Who pays when a post stays sealed

Ordinary crowdfunding punishes the people who believe first: pledge to something that stalls and your money sat frozen for a month for nothing, so the sensible move is to wait and see what everybody else does. When everybody waits, things everybody wanted die quietly. Economists call it the free-rider problem. It is not a failure of enthusiasm; it is a failure of the contract.

What is a dominant assurance contract?

In 1998, the economist Alex Tabarrok published a fix that is one clause long: if the campaign fails, every backer is refunded plus a bonus.

Look at what that does to a backer’s reasoning. If enough other people show up, you back it and get the thing. If they do not, you back it and get paid for having shown up. There is no belief you can hold about the campaign’s chances that makes sitting it out the better move. In game-theory terms backing has become the dominant strategy no matter what anyone else does, which is where the name comes from: a dominant assurance contract.

The insight is that the bonus is not a consolation prize. It is a price paid to buy information out of the crowd. People stop hedging, the true level of demand becomes visible, and, in the ordinary case, the bonus is never actually paid because the thing funds.

Who posts the bond?

A promise to pay backers on failure is worth nothing unless the money is already there. Somebody has to put it up, and in Tabarrok’s paper that somebody is the entrepreneur proposing the project. They do two jobs: propose the work, and escrow the money that pays everyone a bonus if the goal is missed.

That fits the world the paper describes: an entrepreneur with capital, providing a public good, who can reliably escrow the failure payout. For that person the bond is a cost of doing business.

It fits a social network badly. Somebody sealing a post to raise $100 usually does not have a spare $25 to lock away behind it; if they did, the post might not need the goal. Requiring the creator to post the bond would quietly restrict the whole mechanism to people who already have money, which is a poor look for a design whose entire purpose is to get good work funded.

Change one: the bond is an open pool

openmog keeps Tabarrok’s contract intact on the backer’s side and changes one thing on the other side. The bond is an open pool, and anyone can fill it.

When a creator seals a post they set a goal, a deadline and, off by default, a bonus percentage. A post that leaves the bonus off runs as plain all-or-nothing crowdfunding; this post is about what changes when one is on. Those choices fix the pool’s cap at the bonus percentage times the goal, and while the post is live anyone can stake into it: people who know the creator, people who have watched their audience for years, the creator themselves. Then the contract settles both ways.

• If the post unseals, the pool is released and each bonder is credited what they staked, paid out of the raise. openmog’s fee, 10% of each payment with a 75 cent minimum, is included in the price each backer confirmed, so the creator’s goal arrives whole with only that bonus taken from it.
• If it stays sealed, the pool pays the backers their bonuses on top of holds that were never charged. Each payer’s price includes the fee, and the bonders’ charges are now the only payments, so their all-in charges are what carry it. One sealed post, two endings runs both with the actual dollars.

From a backer’s seat the guarantee has not moved: fund it and the post opens, watch it fall short and the pool pays you for having shown up. What changed is who underwrites it.

It changes one more thing, and it is the cost of the design. Tabarrok’s bond is posted before the campaign opens, so the bonus is guaranteed from the first pledge. An open pool is filled during the run, or not at all, which means a post nobody bonds pays no bonus. Backers on such a post are still never charged when it falls short, but they are not paid either. openmog buys the mechanism’s reach with the certainty of its bonus, and the stake sheet shows you which of the two you are looking at before you commit a cent.

• The creator sheds the capital burden. They run a full dominant assurance contract without escrowing a dollar, and a post that stays sealed costs them nothing extra.
• Someone other than the creator can hold the promise. The original design had no place for a supporter who wanted to make the creator’s guarantee real without being the creator. A bonder is that person: their money is what pays the backers’ bonus if the post falls short, and they are paid for standing behind it. That job has its own post.
• Backers get a second signal. A pool that other people fill tells backers something the original could not: someone besides the creator thought this post was worth guaranteeing. And every dollar of it is the very bonus that makes backing safe.

Change two: every payout is capped at its advertised rate

The second change is smaller to describe and just as load-bearing. It came out of asking one rude question: what is the cheapest way to take money out of this?

The obvious way to settle a failed campaign is to split the forfeited pool pro-rata across the backers. It sounds fair, and on a busy post it is. But consider a post with a $25 pool that ends with exactly one backer, who backed one cent. Pro-rata hands that person the entire $25. The mirror image exists on the other side: split the bonus pro-rata across bonders and a single one-cent bond, alone in the pool, collects all of it. Both are large payouts for the smallest possible stake, and both are trivial to automate against every post on the network.

So openmog caps each payout at the rate the post advertised, applied to the participant’s own stake. A backer’s bonus is the lesser of their pro-rata share and the advertised percentage of what they backed. A bonder’s earnings are their share of a bonus that can never exceed the pool that filled. And because the pool only ever pays out what the bonuses cost, bonders on a post that stays sealed are charged the all-in amount that covers exactly that, and no more, with the rest of the hold released.

Three things follow, and they are why the cap is a repair rather than a restriction. The cap costs ordinary participants nothing: at full participation the capped numbers and the uncapped numbers are identical, because it only binds in the cases where a pot was about to be handed to someone for showing up small. It makes the mechanism indifferent to how many accounts you control, since ten one-cent stakes now earn exactly what one ten-cent stake earns. And it puts the design back where the literature had it: Tabarrok’s refund bonus, and the provision-point bonuses studied since, were alwaysrates on the pledge, never slices of a pot.

What do the two changes leave alone?

Neither change touches the clause the whole thing rests on. Back a sealed post and you cannot end up out of pocket: either it opens and you paid for something you wanted, or it stays shut and your money is released without ever having moved. What the pool adds on top of that floor is the bonus, sized by what other people were willing to put behind it. Everything above is about arranging for somebody trustworthy to stand behind the promise, and about making sure the promise cannot be farmed.

The arithmetic, with actual dollars in it, is in one sealed post, two endings. The second way a sealed post can open, with no goal and no bond at all, is pay to reveal. Or go look at a live one.