Creative angles of attacking proof-of-work blockchains

[Note: the following views were originally included in a new paper but needed to be removed for space and flow considerations]

While most academic literature has thus far narrowly focused under the assumption that proof-of-work miners such as those used in Bitcoin will behave according to a “goodwill” expectation, as explored in this paper, there may be incentives that creative attackers could look to exploit.

Is there another way of framing this issue as it relates to watermarked tokens such as colored coins and metacoins?

Below are comments from several thought-leaders working within the industry.

According to John Light, co-founder of Bitseed:1

When it comes to cryptocurrency, as with any other situation, an attacker has to balance the cost of attacking the network with the benefit of doing so. If an attacker spends the minimum amount required to 51% attack bitcoin, say $500 million, then the attacker needs to either be able to short $500 million or more worth of BTC for the attack to be worth it, or needs to double spend $500 million or more worth of BTC and receive some irreversible benefit and not get caught (or not have consequences for getting caught), all while taking into consideration the loss of future revenues from mining honestly. When you bring meta-coins into the equation, things get even murkier; the cost is less dependent on the price of bitcoin or future mining revenues, and depends more on the asset being attacked, whether it’s a stock sale or company merger that’s being prevented, or USD tokens being double-spent.

There’s no easy answer, but based on the economics of the situation, and depending on the asset in question, it doesn’t seem wise to put more value on chain than the market cap of BTC itself (as a rough benchmark – probably not that exact number, but something close to it).

Not a single study has been publicly published looking at this disproportionalism yet it is regularly touted at conferences and social media as a realistic, secure, legal possibility.

According to Vitalik Buterin, creator of Ethereum:2

There are actually two important points here from an economics perspective. The first is that when you are securing $1 billion on value on a system with a cryptoeconomic security margin that is very small, that opens the door to a number of financial attacks:

  1. Short the underlying asset on another exchange, then break the system
  2. Short or long some asset at ultrahigh leverage, essentially making a coin-flip bet with a huge amount of money that it will go 0.1% in one direction before the other. If the bet pays off, great. If it does not pay off, double spend.
  3. Join in and take up 60%+ of the hashrate without anyone noticing. Then, front-run everyone. Suppose that person A sends an order “I am willing to buy one unit of X for at most $31”, and person B sends an order “I am willing to sell one unit of X for at least $30”. As a front-runner, you would create an order “I am willing to sell one unit of X for at least $30.999” and “I am willing to buy one unit of X for at most $30.001”, get each order matched with the corresponding order, and earn $0.998 risk-free profit. There are also of course more exotic attacks.

In fact, I could see miners even without any attacks taking place front-running as many markets as they can; the ability to do this may well change the equilibrium market price of mining to the point where the system will, quite ironically, be “secure” without needing to pay high transaction fees or have an expensive underlying currency.

The second is that assets on a chain are in “competition” with each other: network security is a public good, and if that public good is paid for by inflation of one currency (which in my opinion, in a single-currency-chain environment, is economically optimal) then the other currencies will gain market share; if the protocol tries to tax all currencies, then someone will create a funky meta-protocol that “evades taxes by definition”: think colored coins where all demurrage is ignored by definition of the colored coin protocol. Hence, we’ll see chains secured by the combination of transaction fee revenue and miner front running.

Unsolved economics question: would it be a good thing or a bad thing if markets could secure themselves against miner frontruns? May be good because it makes exchanges more efficient, or bad because it removes a source of revenue and reduces chain security.

Cryptoeconomics is a nascent academic field studying the confluence of economics, cryptography, game theory and finance.3

Piotr Piasecki, a software developer and independent analyst explained:4

If a malicious miner sees a big buy order coming into the market that would move the price significantly, they can engage in front running – the buy order could be pushed to the back of the queue or even left out until the next block, while the miner buys up all of the current stock and re-lists it at a higher price to turn a profit. Alternatively, when they see there is a high market pressure coming in, especially in systems that are inefficient by design, they can buy the orders up one by one by using their power to include any number of their own transactions into a block for free, and similarly re-list them for people to buy up.

Or in other words, because miners have the ability to order transactions in a block this creates an opportunity to front run. If publicly traded equities are tracked as a type of colored coin on a public blockchain, miners could order transaction in such a way as to put certain on-chain transactions, or trades in this case, to execute before others.

Robert Sams, co-founder of Clearmatics, previously looked at the bearer versus registered asset challenge:5

One of the arguments against the double-spend and 51% attacks is that it needs to incorporate the effect a successful attack would have on the exchange rate. As coloured coins represent claims to assets whose value will often have no connection to the exchange rate, it potentially strengthens the attack vector of focusing a double spend on some large-value colour. But then, I’ve always thought the whole double-spend thing could be reduced significantly if both legs of the exchange were represented on a single tx (buyer’s bitcoin and seller’s coloured coin).

The other issue concerns what colour really represents. The idea is that colour acts like a bearer asset, whoever possesses it owns it, just like bitcoin. But this raises the whole blacklisted coin question that you refer to in the paper. Is the issuer of colour (say, a company floating its equity on the blockchain) going to pay dividends to the holder of a coloured coin widely believed to have been acquired through a double-spend? With services like Coin Validation, you ruin fungibility of coins that way, so all coins need to be treated the same (easy to accomplish if, say, the zerocoin protocol were incorporated). But colour? The expectations are different here, I believe.

On a practical level, I just don’t see how pseudo-anonymous colour would ever represent anything more than fringe assets. A registry of real identities mapping to the public keys would need to be kept by someone. This is certainly the case if you ever wanted these assets to be recognised by current law.

But in a purely binary world where this is not the case, I would expect that colour issuers would “de-colour” coins it believed were acquired through double-spend, or maybe a single bitcoin-vs-colour tx would make that whole attack vector irrelevant anyway. In which case, we’re back to the question of what happens when the colour value of the blockchain greatly exceeds that of the bitcoin monetary base? Who knows, really depends on the details of the colour infrastructure. Could someone sell short the crypto equity market and launch a 51% attack? I guess, but then the attacker is left with a bunch of bitcoin whose value is…

The more interesting question for me is this: what happens to colour “ownership” when the network comes under 51% control? Without a registry mapping real identities to public keys, a pseudo-anonymous network of coloured assets on a network controlled by one guy is just junk, no longer represents anything (unless the 51% hasher is benevolent of course). Nobody can make a claim on the colour issuer’s assets. So perhaps this is the real attack vector: a bunch of issuers get together (say, they’re issuers of coloured coin bonds) to launch a 51% attack to extinguish their debts. If the value of that colour is much greater than cost of hashing 51% of the network, that attack vector seems to work.

On this point, Jonathan Levin, co-founder of Chainalysis previously explained that:6

We don’t know how much proof of work is enough for the existing system and building financially valuable layers on top does not contribute any economic incentives to secure the network further. These incentives are fixed in terms of Bitcoin – which may lead to an interesting result where people who are dependent on coloured coin implementations hoard bitcoins to attempt to and increase the price of Bitcoin and thus provide incentives to miners.

It should also be noted that the engineers and those promoting extensibility such as colored coins do not see the technology as being limited in this way. If all colored coins can represent is ‘fringe assets’ then the level of interest in them would be minimal.

Time will tell whether this is the case. Yet if Bob could decolor assets, in this scenario, an issuer of a colored coin has (inadvertently) granted itself the ability to delegitimize the bearer assets as easily as it created them. And arguably, decoloring does not offer Bob any added insurance that the coin has been fully redeemed, it is just an extra transaction at the end of the round trip to the issuer.

  1. Personal correspondence, August 10, 2015. Bitseed is a startup that builds plug-and-play full nodes for the Bitcoin network. []
  2. Personal correspondence, August 13, 2015. []
  3. See What is cryptoeconomics? and Formalizing Cryptoeconomics by Vlad Zamfir []
  4. Mining versus Consensus algorithms in Crypto 2.0 systems by Piotr Piasecki []
  5. As quoted in: Will colored coin extensibility throw a wrench into the automated information security costs of Bitcoin? by Tim Swanson; reused with permission. []
  6. This example originally comes from Will colored coin extensibility throw a wrench into the automated information security costs of Bitcoin? by Tim Swanson; reused with permission. []
Send to Kindle

Leave a Reply

Your email address will not be published. Required fields are marked *