Bet cancellation clarifications

BitBet FAQ says :

Can bets be cancelled?

Bets can indeed be cancelled, but only if BitBet judges that material changes make the bet meaningless. Once a bet is cancelled all BTC bet is refunded at the respective requested payout addresses.

Bets will not be cancelled because it became impossible for one side to win - in such cases bets are closed and paid early instead. Bets will not be cancelled simply at the request of a third party. We do review comments made for cases where bets should indeed be cancelled or closed early.

To shed a little light and add more detail to that, we take a very narrow view to bet cancellation, on the general principle that once a user entrusts his BTC to BitBet on a proposition, he expects a resolution rather than an excuse. Moreover, it is the user's responsibility to filter what he bets on - the "it wasn't clear to me" excuse may work well in school, or in the US, but it has no place among adults. Consequently, all bets should be resolved one way or the other.

Also worthy of note, should cancellations be anything but extremely rare then from a purely game-theoretic perspective it becomes a +EV proposition for any party on the losing side of a bet to argue anything and everything from any angle possible to get that bet cancelled. We have historically seen this at work in highly contested casesⁱ and it's not something that can generally be contemplated. It is better to have a situation where people know they can depend on their winnings being paid out and Gordian knots are being cut by the sword as they appear than to have a situation where everything's suddenly a Gordian knot and nobody can trust any particular outcome of any bet. Furthermore, it is the job of prospective bettors to bring constructive criticism to bets as they get added. Consequently, we will not cancel bets that are oldⁱⁱ, that have garnered interestⁱⁱⁱ or that are contentious^iv.

There are two broad classes of exceptions to this policy. One is that if a bet with material errors^v was recently added, we will cancel it to allow a proper formulation to proceed.

The other is that if the subject of the bet disappears, then we will sometimes cancel and refund the bet. There was exactly one such case to date, where the winner of some pageant could no longer be ascertained because the announced pageant was cancelled. The key consideration that led to the bet being cancelled was the lack of interest for it among users^vi. Had the bet been popular we would have probably resolved it anyway^vii.

So as a rule of thumb : if we judge that the bet being cancelled makes no difference to anyone in the world we might do it. If we judge that the bet being cancelled does make a difference to users, we will not do it. Bet cancellations are a cruft removing tool, to keep BitBet more or less clean of nonsense. They are not a tool of great fairness to bring the light of justice to users who consider themselves slighted by the flow of history. This said, users that have bet in error have at their disposal a failproof method of cancelling their own bets^viii for a modicum of cost.

This theory has some direct application right now, unfortunately. Users pointed out that the Litecoin to fall below 0.006 Bitcoin each before Dec 25th bet suffers from a mismatch between title and spec : it claims to close on Dec 25th, but in fact closes on November the 25th. The bet was introduced 19 hours ago, and it was cancelled 13 hours ago. This would be more or less typical of a cancelled bet due to material error.

This unfortunate discovery also prompted us to do a review of all currently opened bets for similar mistakes, and indeed one was found : Litecoin to rise above 6 USD before Dec 25th. This is a week old bet, rather than something entered the same day ; it has garnered significant interest, rather than consist of just the author's zeroconf^ix and BitBet's seed. It is also highly contentious, with LTC moving significantly in the interval. Consequently, it will not be cancelled. We can only hope that either the Yes condition is satisfied in the next few days, rendering the mismatch moot, or else that all bettors fully understood that in case of disagreement between text and spec, it's the spec that controls.

Obviously allowing this latter bet was a mistake on the part of BitBet. Mistakes do happen as part of business, and while we strive to keep them to a minimum, there has to be a balancing act in place between the effort expended to guard against mistakes and the costs of that effort.^x It is for this reason that BitBet is offered as-is, and users are expected to do their own homework.

That said... good luck to everyone.

1. Such as the BitFury delivery bet, or the BFL delivery bets and so on. [↩]
2. If nobody found anything to object to some bet for a year, that means there's nothing to validly object to that bet. Whether this seems reasonable to you or not is immaterial, time beats opining, every time. [↩]
3. If nobody found anything to object to some bet out of a bunch of people betting, that means there's nothing to validly object to that bet, to a larger degree than simple age. [↩]
4. If a bet has changed significantly from where it started (say it seemed 10:1 originally but now looks more like even odds) cancelling it would be monstrously unfair to those who have had their BTC locked in for the interval, and we'll try anything within reason to resolve rather than cancel. [↩]
5. The most frequent cause of this is a mismatch between the expiration date referenced in the bet text or title and the same referenced in the actual bet specifications. And since we're on the topic of bet specifications : they control the bet. No matter what the title or text may say, if the specs call for X date of closure then X date of closure it is. [↩]
6. If I recall correctly nobody had bet. [↩]
7. For instance, if the bet is for X to win Y and Y gets cancelled, absent specification to the contrary in the text of the bet then obviously X hasn't won Y and so the outcome is No. It is the responsibility of the bet authors to correctly specify what the outcome should be in edge cases if they're not happy with what BitBet defaults on. [↩]
8. How to cancel your BitBet bet. [↩]
9. Zeroconf is what we call BTC sent at the time the bet is made (ie, before it is published on the site), something only the author can do. [↩]
10. Especially important to consider is the point that no finite expenditure can guarantee complete absence of error. Consequently, a BitBet charging 100% of all bets made in administration fees could still not guarantee all bets accepted are correct - presuming anyone were still using it then. [↩]