Topic: Attack insurance (Read 1212 times)

No need to apologize. I have to put you on ignore because of annoying concept of "attack insurance" which could never work.

Well, bitcoin is all about new kinds of commerce the history of the world has never seen before. So, if one thing is certain is that the current state of affairs is not a good indicator for the future. And the reason is that people do all kinds of things for all kinds of reasons, you probably have not ever thought about. The best historic parallel for the events which are unfolding is the Renaissance, when insurance, stock ownership, bonds, etc. were invented. See: http://szabo.best.vwh.net/insurance.html

P.S. I have to put you on ignore, because of annoying advertisement. Sorry.

Sure, Bitcoin in country X might become illegal. But nobody is going to pay people in that country for those bitcoins via "insurance". All they will do is sell the bitcoins on an exchange and get their money back.

I'm using the term insurance and attack very loosely.

Attack could for example also mean, government X bans bitcoin. Now holding Bitcoin in X becomes illegal. Then coinholders in X get compensation. The example is hypothetical and I don't think such a thing would work. But a betting/insurance market might work.

I believe a collapse is extremely unlikely, but the probability is not nearly as small as a collapse in the financial system. Central banks around the world are printing a lot of money, but the probability of the USD and EUR becoming worthless over night is much closer to zero.


Yes, insuring Bitcoin would be extremely costly. I don't know who you refer to by "we". Bitcoin is obviously much more than an "experiment" now. If it the exchange rate would go to zero tomorrow that would have some effect. Which is why  governments, if they could, probably would not shut it down. 10B$ is too much to simply destroy. If Bitcoin grows by another 100x, it will be 1 trillion dollars. And then things might change. 10B$ loss, happens all the time in the stock market for example, 1T$ not.

People have been successfully reversed and respent— many times in the case of unconfirmed transactions, but also in the cases of incidental reorgs and other events. Reports of bitcoin's demise may be exaggerated.

I'm not sure what type of attack you are imagining if you could "smoothly reboot the system" after ..... the bitcoin protocol was built from the ground up to prevent "attacks". Its not been attacked yet, and if it does get successfully attacked, it will just fall apart. Nobody is going to pay out billions of dollars to coin holders.

Bitcoin holders knowingly take on the risk that this experiment may or may not work. We don't need insurance and insurance can never be free.

I described a notion in #bitcoin-wizards a while back that I called "fail inflationary":

<@gmaxwell> So cryptocurrencies which "fail inflationary"... which is a property that things like USD has, if you print really good fake USD .. well everyone takes the cost. Not the person that accepted the fake USD, at least not if its sufficiently good.
<@gmaxwell> It would be totally plausable to make it so that if you get tricked by a recentl valid looking bitcoin fork, that both parties get paid. Thus moving the cost of such an attack to everyone holding bitcoin and not just the guy accepting it.

You could obviously do things like limit the amount of inflation created this way— perhaps in a first come first server exponentially declining way so that the ability to get compensated never runs out but is still finite...

The tricky part is how this could be constructed in a way where miners wouldn't just intentionally create forks in order to collect inflated coins.

It could be either people who are willing to underwrite insurance or it could be part of the network protocol. For example someone who strongly believes that Bitcoin can't be attacked, or attacked in a certain way, could post an insurance offer, proving his trust in the network. That way one would have more visibility into certain risks. Most people who adopt Bitcoin can't really make an informed judgement, so they might want to mitigate that to a third party.

One imagine a scenario where some kind of attack is successful, but build in features make it possible to rather smoothly reboot the system. If Bitcoin would fail today, tomorrow some would reboot a different Cryptocurrency which addresses the failure. On the other hand If there would be a viable Alt-Coin, that might introduce different incentives for attacks. An insurance based scheme could potentially address that issue. Its more of a thought experiment at the moment, but these kinds of structures are important in finance. The 2008 GFC was basically like an attack on the network and the central bank stepped in. In the future that might not be possible. Too big to fail is very similar to mining pool aggregation.

Who would pay?

To expand a little. A precursor to this would be a betting exchange, were BTC owners bet on likelihood of attacks. This could be done as smart contracts over the blockchain. There would then a set of attacks which are formally defined. If one imagines the scenario of a complete attack there are various ways this could play out in terms of how such a shock propagates through the network and following events. In an extreme case a new version could be launched which fixes the loophole (if it is possible). Bankruptcy proceedings are ways to ensure claims of some sort in a more structure manner. For the current international financial system, such proceedings don't exist. The IMF operates on a very ad-hoc basis and even the EU/Euro system hasn't managed to formalize such agreements. Recent history of default insurance (CDS) has been very telling in this regard.

At the very least it would highly interesting to see how individuals or groups (markets) consider probabilities of attacks. See https://en.bitcoin.it/wiki/Weaknesses  In a way holding BTC is already a bet on Bitcoin as a system. But there could be people who would hold Bitcoin if they could hedge certain events. Such insurance is likely to be quite expensive, but it might be a way to propagate audits/trust of the system to other people. In a way that is already largely being done.

Has anyone considered attack insurance as a principle? In a way early adopters and/or developers profit from the adoption, so it would make sense to think about how they could insure the system. What would be needed is a formal method of defining an attack. Once that event is triggered a big loss occurs. The loss could be distributed in smart ways, similar to deposit insurance.