Bitcoin & Gresham's Law - the economic inevitability of Collapse

[torservers]

Bitcoin & Gresham's Law - the economic inevitability of Collapse
Philipp Güring & Ian Grigg
October-December 2011

Abstract.  The Bitcoin economy exhibits remarkable and predictable stability on the supply side based on the power
costs of mining.  However, that stability is challenged if cost-curve assumption is not solely expressed by the fair cost
of power.  As there is at least one major player, the botnets, that can operate at a power-cost-curve of zero, the result
is a breach of Gresham's Law:  stolen electricity will drive out honest mining.  This has unfortunate effects for the
stability of the Bitcoin economy, and the result is inevitable collapse.

http://iang.org/papers/BitcoinBreachesGreshamsLaw.pdf

[torservers]

As a first comment: I don't believe in that scenario too much.

[bb113]

How many Ghash would 100,000 crappy graphics cards get you? I doubt most of the bots have access to decent equipment.

[torservers]

abuse.ch: How Big is Big? Some Botnet Statistics

When I’ve started to sinkhole this botnet I was shocked as I saw that more than 1,2 million (yes, 1’200’000) unique IPs connected to my sinkhole just within 24 hours.

about.com: Botnets

According to a study by McAfee, "at least 12 million computers around the world (are) compromised by botnets."

Even if all coins were mined by botnets, it does not directly lead to collapse. You have to read the paper though to understand why the authors claim this.

Several things happen: (a) incentives for easy money naturally cause an increase in criminal participation at all levels, such as direct theft of bitcoins.  This increase across the board encourages (b) honest users to pack up and leave.  Both of these effects combine to create rising criminality, and (c) at some stage the Feds get involved.  Finally, (d) the system collapses.

[caveden]

People are already starting to use FPGAs, and maybe one day we'll even see ASICs.
Botnets won't have such specialized hardware. Ok, they have free electricity, but they are limited in size.

And by the way, in what concerns the bitcoin network, botnets are not a problem. It only becomes a problem if one single botnet is too large and manages to get >50%.

PS: I confess I did not read the paper so sorry if I'm saying something which is already addressed by it.

[Vandroiy]

It only becomes a problem if one single botnet is too large and manages to get >50%.

And even that only helps if they can sustain it during and following their attack.

In effect, this is just one special case of the question of insurance mining to secure the network after multiple minting cuts. And as it is one of the more likely attack scenarios, enough people know of this issue already.

Also, if I read "the inevitability of collapse", I smell someone lacks attention.

[caveden]

Also, if I read "the inevitability of collapse", I smell someone lacks attention.

hehe, yeah. Researches often desperately need to publish.

[lonelyminer (Peter Šurda)]

Notwithstanding the arguments about profitability of Bitcoin botnet mining (which I won't address, I'll just assume they are correct), I humbly submit that this has nothing to do with Gresham's law and the economic conclusions presented in the paper are incorrect.

The cost of Bitcoin mining has a practically negligible effect on the changes of the supply thereof (I calculated the effect of difficulty fluctuations on the ideal, 1block/10min, function, a couple of months ago and the difference was about 1.9%). In fact, Bitcoin is specifically designed to counter supply shocks. Also of noting is that it is the change in hashing power, not its level, that causes these minuscule effects on the supply. If someone manages to create a botnet with, say 1PH/s, there will be one (maybe two) jumps and afterwards the production rate will be back to normal. It is misleading to say that by faking that you paid for the power, you somehow subvert Bitcoin. You don't. You merely steal the zombie computer owners' money through his electricity bill. Bitcoin, in that case, is just a tool used for this theft, not the consequence thereof. The claim of the paper's writers that "The Bitcoin economy exhibits remarkable and predictable stability on the supply side based on the power costs of mining" (emphasis added) is erroneous.

From the perspective of price, whether the mining is done by botnets or "legitimate" miners is almost entirely irrelevant. The claim that the cost of mining has an effect on the price of Bitcoin is not supported by empirical data, for example, the former seems to lag after the latter, and many more Bitcoins are traded than produced. Nowadays about 7200 BTC are produced daily, while about 85000 traded on Mt. Gox only. In 10 months, only 3600 BTC will be produced daily, while there is no reason why this should affect the amount traded. So the argument presented in the paper reverses cause and effect.

It's just an argument about externalities. Botnet mining creates an externality. But this is between the zombie computer owner and the botnet operator, not between the botnet operator and the rest of the Bitcoin economy. Since the effect on supply is countered, this is, from economic point of view, similar to a bank robbery. It's merely a different distribution of goods, not a change in the supply of goods.

The authors also claim that there could be a motivational link between the botnet operation and direct theft of Bitcoins. I find this unconvincing. The motivation to steal Bitcoins is influenced by the comparison costs vs. profits. If the price of Bitcoin decreases (which is what they claim would be the effect of botnet mining, ed.), that decreases the potential profits as well and therefore decreases the likelihood of further theft. So this reinforcing effect they claim to be underlying the collapse is not there. Also, if someone steals Bitcoins, their model assumes that they will try to sell them right away (which is probably correct), so they stay in the Bitcoin economy anyway. Since exchanges have now started enforcing AML and KYC rules, this creates a risk for the criminals, due to the withdrawals being trackable. And I'm not even talking about the possibility of privately funded Bitcoin "police" which can counter this as well.

There are still risks associated with botnet mining, for example the overpowering (51% attack). But the other, economic, effects, in my opinion, would not be reflected in the price of Bitcoin, the supply thereof, and wouldn't create a meaningful dichotomy between "good" and "bad" Bitcoins with a price fixed at a non-market price (i.e. Gresham's law). What could happen in long term is a decrease in the equilibrium transaction fees. Since the costs of mining would be externalised, they (the equilibrium transaction fees, ed.) would be below the marginal cost. It could also have a minor redistribution effect, assuming legitimate miners hoard more than botnet operators, the result of an increase of botnet mining would be more money on the exchanges and less money with the legitimate miners.

Summary: the authors did not think it fully through and make arbitrary implicit assumptions. Bitcoin breaks implicit assumptions.

EDIT
I spotted one more thing: a false dilemma fallacy, where they argue that if it's not a contract or money, it's nothing. This sort of false dilemma fallacy which I dub "money or nothing" is common to almost all critiques of Bitcoin.

EDIT2
I would also like to submit that the goals of subverting (e.g. 51% attack) Bitcoin and using it for mining profit might not be congruent. For example, if you just want to mine for profit, you'll probably use an existing mining pool, while if you want to subvert it, you must not. The examples that Peter Kleissner showed at the Prague conference and also explained to me in an email showed that those botnets that were backtracked, were indeed backtracked to an existing (legitimate) pool. I'm not going as far as saying that these requirements are mutually exclusive, I'm just pointing out that arbitrarily putting them together as a common risk is unsubstantiated.

[ribuck]

the authors did not think it fully through and make arbitrary implicit assumptions

This.

There are many reasons why people mine, other than the price of electricity. An exchange or a retailer might choose to mine to keep their transactions flowing smoothly; and there will always be altruists and idealists who will mine regardless of the price of electricity.

Also, there are plenty of people other than botnet operators who pay zero marginal price for their electricity.

[matonis]

Notwithstanding the arguments about profitability of Bitcoin botnet mining (which I won't address, I'll just assume they are correct), I humbly submit that this has nothing to do with Gresham's law and the economic conclusions presented in the paper are incorrect.

I would dispute the implicit profitability of bitcoin botnet mining which would not even get us to the faulty economic conclusions.

The thread here has shown that the authors are not hashing experts and now you have shown that nor are they economists.

[cunicula]

Notwithstanding the arguments about profitability of Bitcoin botnet mining (which I won't address, I'll just assume they are correct), I humbly submit that this has nothing to do with Gresham's law and the economic conclusions presented in the paper are incorrect.

Yeah, why do they even bring up Gresham's law? It has no relationship to anything they write about.

[lonelyminer (Peter Šurda)]

I would dispute the implicit profitability of bitcoin botnet mining which would not even get us to the faulty economic conclusions.

You are right, it might be possible do this, but I am not always an Austrian, I'm sometimes a falsificationist, and a lazy one. Since they describe their economic model in about half of the paper, avoiding the question of validity thereof and simply assuming it's true saves a lot of time that I don't have to spend on verifying their calculations :-).

The thread here has shown that the authors are not hashing experts and now you have shown that nor are they economists.

I only saw this thread after I wrote my post. The points about Bitcoin favouring specialised hardware is a very nice empirical point to counter the assumptions of the authors. It might have other consequences (e.g. subversion), but as I said above, subversion and mining "theft" are two different issues.

The whole paper is odd, it's kind of like saying: criminals can enslave people and force them to work in gold mines, therefore, a monetary system based on gold is deficient and needs to be fixed. It's a non-sequitur.

Yeah, why do they even bring up Gresham's law? It has no relationship to anything they write about.

Well, they attempt to argue that botnet mining creates a new type of Bitcoins, so there would be "good" and "bad" Bitcoins. But you're right, their explanation for this dichotomy is insufficient. As soon as the "bad" Bitcoins are sold on the exchange (which is the model they propose), this dichotomy vanishes. Now, it is thinkable that a world dominated by botnet mining might be somewhat different from one dominated by "legitimate" mining. If we disregard the control issues, however, the differences are practically negligible, and more likely to be different that those that the authors present.

[miscreanity]

Notwithstanding the arguments about profitability of Bitcoin botnet mining (which I won't address, I'll just assume they are correct), I humbly submit that this has nothing to do with Gresham's law and the economic conclusions presented in the paper are incorrect.

I would dispute the implicit profitability of bitcoin botnet mining which would not even get us to the faulty economic conclusions.

The thread here has shown that the authors are not hashing experts and now you have shown that nor are they economists.

It seems that the authors might have been aiming for a Minskian collapse.

[99Percent]

Horrible (mis)application of Gresham's Law if at all. These guy should stick with cryptography and stay out of economics. And not even that because a botnet has virtually no chance in competing with the current hashrate, unless (unlikely) such a botnet focuses on targeting gamers. And not even now with specialized hashing hardware like FGPAs.

[proudhon]

Horrible (mis)application of Gresham's Law if at all. These guy should stick with cryptography and stay out of economics. And not even that because a botnet has virtually no chance in competing with the current hashrate, unless (unlikely) such a botnet focuses on targeting gamers. And not even now with specialized hashing hardware like FGPAs.

And they can't merely target gamers.  They've got to target gamers using AMD hardware.  And the gamers need to not notice that when they play Skyrim their framerate has gone in the shitter, or not notice that their GPU fan is always on.  This is stupid.

[BlackPrapor]

And they can't merely target gamers.  They've got to target gamers using AMD hardware.  And the gamers need to not notice that when they play Skyrim their framerate has gone in the shitter, or not notice that their GPU fan is always on.  This is stupid.

Remember these words! botnets have to be very sophisticated to harness any Mhs, and it's still will be at most just another miner's rig power.