people
× Showing problematic gambling behaviour a transfer from other products, not measured for this one, and behaviour rather than a diagnosis: between the 17 percent of sports bettors and the 24 percent of fantasy sports players who report problematic behaviour many times a year; slot-style play produces more disorder per hour than other formats [3] 20 % 600,000 people
× Helped by removing the product Setting, range 10 to 45 percent: the rest move to licensed casinos, offshore sites or sports betting 25 % 150,000 people
× Quality-adjusted years lost per person a year Setting, range 0.04 to 0.12: 0.2 for the clinical share, from 2.5 million adults with a severe problem against 5 to 8 million with a milder one; a tenth of that for the rest [9] 28 % × 0.2 + 72 % × 0.02 = 0.07 10,500 quality-adjusted years
× Value of the years the value of a healthy life year used across this site 40,000 euro-equivalent each 420 million euro-equivalent
÷ Normalised Impact scale of this evaluation 500 million euro-equivalent a point 0.84
Score 0.84 Impact × 9 Value × 4 Plausibility ÷ 10 = 3 of 100

Plausibility

The direction is as well supported as anything in gambling research and the sizes are not. That continuous slot-style play produces more disorder per hour than other formats is a consistent finding across decades of clinical and machine-level data; the counterfactual in that work is other gambling products rather than no gambling, which is exactly the comparison this argument needs. What has no source is the American sweepstakes population itself: the operators publish nothing and no regulator collects it. Every number specific to this market is a setting: the 3 million regular players, the 20 percent with problematic behaviour, the quarter helped and the loss per person; no study of dual-currency casinos measures any of them. The counter-mechanism is real and only partly answered — a player whose sweepstakes site closes has licensed online casinos in seven states, offshore sites everywhere, and sports betting in most of the country, which is why only a quarter of the affected group is assumed to be helped rather than most of them. Reverse causation runs the usual way for prevalence figures and is unresolved: people prone to a gambling problem seek out these games. The Plausibility is below the middle: the harm of the format is well established and everything specific to this market is estimated.

Evidence: Mechanism · plausibility at most 5.5 Evidence of effect: Associational · at most 5.5 Placement: Chain closed, unevidenced · plausibility 4–5

Open: The states that have banned the model since 2025 did so at different dates. Comparing helpline contacts naming online casino games in those states against the rest, before and after, would measure this directly and could carry P to 6.

Licensed operators stop competing with an untaxed rival

0.78points

A licensed online casino pays state tax, funds treatment, runs an exclusion register and files reports. A sweepstakes site does none of it and sells the same games. In the seven states where both are running, one of them is playing by rules the other has opted out of.

Value 5 · Public financesImpact 0.3Plausibility 5
▸ Show reasoning & sources ▾ Hide reasoning & sources

Value

The stream is state revenue, priced at the middle of the scale like any other public money. The players pay the same either way; what changes is that part of what they lose reaches a treasury instead of staying with the owners of an untaxed operator. Nothing is priced here for fairness between operators as such: a company disadvantaged by a rival's regulatory arbitrage has a complaint, but it is a complaint about its own profits rather than a public good. What makes this a gain is that the same money is worth more in a budget that funds something than as a receipt for which nothing was given back — the receiving end it comes from is counted as its own argument against. The value is the middle of the scale, the level this site uses for public money whatever it is spent on.

Impact

A fifth of the 3.6 billion dollars players lose is assumed to move to a licensed operator where one exists, which is about 720 million dollars of operator revenue [4]. Seven states run licensed online casinos, so the volume that can land somewhere taxed is small and concentrated. Maine authorised online casino games in January 2026 and had not launched them by October; a start within the five-year horizon is assumed, and with 0.4 percent of the population it does not change the shares used here [7][8]. Those states tax online casino revenue at rates from 15 to more than 50 percent of what operators keep; 25 percent is used, in a range from 15 to 50 percent, which gives about 180 million dollars a year of tax [2]. Public money counts at face value, the standard figure on this site, so the amount counted is the 180 million itself. The same 180 million no longer reaches an untaxed operator, and that is counted against this measure at one value step lower — the gain is the step between the two, not the whole sum. What is not counted is the compliance and treatment funding that comes with a licence, which is a cost to the operator rather than a public gain. The Impact is the smallest in this debate, because the licensed channel that could absorb the volume runs in seven states, with an eighth authorised, and not in the other forty-two.

▸ Show calculation ▾ Hide calculation
Net player losses to sweepstakes casinos net revenue after prize redemptions [4], at 1.15 dollars to the euro (setting, range 1.05 to 1.25); the analysts' base case already assumes operators leave California and the other states that have banned the model [4] 3.6 billion dollars a year, 2026 base case 3.1 billion euro-equivalent
× Share moving to a licensed operator Setting, range 8 to 35 percent: online casino games run in only seven states, with Maine authorised and not yet launched 20 % 620 million euro-equivalent
× Reaching a state treasury as tax Setting, range 15 to 50 percent: online casino tax rates run from 15 to over 50 percent of what operators keep [2] 25 % 155 million euro-equivalent
× Weight of a dollar in a state budget the standard weight for public money on this site 1.0 155 million euro-equivalent
÷ Normalised Impact scale of this evaluation 500 million euro-equivalent a point 0.31
Score 0.31 Impact × 5 Value × 5 Plausibility ÷ 10 = 0.8 of 100

Plausibility

The tax rates are published and the mechanism is arithmetic: revenue moving into a licensed channel is taxed at the licensed rate. The counterfactual is the current position, where the same spending generates nothing. What is estimated is how much moves, and the answer is bounded by a structural fact rather than a behavioural one — most players have no licensed online casino available to them, so the ceiling on this argument is low regardless of how players behave. The confounder that matters is that some of the spending would move to sports betting instead, which is taxed differently and in more states; that would raise this figure and is not counted, which makes it conservative. Reverse causation does not arise. The Plausibility is at the middle: the tax arithmetic is certain and the volume reaching it is small and estimated.

Evidence: Projection · plausibility at most 6 Evidence of effect: Definitional

Arguments against4 arguments

What the operators no longer take

6.1points

The money households keep is money somebody else stops receiving. Sweepstakes operators are privately held companies that pay no gambling tax, and what they collect is the difference between what players stake and what comes back as prizes. Closing the model ends that receipt.

Value 3 · Income without a returnImpact 3.7Plausibility 5.5
▸ Show reasoning & sources ▾ Hide reasoning & sources

Value

The stream is money a company takes in, and money counts in full wherever it lands — an operator's dollar is not worth less because of who owns the company. What places this below ordinary money is what stands opposite it. A grocer's revenue pays for groceries; a stake buys a few hours of a game worth a small fraction of what is lost, and for a large part of this revenue, which comes from people whose play is disordered, it buys nothing that person would choose. Income taken with so little given back is real income and is counted as such, two steps below the middle of the scale. The same money keeps the middle of the scale once a tax takes it and a legislature decides what it buys. The value is well below the middle, because the money is real and what is given for it is not.

Impact

The 2.1 billion dollars a year that stops leaving households is the same 2.1 billion the operators stop receiving; it is one payment seen from its two ends, and it is counted here at the same amount and the same plausibility as in the argument opposite [4]. Nothing is deducted for the cost of running the games, which is small beside the margin and is a real resource the measure also frees. The remaining fifth of the market that moves to a licensed operator and the fifth that moves offshore are not counted here: those losses change which company receives them, and one operator's receipt is worth what another's is, so the move nets out. What does not net out is the slice that reaches a treasury instead, which is counted in its own argument on both of its ends. The Impact is the second largest in this debate, and it is the necessary other half of the largest: money cannot stay with households without ceasing to arrive somewhere else.

▸ Show calculation ▾ Hide calculation
Net player losses to sweepstakes casinos net revenue after prize redemptions [4], at 1.15 dollars to the euro (setting, range 1.05 to 1.25); the analysts' base case already assumes operators leave California and the other states that have banned the model [4] 3.6 billion dollars a year, 2026 base case 3.1 billion euro-equivalent
× Share that simply stops the same share as in the argument opposite: a fifth moves offshore and a fifth to sports betting or a licensed casino [2] 60 % 1.86 billion euro-equivalent
× Weight of a dollar received money counts in full wherever it lands; what is given back for it is carried by the value, not by the weight 1.0 1.86 billion euro-equivalent
÷ Normalised Impact scale of this evaluation 500 million euro-equivalent a point 3.72
Score 3.72 Impact × 3 Value × 5.5 Plausibility ÷ 10 = 6.1 of 100

Plausibility

This rests on exactly the same figure as the money argument in favour and carries exactly the same plausibility, because it is the same quantity counted once from each end: if three fifths of the market stops, it stops for the operators too. The counterfactual is the current position in the states that have not banned the model. The market size is an analyst's base case rather than a regulatory return — nobody files anything — with a low case of 2.8 and a high case of 4.55 billion dollars [4]. What is estimated is the share that stops rather than moving, and that estimate is structural: online casino games are running in seven states and authorised in an eighth, Maine, which has not launched [7]. Reverse causation does not arise. The Plausibility is at the upper end of what a projection can carry, and is deliberately identical to that of the argument it mirrors.

Evidence: Projection · plausibility at most 6 Evidence of effect: Definitional

Millions of people liked playing them

0.7points

Most sweepstakes players spend a few dollars a month on a game they enjoy and stop when they are bored. They are not in difficulty and they did not ask to be protected. A ban takes the game away from all of them to reach the minority who are.

Value 4 · EnjoymentImpact 0.3Plausibility 5
▸ Show reasoning & sources ▾ Hide reasoning & sources

Value

The stream is enjoyment — a game people choose to play and pay for. This site places it in the class it uses for comfort and everyday pleasure, well below health or money in a strained household. It is not dismissed: an evaluation that treated a freely chosen pleasure as worthless because someone disapproved of it would be doing something other than measuring. But it is counted only where the choice is genuinely free. Where the play is already disordered, the payment is not a purchase and nothing is counted on this side for it; that part of the money appears only as the loss it is. The value is in the lower part of the scale, because what is lost is a chosen pleasure rather than anything anyone depends on.

Impact

What is lost is time spent on a game, so it is counted per player rather than as a slice of the money. About 3 million people play these games regularly; three fifths of them lose the product altogether when it closes, the same share of the market that simply stops. Of those, the play of roughly one in five is already disordered and is not counted here at all — for them the payment is the harm, not a purchase. That leaves about 1.44 million people losing a pastime they freely chose. What it is worth is set at what comparable entertainment costs for the same hours: a streaming subscription runs 12 to 17 dollars a month and a cinema ticket about 17 dollars, and about 140 dollars a year is used, in a range from 70 to 210. That gives a Normalised Impact of 0.35 on the scale of this evaluation, since what is lost is the pastime rather than money. Earlier versions priced this at three quarters of everything the players staked, which counted the losses of the heaviest players as though they were buying entertainment with them. What is not counted, and would raise this figure, is that much of the appeal is social — leaderboards, streaks, friends — and none of it transfers elsewhere. The Impact is a small fraction of the money gain it offsets, which is the honest shape of a prohibition: most of the people it stops were not in trouble, and what they lose is an evening's amusement rather than the sums that pass through the machine.

▸ Show calculation ▾ Hide calculation
Regular players Setting, range 1.5 to 6 million: spending is heavily concentrated and most accounts are trivial [6] of an estimated 12 million accounts 3 million people
× Who lose the product when it closes the share of the market that simply stops — the same figure carries the money argument 60 % 1.8 million people
× Whose play is freely chosen Setting, range 70 to 90 percent: the play of about one in five is disordered, and for them the payment is the harm rather than a purchase [3][5] 80 % 1.44 million people
× Comparable entertainment for the same hours Setting, range 60 to 180 euro-equivalent (about 70 to 210 dollars): a streaming subscription runs 12 to 17 dollars a month, a cinema ticket about 17 dollars; no source records how long these games are played 120 euro-equivalent a year (about 140 dollars) 173 million euro-equivalent
÷ Normalised Impact scale of this evaluation 500 million euro-equivalent a point 0.35
Score 0.35 Impact × 4 Value × 5 Plausibility ÷ 10 = 0.7 of 100

Plausibility

That people get something from these games is not in doubt; they return to them daily and pay for the privilege. What has no measurement is how long they play. No operator publishes session data, no regulator collects it, and the figure used here is a stated price for a year of a pastime rather than an observed one. The counterfactual is the same players under current rules. Pricing the hours at what other entertainment costs is the conservative choice available: it does not assume the game is worth what it takes, which for this product would be circular, and it does not assume it is worthless. The share treated as freely chosen follows from the share of players whose play is disordered, which is measured for online betting rather than for this product [3][5]. Reverse causation does not arise. The Plausibility is at the middle: the stream certainly exists and the hours behind it have never been counted.

Evidence: Plausibility · plausibility at most 5 Evidence of effect: Definitional

The taxed share leaves the operators too

0.47points

The part of the money that ends up in a state treasury also stops arriving somewhere. It is the same amount, seen from the end that loses it, and it is worth slightly less there than in a public budget.

Value 3 · Income without a returnImpact 0.3Plausibility 5
▸ Show reasoning & sources ▾ Hide reasoning & sources

Value

The stream is money an untaxed operator no longer keeps, counted in full like any other money. It sits two steps below the middle of the scale because so little is given in return for it: what the player receives for the stake is a few minutes of a game, not a good worth what was paid. The distinction from the public budget opposite is the whole content of this pair. The same 180 million dollars is worth two steps more once it funds something a legislature has chosen than it is as a receipt for a wager. That gap, and not the sum itself, is what the two arguments together actually show. Nothing is priced here for the operator's disappointment as such, which is a matter of business rather than a public loss. Nor is anything deducted for the cost of collecting the tax, which is small at this scale and would belong to the state's side of the ledger. The value is well below the middle, the level this site uses for income taken with little given back.

Impact

This is the other end of the tax argument and carries the identical amount: about 720 million dollars of losses moving to a licensed operator, of which 25 percent — about 180 million dollars a year — is taken in state tax [2][4]. Whoever would otherwise have kept that money keeps it no longer. Counted at full weight, as all money is, the amount is the 180 million itself, in a range from 110 to 360 million following the tax rate. Set against the public budget on the other side, what the pair actually shows is the two value steps between the two ends, which is why the public side of the pair outweighs this one rather than cancelling it. The Impact is the smallest in this debate and exists to keep the tax gain honest rather than to move the balance.

▸ Show calculation ▾ Hide calculation
Net player losses to sweepstakes casinos net revenue after prize redemptions [4], at 1.15 dollars to the euro (setting, range 1.05 to 1.25); the analysts' base case already assumes operators leave California and the other states that have banned the model [4] 3.6 billion dollars a year, 2026 base case 3.1 billion euro-equivalent
× Share moving to a licensed operator the same share as in the tax argument: online casino games run in only seven states, with Maine authorised and not yet launched 20 % 620 million euro-equivalent
× Taken in state tax rather than kept Setting, range 15 to 50 percent: online casino tax rates run from 15 to over 50 percent of what operators keep [2] 25 % 155 million euro-equivalent
× Weight of a dollar received money counts in full wherever it lands 1.0 155 million euro-equivalent
÷ Normalised Impact scale of this evaluation 500 million euro-equivalent a point 0.31
Score 0.31 Impact × 3 Value × 5 Plausibility ÷ 10 = 0.5 of 100

Plausibility

The plausibility is that of the argument it mirrors and is set equal to it deliberately, because it is one quantity counted from each end. The tax rates are published and the arithmetic is certain; what is estimated is how much volume moves into a licensed channel, and that is bounded by a structural fact — most players have no licensed online casino available to them [2]. The counterfactual is the current position, in which the same spending generates nothing for any treasury. Reverse causation does not arise. The Plausibility is at the middle, identical to the argument this mirrors.

Evidence: Projection · plausibility at most 6 Evidence of effect: Definitional

A fifth of it goes offshore

0.12points

Offshore online casinos have taken American customers for twenty years and are one search away. The players who look hardest for a replacement are the ones who were playing most. What they find has fewer protections than the model being banned, not more.

Value 5 · Household budgetsImpact 0.08Plausibility 3
▸ Show reasoning & sources ▾ Hide reasoning & sources

Value

The stream is money lost where nothing at all applies, priced at the middle of the scale. An offshore casino has no age check that means anything, no obligation to pay out, no segregated customer funds and no authority to complain to. What is priced here is only the additional harm of losing money somewhere with no recourse whatever; the losses the ban fails to prevent are simply absent from the money kept on the other side. The gambling disorder that follows a player offshore is counted in the argument on health rather than a second time here. The value is the middle of the scale, because the stream is money and the absence of protection around it is priced in the Impact.

Impact

A fifth of the 3.6 billion dollars players lose is assumed to move to operators outside any American jurisdiction, in a range from a tenth to two fifths — about 720 million dollars a year [4]. That share is already left out of the money kept in the argument in favour, and it goes from the same players to an operator in place of another, so the move itself changes nothing that is not already counted. What is added is what an offshore site can do that a domestic sweepstakes site cannot get away with: refuse to pay out, freeze a balance or disappear, with no court within reach. Five percent of what the migrating players lose is assumed to be taken that way on top, in a range from 2 to 15 percent: about 36 million dollars a year, counted at 1.3 like the players' other money, a Normalised Impact of 0.08 on the scale of this evaluation. Nothing is counted on the other side for it: money taken by refusing to honour a bet is not a receipt anyone is entitled to, and it is not set against the loss. The players who look hardest for a replacement are the heavy ones; for their health that is already reflected in the argument that helps only a quarter of the affected group. The Impact is the smallest in this debate: moving offshore mostly means losing the money in a different place, and that money was never counted as saved.

▸ Show calculation ▾ Hide calculation
Net player losses to sweepstakes casinos net revenue after prize redemptions [4], at 1.15 dollars to the euro (setting, range 1.05 to 1.25); the analysts' base case already assumes operators leave California and the other states that have banned the model [4] 3.6 billion dollars a year, 2026 base case 3.1 billion euro-equivalent
× Share moving offshore Setting, range 10 to 40 percent: offshore sites lack the app-store distribution and social features that made this model popular 20 % 620 million euro-equivalent
× Withheld winnings and frozen balances on top Setting, range 2 to 15 percent of what the migrating players lose: no obligation to pay out and no court within reach 5 % 31 million euro-equivalent
× What the money is worth to the players losing it taken from players under financial pressure; nothing is set against it, because money kept by refusing to pay out is not a receipt anyone is entitled to 1.3 40 million euro-equivalent
÷ Normalised Impact scale of this evaluation 500 million euro-equivalent a point 0.08
Score 0.08 Impact × 5 Value × 3 Plausibility ÷ 10 = 0.1 of 100

Plausibility

The pattern is old and the size is unmeasured. Offshore online casinos have accepted American customers for two decades and payment blocking has never stopped them entirely; crypto settlement has made it easier again. The counterfactual is the current position with a domestic alternative available. The chain is short and every link visible: the product closes, the player wants to continue, an offshore site is reachable. What no source supplies is the share. Several states have banned the model at different dates since 2025, which is exactly the variation that would answer this, and nobody has used it [1]. The counter-mechanism is partly answered: offshore sites lack the app-store distribution and social features that made the sweepstakes model popular, so the substitution is imperfect, which is why the share used is a fifth rather than a half. Reverse causation does not arise. How often offshore sites keep what they owe has no systematic measurement either. The Plausibility is low because neither the migrating share nor the withheld share has been measured, although the state bans since 2025 would allow the first.

Evidence: Mechanism · plausibility at most 6 Evidence of effect: Mechanistic · at most 6 Placement: Chain open · plausibility 3–3.5

Open: App download and payment data by state, before and after each state ban, would give the migrating share directly and could carry P to 6.

How robust the result is

Under the assumptions used, the proposal brings more than it costs in more than 99 of 100 model runs.

gains more than it costscosts more than it gains

Each dot stands for 200 of the 20,000 model runs.

How much of the money is freely spent decides the order of the arguments, together with what it is worth where it lands. The share of losses that comes from people whose play is already disordered is treated as no free choice at all, and no entertainment is counted against it. Connecticut's impact study puts 51 percent of sports betting revenue and more than 70 percent of all gambling revenue on the small group with a problem or at risk of one [5]. Slot-style play is the format most closely tied to disorder, and 69 percent is used. Where the money lands, an operator's receipt counts in full, but a dollar taken without anything much given back counts one step below ordinary money. The ban comes out ahead in practically all runs.

Show the distribution of the model runs

20,000 simulated futures

Very likely better · 100 %

In more than 99 of 100 simulated futures the proposal brings more than it costs.

No effect is known exactly; every argument has a range. The calculation draws a possible size for each argument from its range 20,000 times and counts how often the benefits exceed the costs.

Net effect in points per future. Left of zero the drawbacks outweigh the benefits, right of zero the benefits win. Eight in ten futures fall between +6.9 and +13 points.
Scale and assessment

Scale of this evaluation: Normalised Impact, unitless and calibrated to this topic. For comparison, one point here corresponds to about 500 million euro-equivalent a year.

Size class:medium

Effect medium (+4.8) · 93 % of the effect is positive

Areas show individual argument weights. The positive share of the effect nets out money transfers between the two sides; the probability of a positive net effect is determined separately.

How we score →

Review and sources

AI evaluation · not yet reviewed by a human

This evaluation was produced and sourced by an AI model; a human review is still pending. Figures and conclusions may still change. The review log is at the foot of the page.How review works →

Last reviewed by Claude Opus 5.5 · October 29, 2026 · 10× AI, not yet reviewed by a human
  1. October 29, 2026AI reviewClaude Opus 5.5record updated

    Transfer pairs pro-2/con-4 and pro-3/con-5 linked by gegenbein, con-2 booked net with netto_grund; balance recomputed, category unchanged.

  2. October 29, 2026AI reviewClaude Opus 5.5record updated

    Page texts for the new page layout (draft G) with quotes from primary sources; numbers unchanged.

  3. October 21, 2026AI reviewClaude Fable 5.1record updated

    Money amounts in argument texts stated in dollars per style-en.md; scores, chains and balance unchanged

  4. October 20, 2026AI reviewClaude Fable 5.1record updated

    Summary and scale note rewritten to the style of 20 Oct 2026; numbers unchanged

  5. October 14, 2026AI reviewClaude Opus 5.5re-scored

    Prüfbericht 11.10.: Maine (LD 1164, Jan. 2026) als autorisiert, nicht gestartet ergänzt; pro-1 Setzungen sichtbar (3 Mio. Nutzer, 20 %, Schwere), NGAGE-Verhalten ≠ Störung (0,07 statt 0,2 QALY), P 4,5 → 4; r 0,76 → 0,70, Kategorie bleibt besser.

  6. September 16, 2026AI reviewClaude Opus 5re-scored

    Anbieterempfang nach Julians Entscheidung vom 16.09. von V 4 auf V 3 gesenkt (con-4 und con-5): Pro 11,90 zu Con 3,71, r 0,71 auf 0,76 — Kategorie nach altem Maßstab unverändert besser.

  7. September 16, 2026AI reviewClaude Opus 5re-scored

    Glücksspiel-Regel 16.09.: Buchungsregel gluecksspiel-gegenleistung/1.0: Empfängerseite Gewicht 1 statt 0,5 — Spielerbein (V 5) und Betreiberbein (V 4) als Paar gebucht, Unterhaltung aus Spielzeit statt drei Vierteln des Einsatzes (i 0,84 → 0,17); r 0,84 → 0,71, deutlich besser → besser.

  8. September 14, 2026AI reviewClaude Opus 5re-scored

    Gesamtprüfung 08.09. nachgearbeitet: Basis auf Nettoumsatz nach Preisen (EKG 2026: 3,6 Mrd $) statt unbelegter 7,2 Mrd; con-3 (Preisausschreiben, außerhalb F1) entfällt, con-2 nur Offshore-Mehrschaden, pro-3 Steuer netto, con-1 ohne w; r 0,77 → 0,84.

  9. September 6, 2026AI reviewClaude Opus 5record updated

    i_spanne an allen 6 Argumenten, normalisierung erstmals an allen 6 (nur globale Anker), kein gegenbein (beide Transfers genettet), massstab_hinweis ohne r, drei deutsche Scoring-Token aus der englischen Prosa entfernt, Wechselkurs als Setzung mit Spanne benannt. Kategorie steigt von Besser (r 0,77) auf Deutlich besser (P(D>0) 0,99).

  10. September 6, 2026AI reviewClaude Opus 5First evaluation

    Created for the English side: no licensed alternative exists in 43 states, which is why this ban ends spending rather than moving it.

Evaluations are produced with AI support and reviewed on a schedule for new developments; human passes are marked separately.How we review →

Show all 9 sourcesHide sources
  1. InfoLawGroup: Sweepstakes Casino Laws in 2026: New Bans, Proposed Legislation, and Regulatory Trends. infolawgroup.com
  2. Sports Handle: Legal US sports betting and online casino revenue, handle and state tax database. sportshandle.com
  3. National Council on Problem Gambling: National Survey on Gambling Attitudes and Gambling Experiences 3.0. ncpgambling.org
  4. Sweepsy: Eilers & Krejcik Forecasts Downturn For Sweepstakes Casinos. sweepsy.com
  5. Connecticut Department of Mental Health and Addiction Services (Gemini Research): Socioeconomic Impacts of Gambling in Connecticut: Final Report. portal.ct.gov
  6. Track360: Sweepstakes Casino Market Statistics 2026. track360.io
  7. Maine Legislature: LD 1164, An Act to Create Economic Opportunity for the Wabanaki Nations through Internet Gaming (Public Law chapter 538, January 2026). mainelegislature.org
  8. Maine Gambling Control Unit: Internet gaming. maine.gov
  9. National Council on Problem Gambling: National Survey on Gambling Attitudes and Gambling Experiences and Helpline Annual Report. ncpgambling.org

More debates in this area

Unclear · 35 %
Size: small
Close the Event-Contract Gap

Say in federal law that a contract paying out on the result of a game is a bet, and let the states regulate it as one.
Very likely better · 98 %
Size: medium
Federal Sports Betting Standards

Set a federal floor under state sports betting: a limit on deposits, affordability checks, no bonus bets, and no advertising during play.
All debates in Gambling and Betting →