FAQ: a skeptic’s questions—answered with numbers

Who confirms that the sale happened?

In the first version, nobody—because there’s nothing to confirm. Antistaking starts only with things that are sold directly on the blockchain: NFTs and tokens. The lot is listed through a contract, and the sale is the transaction itself: the buyer pays, and the item goes to the buyer. The contract sees this on its own. The item stays locked in the contract for the whole life of the lot, so you can’t sell it outside the platform.

You can buy from yourself through a second wallet—that’s exactly the “fake sale,” and it costs the seller more than giving up honestly (withdrawing the lot early and paying what has dripped plus a small compensation payment). See the numbers in the “Cheaters” section below, or check them yourself in the demo.

Cars, apartments and tickets come later, once there’s a reliable way to confirm a sale that happens off the blockchain. One rule for that is already clear: whoever confirms the deal must never be paid a percentage of sales volume—otherwise they’d profit from “confirming” as many sales as possible.

What if it sells below the listed price—say, at 90%? Does that count as a sale?

In the first version, that can’t happen. The item sits in the contract, and the contract hands it over only to someone who pays the listed price or more. Want to sell for less? Give up early (withdraw the lot) and sell it anywhere you like.

One option for the future: allow a discount, but pay everyone except the seller as if the item had sold at the listed price, with the seller covering the difference out of their own pocket. Then “sell to yourself at a discount and take back the deposit” gets you nothing. For now this is just an idea, not a rule: we have to run the numbers first.

What stops the seller from “selling” the lot to themselves and walking off with the crowd’s money?

The price of cheating. A fake sale costs the seller three things: the platform fee on the price; the part of the dripped deposit that isn’t refunded on a sale (60% for NFTs, 45% for expensive items); and the announced reward, which is paid to supporters no matter what and never comes back to the seller. Giving up early (paying what has dripped plus a small compensation payment to doubters) is always cheaper. How much cheaper: a fake costs the seller 2.9–4.7% of the price more than giving up, even if the cheater picks the most favorable day before the deadline (the figure depends on the type of item).

And a fake “sale” doesn’t put anyone else’s money in the seller’s pocket. Doubters lose only earnings that weren’t theirs yet, and their stakes stay intact. Supporters get their stakes back plus the reward, and they get back the same share of their dripped drops as the seller gets of the deposit (40% for NFTs). The platform keeps the rest, not the seller.

What if the seller poses as a crowd of supporters and wins their own reward back?

They can—and it’s the hardest attack to kill (“planted supporters,” or shills). But planted stakes pay the drop and the entry fee, just like real ones, so the gain is capped: the cheater can shortchange their real supporters out of part of the promised reward, but no more than a few percent of the price (on an NFT lot in the demo, you can’t squeeze out more than about 3.6%). Plant too many stakes, and the drop eats up the profit.

The next step to close this gap: before paying out the reward, check whether the supporters’ money came from the seller’s own wallets (that’s visible on the blockchain), and take the seller’s reputation into account. This is a residual risk, and we say so openly.

A fake sale and a planted crowd at the SAME TIME?

We ran the numbers on that combo too. In a fake sale, the planted supporters get back both the reward and most of their drop (85% in the first draft), and that nearly broke the defense. The fix: we cut the share of what dripped that is refunded on a sale. For expensive items and real estate, 55% now comes back, not 85% as in the first draft.

After that change, the combo costs the cheater 0.8–2.1% of the price more than giving up, even if they pick the number of planted stakes and the day that suit them best. This check is built into how we choose the settings for each type of item.

What about betting AGAINST yourself from a second wallet and sinking the sale?

That’s “betting on your own failure”: the seller bets “against” themselves, sells nothing, and at the deadline collects their own dripped deposit and the supporters’ drops from the pot. In a live market this backfires: a lot that’s clearly doomed, with a fat pot, draws real doubters like a magnet. Their money dilutes the cheater’s share, and the cheater loses almost the whole deposit (on the demo lot, 14–15% of the price versus 0.75% if they had given up right away). The bigger their own stake “against,” the more outside doubters it takes to make the scam cost more than the honest path. You can see this with the sliders in the demo.

A caveat: in a completely empty market, arithmetic alone can’t close this attack. It’s the only place where the defense relies on other people’s money being around, and we say so plainly (more in the question about an empty market below).

If I split my stake across a hundred wallets, can I game the payouts?

No. Every payout is split strictly in proportion to “how much you put in × how long you held it,” with no thresholds and no per-wallet bonuses: the payout to a hundred wallets equals the payout to one to the 12th decimal place (we checked this by calculation for every event: twice the stake means exactly twice the payout). Splitting up is pointless—and so is pooling everything into one whale wallet.

What if I keep listing lots and canceling them—can I wear everyone out?

Doing it just to cause damage is expensive: that game costs the cheater about 9 times more than the harm they do, and everything they pay goes to the doubters, so they end up funding the other side.

The clever version is “list-and-cancel”: the seller bets “against” themselves, keeps listing and pulling lots, and pockets the drops of the honest supporters they lured in (the cancellation compensation they pay goes to themselves). Here, a gap does remain. If real supporters join the lot, the cheater makes about 0.4–1% of the price per cycle. The fix: the history of cancellations is public, so people stop trusting a serial canceler. There’s also a small fee for listing a lot—it has been designed but isn’t switched on yet (the author’s call).

A lot split into parts: can I buy the parts back from myself one by one, like climbing a ladder?

That won’t help. Each part carries its own share of all the costs of a fake, so a ladder is just several small fakes, and each of them costs more than giving up. In every variant we checked on the demo lot, the ladder costs 9.0–14.7% of the price more than giving up right away.

I’m a doubter. What can I lose?

Your stake itself: never, no matter what. This is a core rule: it is checked on every run, and if the math breaks it even once, we redo the mechanics. But a doubter’s vote isn’t free: you pay the entry fee (a fraction of a percent), and your money is frozen while your stake is in (you can leave at any time, and your stake comes back right away). Your earnings from the pot become yours only if no sale has happened by the deadline. What that means in dollars for a $100–500 stake is on the home page. That’s why a doubter’s signal is weaker than the seller’s deposit: the seller risks the money itself, while you risk only time and a fee.

I’m a supporter. What can I lose?

Only the drop—a small share of your stake (3–8% over the life of the lot, depending on the type of item) that goes to the doubters’ pot if the lot doesn’t sell—plus the entry fee. You know the size of the drop before you join; the rest of your stake comes back. Take the Lamborghini example on the home page (a car is an off-blockchain item, which comes later; the first version covers only NFTs and tokens—see the first question): a blogger puts in $2,000 and, if the lot fails, loses a $60 drop. The drop works as a filter: backing “it will sell” only makes sense if you believe in the sale strongly enough to justify the drop. There’s an example for a $100–500 stake on the home page.

Where do the payouts come from? Isn’t this a pyramid scheme?

Every payout comes from money that was paid in earlier, for that very purpose: the doubters’ pot is filled from the seller’s deposit and the supporters’ drops; the supporter reward comes from the proceeds of a sale that actually happened. Payouts can’t exceed what was paid in—that’s built into the math. New participants never pay off earlier ones.

The platform earns from fees and from the part of what dripped that isn’t refunded on a sale, so it earns the most when the item sells, not when someone loses (in the Lamborghini example, about $15,900 versus $3,100). That’s by design: the platform has a reason to bring in buyers, and it can’t confirm a sale on its own—the contract does that. The legal line between a sales commission and a pyramid scheme is well established: a pyramid pays for recruiting new participants, not for selling to a real buyer. Here the reward is paid only for a sale to an outside buyer, which puts it on the safe side of the line.

Source: the US test for pyramid schemes from the FTC’s Koscot case, applied in Webster v. Omnitrition (9th Circuit, 1996) and reaffirmed in FTC v. BurnLounge (2014).

Will there be a token? Aren’t points just laying the groundwork for a token giveaway?

There is no token, and there won’t be one until the mechanics are proven. Points are participation points that the platform gives to early participants, in proportion to the fees they pay. So anyone who churns fake volume to farm points pays for those points themselves. The mechanics have to work with no points at all.

What does it cost an honest seller? More than a real estate agent?

About the same as the usual middlemen: toward the upper end of their range, and a bit above it for tickets. If the item sells, the honest seller pays 7.9% of the price for cars and art (consignment, i.e. selling through a dealer or gallery: 5–10%), 5.3% for real estate (an agent: 3–6%), and 10.9% for tickets and editions (a launch platform’s fee: about 10%).

An NFT launch costs 14.9%, versus 10–15% for a marketplace plus marketing. That’s the very top of the range, and for NFTs it’s a lot. It’s the price of a short deadline and a big reward for supporters. So Antistaking doesn’t replace a regular NFT marketplace for ordinary sales: it’s for lots where a public debate over “will it sell or not?” is worth something in itself.

The difference from an agent: here everyone involved has a reason to sell faster. The deposit drips every day, and a faster sale means the same supporter reward for fewer days of waiting.

“Money that loses value if you sit on it”—isn’t that the Wörgl legend?

In 1932, Wörgl, a small Austrian town, issued local money that lost value if you didn’t spend it. There’s a nice legend about it, and we debunked it ourselves: the famous numbers (“the money circulated 20 times faster”) aren’t backed by primary sources, and you won’t find them anywhere in our materials.

What survived is a duller but more useful fact: in the US in the 1930s, similar local money worked when the fee was charged for the time you held it, and failed right away when it was charged on every purchase. That’s why we have an iron rule: the deposit drips only as time passes, never in response to an action.

All three AI reviewers debunked the legend, and all three confirmed the fact. Source of the fact: Federal Reserve Bank of Cleveland, 2008.

Why not ban fake sales instead of putting a price on them?

Two reasons. Theory: it’s been proven that a mechanism like this can’t have all three of these properties at once—simple rules, “telling the truth pays best,” and protection against the seller’s planted stakes. You can pick any two of the three; we chose simplicity and protection against fakes. The fake sale is fully closed, while planted stakes “for” yourself remain a limited gap (see the question about the planted crowd above).

Practice: on the blockchain you can’t verify who someone is, so a rule like “the seller may not…” can’t be enforced—only a price can. A ban needs someone to enforce it; sending the bill doesn’t.

Source: the theorem by Komo, Kominers and Roughgarden (EC’25 conference, peer-reviewed); three AI reviewers confirmed it unanimously.

Has anyone done this before? Why didn’t it work for them?

The whole combination—no one (we searched from five different angles). But each piece has been tried on its own by other projects: Kleros (a blockchain “court” for disputes) has a deposit that anyone can challenge; the Origin marketplace paid a reward for actual sales (but its promoters risked nothing—our drop closes exactly that hole); and Boson has paid cancellation of a deal.

We studied the failures too. The NFT marketplace LooksRare handed out rewards for trading volume, and 94.5% of that volume turned out to be fake; here the reward is the seller’s money, not newly minted tokens handed out for free. The NFTPerp exchange paid out more than was paid into it; here that’s impossible—the math rules it out. Polymarket’s short five-minute markets were exploited by people who knew how to nudge the outcome; our defense against that is a long enough lot deadline.

More about each project on the “Idea review” page.

The crowd “promotes the lot”—but is that even worth it for them?

The short answer: with five or more supporters, not by itself. Our calculation showed that each of them is individually better off not trying and waiting for someone else to make the sale. A small supporter is a vote, not a sales team: the blogger with a $2,000 stake in the Lamborghini example gets about $50 as their share of the reward—more of an “I believed in it” badge than payment for a review.

So the sale is driven by the seller and a few big supporters (the seller’s partners, big fans), for whom their share of the reward is real money: a supporter with a $100,000 stake gets about $2,500. A crowd of small supporters still helps: it’s a signal the buyer can see—“people believe in this.”

There’s a fix designed for the crowd: a separate payment to whoever brings in the buyer. If a third of the reward goes there, most of the crowd’s effort comes back (78–94% of the ideal). But it isn’t switched on yet: it changes the math of cheating, so we’ll run the numbers first. More in “The honest truth about supporters” on the home page.

Basis of the calculation: the “contest for a prize” model (Nitzan, 1991).

What if the market is empty at launch, with nobody to bet “against”?

Yes, that’s the weak spot of the early days. Some attacks are closed by arithmetic alone and don’t depend on how many people show up: the fake sale, the self-buy ladder (a fake sale split into many small resales), and the sniper (someone who jumps in as a supporter at the last moment). But the defense against betting on your own failure relies on other people’s money being around: it only works once enough real doubters join the lot.

The platform puts small amounts on both sides of every lot, so that every stake has someone on the other side. But in an empty market that doesn’t stop a seller from betting on their own failure: the cheater can bet more “against” themselves, while the seed stake is fixed. There’s an idea for extra insurance—a seed stake that grows with the pot—but it isn’t switched on yet.

What have you NOT proven yet?

The research is finished. Two questions remain open:

(1) Can someone buy up other people’s right to the reward? Picture a service that says “I’ll pay your drop for you, and we’ll split the reward” (“drop outsourcing”). Nobody in the industry has a ready-made fix: something similar has already happened in other crypto projects, where middlemen bought up voting power based on “how much you put in × how long you held it.” It’s an open problem; we’ll work on it by running the numbers.

(2) The crowd of small supporters doesn’t put in effort—see the question above and “The honest truth about supporters”.

Beyond that, the first version has some limits. These aren’t open questions but deliberate choices: the defense against betting on your own failure in an empty market relies on outside money coming in (see the question above); the legal review has been postponed, so items that look like securities aren’t allowed in the first version; and selling below the listed price isn’t allowed in the first version either (see the second question). And a market where you can legally influence the outcome has barely been studied in theory—that’s why we calculated everything ourselves rather than taking anyone’s word for it. You can rerun the calculations yourself—the archive with the programs is on the “How we checked” page.

The case of buying up voting power: Curve (a token exchange) and Convex (a middleman that bought up its votes).