Topic: Difficulty Increase Insurance (Read 3895 times)

The lockup idea is actually fairly trivially from a technical perspective.  The hard part is the arbitration service actually being able to be trusted and unlocking things.  They need a way to be able to judge.  This likely isn't straightforward, and it's hard to know if that entity can be trusted.  That entity may conspire with the parties to hijack the coins, etc...  Judging these things is also going to be expensive.

That being said, I certainly wouldn't be opposed to doing this.  In the end, it's something I'd like to get more into, and it's a bit more general purpose than I envisioned (I wanted to stick to arbitrating things that are easy to judge, such as difficulty), but it's not impossible.

You may want to see my external state posts for more information on this.  Escrow is one key thing that it can be used for.  I've already gotten the lock-up functionality to work on testnet, so it's just a matter of building it.

Forget "difficulty insurance" and pursue the lockup idea. You're on to something important. Come up with a way to do distributed escrow as a two-phase cryptographic protocol. Something that works like this:

- A sends N locked bitcoins to B to buy something.  A cannot spend those Bitcoins again, but B can't spend them yet.
- A gets whatever B was supposed to send them.
- A unlocks the N locked Bitcoins. B can now spend them.

That's enough for little transactions, up to maybe $10 or so. This is better than "escrow services"; there's no escrow service which can run off with the money. (It happens. Big problem on eBay.) Both sides can lose, with un-spendable Bitcoins in limbo. Since neither side has the money, both sides will probably  try to come to some agreement. Make that work and hook it into some popular shopping cart program.

There's a fancier version, with an arbitrator.

- A sends N locked Bitcoins to B, with an arbitration service listed in the transaction.
- A doesn't get whatever B was supposed to send them.
- A sends a token to the arbitration service requesting arbitration.
- Cases:
   - B sends their token, accepting arbitration.
     - A wins. The arbitration service unlocks the Bitcoins in favor of A.
     - B wins. The arbitration service unlocks the Bitcoins in favor of B.
   - B does nothing, and after some period of time, loses by default.
 
- or
- A fails to unlock the locked Bitcoins they owe to B
- B sends a token to the arbitration service requesting arbitration.
  - cases as above.

A solution has to have the following properties:
- The transaction is anonymous unless submitted to arbitration.
- The unlocking operation is a 2 out of 3 cryptographic protocol.
   A and B can unlock, or A and the arbitrator can unlock, or
   B and the arbitrator can unlock.
- Operations involving an arbitrator have a delay (days to weeks)

That may be overkill, but it would be useful to have it available for larger transactions.  (Like "pre-orders" from Butterfly Labs, perhaps.)

This would make Bitcoin ripoffs much harder, and make the sale of goods using Bitcoins much safer. 

Intrade is long gone, and would be something Bitcoin could easily be used to replace it.  RIP Intrade.

At today time, the insurance of every person is must because this will give the feeling of safe for you and your family. So Insurance is the big point to be noted by the every person, and he must insured himself for him and his family. Insurance risks have great importance.

Intrade anyone?

LOL @ you.

Reduction of variance is worth taking a -EV bet sometimes.  That's why people buy insurance.  The goal is not to get a payout.  The goal is to hedge your investment in case something unforeseen happens.  The market prices things to a fair level based on current knowledge, and profit can be made by having better knowledge than the rest.

The economy is not zero-sum.  Almost always you have win-win scenarios, even when there is profit.

Scam.

Nobody starts a business to lose money, unless they are a complete moron.  Anyone thinking they will actually get a significant payout is wrong.

The market can define those probabilities just like the icbit contract.  It's just a different payout structure with the same concept.  To figure out the true value of the icbit contract, you need to factor in those probabilities just the same.  A single binary contract actually requires less calculations than a variable contract and is something that can more easily be guessed.

Since the payout is variable based on how far it is off, you could think it is more likely to be below the trading value, but still have a +EV trade by buying the contract rather than selling, if the payout is high enough with the upside, and the payout on the more likely end is lower.

To define those levels you gave in the example you would need to know the probabilities to start out with. But as you say no one knows those probabilities. And the different contracts would become (un)interesting as new information becomes available and so the community estimate goes up or down. The icbit Dec contract started out at 300M and it last trade was at 484M. This free trading of these future contracts allows the community to bet on the future difficulty without anyone having to guess the probability upfront. It allows the price to reflect news as they become available from the ASIC vendors. And if enough people participate, the efficient market hypothesis predicts that the market value should be a good estimate for the December price. In that case it would not only help miners to hedge but also give them this good estimate as free information. Right now the estimate is probably not very accurate because the volume is low. On the upside, this gives a good opportunity to anyone interested to bet on the difficulty and who think the difficulty will be lower or higher than currently traded.

You don't bet if it's going up or down.  No one is going to bet that it's going down, of course.  The idea is no one really knows what the difficulty will be in December.  Let's use a block 10 increases in the future from now as an example.  If I'm a miner, my profitability will vary tremendously based on what that difficulty is (along with every other along the way).  I cannot make an accurate assessment of whether my investment is profitable.  However, you can have difficulty futures, such that if the difficulty is above X it pays one party, and if it's below it pays the other party.  Both parties put in money up front, and get paid out in December whatever the results are.  It does not matter what X is, there is a price that someone could theoretically pay for both sides.

So you have various levels of X, and they require different contribution shares from each party.  The winner of each contract wins 1 BTC.  For example, you may have:

Difficulty	Price to win if over	Price to win if under
500,000,000	.01BTC	.99BTC
400,000,000	.25BTC	.75BTC
300,000,000	.60BTC	.40BTC
200,000,000	.90BTC	.10BTC
100,000,000	.99BTC	.01BTC

The prices will roughly track the probability of the difficulty being at that level.  For example, with these made up numbers, it would infer there is a 1% chance of being > 500M, a 25% chance of being above 400M, a 60% chance of being above 300M, a 90% chance of being above 200M, and 99% chance of being above 100M.  These kinds of prices will allow him calculate how much BTC he should mine in those 2016 blocks based on current expectations.  In addition, he can also buy some of these contracts, such as above 400M and above 500M, such that it will compensate him if difficulty rises too much, he can at least recover some of his investment.  This has a cost, but if it is lower, he will make up for it in extra bitcoins mined.

Why would someone take the opposite side?  Profit.  Say I think the difficulty will only be above 400M 10% of the time.  So I can bet 75 BTC against it going above 400M.  10% of the time, I lose my 75BTC.  90% of the time, I win 25 BTC.  This means I profit 60BTC.  I still win even if the difficulty rises to 399M.

Taking it an additional step forward, someone could offer a more complicated contract such that the miner only needs to pay a simple price, and the result of the payout would be proportional to the difficulty.  Say a miner only wanted to hedge above 400M in this case.  He could be paid Expectations of Mining at 400M - what he would make mining at the actual difficulty.  So if it was 405M, he would get a small payout, and if it was 600M, he would get a larger payout.  He would pay a small amount for this service, and would use it to make his expectations much more known in the present and make his decision making on expanding rigs, etc... easier to plan.

Well when the difficulty chart is only heading one direction, who would take the other side of the deal.  If you find someone, I will buy immediately.

Insurance was used as a simplification for those who don't compare about implementation, but perhaps it was the wrong choice.

Obviously there are legal issues that need to be brought up.  A P2P nature of such a system might negate some of those challenges.  For example, while sports books might be illegal, there are often no legal challenges with two friends who put up a social wager with each other.  Sports also poses challenges, but things like prediction markets (which this could very well be a component) are certainly less grey.

Decentralizing and splitting it out might get around some of the challenges (for example, someone who merely reports the difficulty wouldn't be likely to be troubled with it, but reporting difficulty in specific formats (say hashes before hand, and nonces afterward) might).  Having peers form their own bets and use external values might get around some of it.  Having a service that helps peers find each other also may work, or just having it as a p2p network as well could work.  Obviously due diligence is needed with legal experts before actually undertaking anything like this seriously, but first understanding if there is even interest is a first step.  Bitcoin difficulty seemed like a good first target, simply because it is interesting to many people who currently hold coins and that it could actually be beneficial for the Bitcoin community to have such a market exist.

Yup, okay - I can definitely see a market for hedging.   Stop using the Insurance word tho...

My main concern [now] would be that the US CFTC would be all over it, if only because it would regain a bit of credibility for them after what'isname said that BitCoin itself was a derivative, and then had to retract and came out of it looking like a fool.

This is just a simple misunderstanding due to my simplification of how it would work.  This is merely a hedge.  This exists today without any need for scamming through securities.  You have someone put up capital to cover a fixed payout and lock it in the blockchain.  You know the funds are there, and you know you can access it once the difficulty is adjusted.

The price would be determined by those willing to bet on the difficulties.  If the price was priced too low, non-mining speculators could invest in it, driving the price up, and if too high, speculators would flock to the other side.  Those who are best able to predict the price profit, those who are worse lose, and the miners who purchase the insurance get a fair price and can accurately calculate the best guess at difficulty (and can hedge against any unexpected rises).

Insurance works because risks spread across a large number of customers can be assessed and forecast to within some defined error.

So I would have no interest in 'difficulty insurance' because I would be sure that any such organisation would be unable to survive - the people behind it would either be fools, or scammers.

I did some basic calculations on a possibly better way to hedge against difficulty increases.  You have a mining rig that can mine at 1GHash/sec.  At the current difficulty, that's worth .01BTC per day.  Over the entire period, it would be worth .14BTC.  If difficulty increased to infinity, it would be worth 0BTC for the next period.  If it doubled, you would be getting .07BTC for the next period.  So you want the contract for the next period to pay .07 to make up for the loss of mining ability if it doubled.  If it quadrupled, you'd get .035 BTC, so you'd want it to pay out .105BTC.  This comes out to being a x^-1 function with a max payout of .14BTC.  Say you had each share pay out .01BTC, someone with a mining rig might buy 14 shares of it.  Chances are, they get a fraction of the payout (it only goes up 20%).  The further futures out are worth more.  But they could buy those (likely for a higher price), and you could even have a structured amount that pays out for every period from now until a year from now.  You could guarantee an exact profit of bitcoins for your mining rig (you know what it costs to buy the futures, then subtract it out).  A very easy way to find profitability and let other people speculate on getting the price right.  Miners no longer need to be speculators on difficulty, and can know exactly what they are getting into.  The only variable left is the BTC Price and cost of electricity.  But they can know for $X of mining equipment, they will get Y BTC after a year.

Another option is icbit.se. They have have a September difficulty future (Diff-9.13 trading at ~81M) and a December one (Diff-12.13 trading at ~350M). Funding is guaranteed by limited leverage and in worse case margin call. (I'm not affiliated with icbit. I'm only using the site as customer and would like more volume on the site).

You could structure a series of all or nothing bets to get a pretty smooth transaction at different difficulties.  The ideal situation would be where a miner could get a flat payout no matter what difficulty happened.  I haven't thought enough about how that might need to be structured, but it seems reasonable it could be done.

As the bitcoin mining is getting more profesional, it would be great to have a liquid market of btc derivates for risk hedging against difficulty and btc prize, it would bring less risk to the investors, i think a linear calulation would be better than all or nothing.

I agree, and I have a solution for that issue.  It basically requires the capital to be locked up.  And I can do with a transaction that gets put on the block chain that can only be unlocked by the "winning" party once the difficulty adjusts.  It can proportionally pay out with a little bit of logic different (or even some other calculation other than linear), but right now I'm focusing on all-or-nothing.  I'm trying to figure out what kind of things would be useful for the Bitcoin community to hedge against (or offer insurance against), or if it should just be entertainment based (Satoshi Dice seems to be handling that well).