Bitcoin Forum

Bitcoin & Gresham's Law - the economic inevitability of Collapse
by Philipp Güring & Ian Grigg, Dec 2011

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

Ian's Blog: https://financialcryptography.com/

the stuff he "predicts" is already being done and still not killing us.
stay positive.

FUD.

These predictions become more likely to occur as the block reward decreases.

Proof-of-work is fatally flawed and in the future alternative cryptocurrency will be based on a more sensible architecture... proof-of-stake.
 
Coding robots come out with a release for me.

That said, the writers of this article don't know shit about economics and abuse terminology. They would be better off sticking to cryptography.

Quoting myself:

If bitcoin achieves critical mass, having criminal botnets bigger than the rest of the network seems extremely unlikely. There's the problem of having proof of work that's extremely more efficiently done by very specific hardware than with common hardware, so a very big party can still create a network of very specific hardware and outmine the rest of the network into submission (making most of them give up mining and achieving 50% then). This also becomes more unlikely as specific hardware is commercialised and acquired by a number of independent miners and for researchers it's much easier and safer to sell it openly than to a cartel.

FUD

The arguments they put forward are very similar to the Ben Laurie blatherings on bitcoin. Zooko Wilcox-O'Hearn had a Google+ post reviewing these 'blacklist' and 'whitelist' architecture issues for bitcoin nodes:

"What Would You Do With >50% of the World's Bitcoin Mining Power?"
https://plus.google.com/u/0/108313527900507320366/posts/6NCqDrZUYiF

My point is that these are fundamentally economics issues, not cryptography issues. The people having these discussions have no idea what they are talking about.

Setting up and maintaining  a botnet takes time and effort. It's not "free."

Not to mention people would notice the sound of their gpus mining at full speed so those are pretty much out of the question.

Network effects are also cryptography issues, or more generally computer security issues.

But actually I'm not going to side with them just because I'm CompSci/EE, the points they bring up are highly speculative and unscientific. In this regard they are talking in terms that belong in economics and sociology, which are not actual sciences. However, that doesn't mean only economists and sociologists can speculate about human behaviour.

Botnets will never work for minigng..they are to valuabel and owners can make more (illegaly) using them in other areas..using them for minign would go the opposite..they would easier be to spot and to spot..no botnet owner right in his head would  ever use it for mining..at 100$ per btc they maybe try lol Also specialised hardwar..etc..cpu minign jsut doesnt woirk out..why can`t these people read a bit ebfore beginnign to FUD

I won't defend economics as an actual science. It isn't. Economics lays out assumptions about human behavior and then derives how agents following those assumptions will behave. The authors are not doing this.

There are two ways of doing economics. Empirically, like physics, or aprioristically, like math. I agree that it is not scientific to do economics empirically, because you simply cannot isolate all the buzillions variables involved, nor repeat your experiments enough times to get good samples. They actually can't even experiment.
But economics can be done like math, using logically sound constructions on top of axioms. This works much better and it is scientific.

This is another "Bitcoin will fail because it's successful" type argument (it's usually about hoarding -- bitcoin will fail because it's so successful that people keep them).

The author's argument is wrong because Bitcoin the network doesn't care in the slightest how many miners there are nor how much they are paying for their electricity.  It is absolutely excellent for Bitcoin if a load of botnets start mining -- the blockchain is more secure.  It's not great for the people who are having their resources abused to do so; but that is their problem.


He should give it a go.  See how quickly his ever cheaper coins get snapped up.  The price is set by the market, not by the miners.  That becomes more and more true every day while we are in our strong inflationary period.

The only potential risk is that a large botnet gets switched off suddenly and the rest of the network can't adjust to it fast enough.  Block production would suddenly slow massively.  However, I think I saw Gavin writing about having potential work arounds for that, so let's not worry too much.

+1, I had the exact same thought. I read through almost to the end of the article, and FPGAs/ASICs are not mentioned - a gaping hole in such an article.

BS

Gosh i so hate these self proclaimed experts who are totally clueless....

Sure, economics is done like math. This is how it is done in academia, and doing this would be a great improvement over what the authors did in this article.

However, deriving a result from a set of assumptions should not be confused with science. Science requires the formulation of testable hypotheses. In economics, particularly macroeconomics, it is typically impossible to credibly test theoretical assumptions. That is why I say it is not an actual science.

Botnets are only a temporary problem for Bitcoin.

The authors have missed several important points:

1 Some honest miners also have a zero cost curve for electricity. Market forces will ensure that their market share will rise. Every electric heater in the world has the potential to be converted into a mining rig.  This market can potentially absorb a huge amount of Ghash/s.
 
2. Most ordinary people (=botnet victims) are ditching their PCs for laptops and laptops for tablets, which have humble GPU power compared to even an amateur miner.

3. Successful botnets operate stealthily.  Botnets that cause too much disruption for their victims are the ones that get shut down.  It is difficult to run a stealthy mining botnet for an extended period of time, because at the very least the fan noise is going to bother the victims.  

4. Botnets have a limited lifetime. Hackers must constantly work hard at infecting new machines in order to make money. A miner, on the other hand, has a passive source of income once he has invested in the rig.

5. CPU => GPU => FPGA => ASIC => highly specialized non-silicon IC

+1

But as I said before, not only economists and sociologists (with their respective lingos) can effectively speculate about human behaviour.


I'd say this is a bug rather than a feature. Traditional botnets never have significant power (in any measure) in comparison with the that of the complete network, they are only of any use because they can profitably attack reduced sets of individuals. They can trounce a carefully selected individual or company. If a 50%+ attack on the whole was necessary they would be completely useless.

However, amassing very specific tools that are out of reach for most people is something a cartel can try and maybe succeed in doing.

For instance, if the higher end of these (http://www.butterflylabs.com/products/) (  discussion (https://bitcointalk.org/index.php?topic=65752.0;all) ) are available, then it would take around $4M to control mining (provided they could ramp up production to that scale).

That's how mathematics itself is done... you don't consider Mathematics to be a science?

EDIT: Correcting myself, mathematics (and sound economics) derive results from a set of axioms, not assumptions, it is not the same thing.

+1

I believe one of the reasons that Bitcoin uses the type of algorithm that can be solved faster with specialized hardware is the issue of botnets. Right now it's already difficult for botnets to do anything with Bitcoin because they don't have specialized hardware. It's still in the realm of possibilities but saying that this will get worse in the future is totally ignorant. As Bitcoin gets bigger there are two things that ensure that botnets will be less and less of an issue.

As Bitcoin grows the value of bitcoins go up, which will cause an increase in the computing speed of the mining network. Bitcoin is fairly safe now but imagine a price of $100+, we would probably have a mining network with over 100 thash. At the same time there are other developments going on such as the mass production of FPGA's and eventually ASIC. These developments combined will make sure that worrying too much about botnets is unfounded.

Right now you could do something with a botnet of millions of regular computers but not in the future. In 5 years you'd probably need billions of computers and nobody has that many in their control.

There is another issue which is actually more significant and that is the centralization of mining pools. It becomes much easier to target Bitcoin if you first ddos all the big pools out of the way and then start hashing. I'm not too worried about this either, we have seen in short time that Deepbit has lost a lot of hashing power and new innovations such as p2pool have been very successful.

I'd say that the future of Bitcoin mining is looking bright.

Science in the classical definition meant "field of knowledge" - in that sense it is. But in the modern, more strict definition of science it doesn't qualify ("natural science" - as described by Sir Francis Bacon, XVII century). It's still the philosophical base for most if not all relevant science.

I brought this semantic distinction in the first place because non-sciences like economics don't prove anything at all in the real world and only give degrees of certainty in repeatable scenarios. In this case they are basically applied statistics.

The author(s) here are making highly speculative assertions about unprecedented situations without providing much base for them and taking a lot of dubious preconditions for granted. This level of discourse is very common in economics and sociology, even at academic levels. Just look at keynesianism.

Topic title: Another respected cryptographer predicts collapse in bitcoin mining

What does his point have to do with cryptography?

You're wrong there, sound economics can prove things, just like math. And sound economics shouldn't even use statistics as a basis for knowledge building, that's why I made the distinction about "empirical economics" (which I consider flawed, for the same reasons you point) and "aprioristic economics".


Keynesianism is not sound economics, it's done empirically.

I couldn't find any mention of transaction fees in the paper. He only mentions the subsidy. Subsidy will decrease to almost nothing, but block rewards will not.

If he's right, we will see Litecoin collapse far sooner, giving the community enough time to develop a replacement like EnCoin. Don't worry.

Mathematics "prove things" within their own models. This is why they are not proof of anything in the real world. Aprioristic economics also prove things within their own models. They can have a lot of merit and be arguably very solid, not discussing that. However there is always a "leap of faith" in taking proof from a model and saying that model is also sound and applicable to reality. This "leap of faith" exists even in extremely developed fields like physics. For instance, between Newton's model and reality there was a leap of faith, and for centuries it was considered absolute truth. Then Einstein proved it wrong and came up with a more precise model. Considering the Einstenian model to be absolute truth also requires a "leap of faith." If that elusive neutrino from the LHC was truly faster than light, then Einstenian model would actually be "wrong" (it would be an approximation just like the Newtonian model, but this time we don't have a "better" one that looks like the real deal - the faster-than-light neutrino observation is still being disputed though, so actually Einstein could theoretically be "right").

Anyway I think we are taking this off-topic and entering philosophical territory.

Let's just agree that this "paper" is highly speculative and not to the scientific standards one would expect of a Computer Science paper. I was arguing that in economics you read this level of "thing" even in the most reputable publications. Sad but true.

There are such things as "proofs" in mathematics; given certain axioms.  As muyuu says, they are disconnected from the physical universe and are valid only within the abstract universe of mathematics.

The is no such thing as a "proof" in science.  All there is is a weight of evidence.  We have models that fit all currently observed data -- that's as good as it gets for there is no way to know that tomorrow an apple won't fall upwards from a tree.

Since economics is a study of things that happen in the physical universe; it can offer proofs that are no stronger than those of physics (and given the enormous number of confounding variables, probably not even that level) -- i.e. no proofs at all.

Economics is all empirical.  If any proof you offer isn't empirical then it isn't economics, it's mathematics.

No, I don't consider mathematics a science at all.

A bit like astronomy then :) 

That is a bit of a stretch. Keynesian makes up a large minority viewpoint in academic macroeconomics. It is not dominant. It is only among government policymakers that Keynesian is the dominant viewpoint.

[Don't get me started on Austrian economics because that is complete wackjob rubbish that asshats believe in. I'm not responding to asshats here. It is not an enjoyable use of my time.]

It is not that Keynesian is necessarily wrong. In fact, I think it is correct, but exceptionally difficult to prove or disprove. From the empirical side, it is impossible to perform random experiments with national economies. From the axiomatic side, it is very difficult to derive a Keynesianism model based on seemingly sensible assumptions. The typical assumptions are that people are rational, have rational expectations about the future, and that prices adjust freely.

However, this is not surprising. The idea that people behave irrationally is at the core of Keynes' thought... "animal spirits" and such. If you allow people to deviate from rationality, then the axiomatic approach becomes pretty useless. People can deviate from rationality in many ways. Oftentimes there are multiple types of deviations which could generate similar macroeconomic data. That is multiple potential axioms could be broadly consistent with the same data. In this case, who is to say what the right way to set up axioms is? It becomes faith-based. This is like math, where people can disagree about whether parallel lines ever meet. This kind of theorizing is pretty useless in my view.


By the way, Francis Bacon and the scientific method is what I have in mind when I think of science.

The economics that I do is more in the way of applied statistics as the person above mentioned. It is not very scientific though it strives to be. The point is to tell a convincing story, where your audience believes you have thoroughly analyzed a situation and considered many possible explanations for it. There is no sense of certainty like you get in the natural sciences in which I did research prior to becoming an economist.

Actually, it would be simple to turn economics into a science. All you have to do is add, at the end of any theorem, the words that Richard Feynman said should be etched in gold on the doors of every lab: "If it disagrees with experience, it is wrong".

Competing botnets could take over the Bitcoin network, but that will strengthen it, not collapse it. Botpools: it's the future  ;)

Now that is complete BS. Link to a recently published example in a leading journal.

But one more thing. It is true that the other sciences also just generate a weight of evidence rather than proof. It is a matter of degree, not black or white. It is just that the level of certainty attached to a laboratory result is much, much greater than in economics.

Astrophysics and nuclear physics, both of which rely heavily on statistical data.

I didn't say "journal", I said reputable publications. People actually respect "The Economist", "Financial Times", and the economy section in any major generic newspaper. And they don't come short on BS.

But these are like Popular Science. I'm sure you could find BS in Popular Science too. Were you referring to popular science when you said "Computer Science paper?" If not, then it is not really a valid analogy, is it?

If the majority of professional CompScis read, quoted and applied Popular Science at work, then I'd have to say Popular Science is "considered reputable." There's simply a different standard between these two fields. The evidence is out there in the streets ;)

http://i.minus.com/dbxZNwKU8m0RWq.png

The OP is wrong, I can prove it. No matter the price is high or low, the effective miner is profitable.

What are you talking about, specifically? You seem to be obfuscating.

and the cryptographer's reputation is ruined by this prediction.

It is evidence-based and therefore "not sound". Intelligent Design, huh?

I was just addressing your analogy here, sorry if I didn't explain myself clearly  ???  I said "Computer Science paper" but I didn't say "Economics paper" nor was I comparing there. I just pointed out that "reputable publications" about economy, that have a lot more readership and repercussion that this "paper" can dream to have, very often have this level of quality and even worse.

The fact that the blockchain chugs along just fine and that bitcoin continues to be useful right now has a lot more weight and "proves" much more than any "analysis" of this sort. Most people with sense will agree.

In any case I don't really think this discussion about fields is going to be fruitful. Bitcoin is susceptible to a number of fields and any decent general analysis will have to be multidisciplinary.

If you want we can probably take this conversation to the off-topic section. I tend to go on tangential points a bit too much. Sorry about that.

No that's okay. I actually agree with what you say here, so I think we can amicably conclude the exchange.

Joseph Affonso Xaxo has reminded me on Google+ that the authors confuse Bitcoin with Tenebrix, a CPU-friendly, GPU-hostile cryptocurrency.  https://bitcointalk.org/index.php?topic=45667.0

Also see, Would any of the current botnets be able to launch and sustain a 51% attack?  (http://bitcoin.stackexchange.com/questions/1975/would-any-of-the-current-botnets-be-able-to-launch-and-sustain-a-51-attack)

As someone that has dealt with botnets a lot in the last year on BTC Guild, I don't see them pushing out honest miners.  Not unless botnets can find a way to mask themselves from the pool ops.  Solo mining isn't really an option for botnets as far as I can see due to the inability to mask the synchronization of a more than 1 GB blockchain on every bot.

Right now spotting a botnet fairly easy to do manually, a little trickier to automate.  But there is not a single botnet out there that I've seen that has an efficiency ratio close to a legit miner, paired with hitting the server with far more longpoll connections than any regular worker.

The only way I see botnets replacing honest miners is if a way to mine coins solo is designed that doesn't require a bitcoin client running on the system.  This would let a botnet mine coins for a specific wallet address across all their solo miners. 

This makes no sense.

What is the problem if botnets do the mining? Honest miners are those that care about mining legitimate blocks and has nothing to do with where they get their power from and at what price (even if they steal energy for free to mine). Also the final arguments can be said about any currency. If I am a criminal I could steal peoples dollars and sell them on the market to create volatility. I don't see the point here...

This is the part at which I realized the writer is a cryptographer, not an economist, and concluded that this article is irrelevant. Even if all of Bitcoin is mined by competing botnets it will still survive just fine. Don't forget that Bitcoin mining rewards will eventually decrease to zero, and mining will mainly be supported by transaction fees. If all the mining is done by free botnets, it will just mean that transaction fees will be able to remain low.

This guy is wrong on everything.

Here is another one...Incentive.
You control a huge botnet and mine a million Bitcoins for free.

What is the most likely thing to do?

A. Crash the Bitcoin market..Because?

Or B.

B. Hold them and become a Billionaire?

Ofcourse you might get hired to do this...
But if thats the case, make sure they pay you well.

He also assumes that the Botnet will get enough Bitcoins each day to be able to drive the price down.
If the Botnet gets 1000 coins, that might not have any effect at all unless he hold them for 100 days.
But if he holds them for 100 days, the price can rise while he hold them and as the price gets lower it becomes easier to buy them.

There are more.
There is this javascript mining program which can mine on peoples computers while they are looking at a webpage.
What if a few large newspapers or even games started to use this as a kind of payment for reading the news for free...
 

Connecting botnet to mining causes botnet to be exposed to infected users and therefore liquidated. Not profitable activity.

Keynesian propositions are not economics but political dogma. And what is a Keynesian doing around here? Chartalist (http://en.wikipedia.org/wiki/Chartalism)!

A common logical fallacy (http://en.wikipedia.org/wiki/Logical_fallacy) is the argumentum ad hominem. As Ludwig von Mises wrote, "Only ideas can overcome ideas." Have you even read the magnum opus Human Action (http://mises.org/Literature/Author/280/Ludwig-von-Mises) which is available for free? I can understand why it may not be enjoyable for you, if you can even understand it. Most all of the issues raised in this entire thread are thoroughly discussed in the first 70 pages.

The white paper is flawed for several reasons including asymmetrical knowledge, specialty capital goods, opportunity cost, arbitrage and tax considerations.

What kind of hash rate can a botnet pull? I guess the threat is if botnets are king they can perform double spends but one might think it be more profitable to just run the network legitimately and not devalue bitcoin to zero.

OPs point would be valid if Botnets were a lot easier to get hold of and that the reward were greater (preventing opportunity cost). As it is, the economic theory is there but in practice botnets just aren't strong enough. The "parent's/office's electricity" problem is of far greater magnitude, as is the "losing money but cool" problem. None of these drive bitcoin below cost of electricity, and the latter 2 help to secure bitcoin's value even further by removing the short term profit motive and making sure the blockchain remains secure, even when the Profit:Loss ratio is temporarily skewed against them.

As long as botnets are running according to Bitcoin protocol to secure transactions, the Bitcoin are not threaten in any way, it only benefits.

And most importantly botnets are ineffective for long term mining operation. The software runs on CPU, the CPU are in orders of magnitude less efficient than high-end GPU, the computer are sluggish as hell and as a result the opreating system are reinstalled and bot is lost.

I have a impression that author of that paper have got it all wrong.

I don't follow the "respected cryptographer": botnets may be one day similar to FPGAs or ASICS, if they can improve the efficiency of mining. In what way do they bring an incentive to crash the market ?
If botnets provided a low cost resource (certainly not zero cost because development, distribution ,maintenance may be cheap but not free) so easily why assume that honest miners would not be majority in using botnets?

Any "expert" who cites an economist from the 16th century (whose reasoning was based on gold money) better have a sound logic, otherwise he makes a fool of himself.

The interesting thing about this unedited and not well thought out paper is why it was written.  It's not clear to me.
Perhaps they are fishing for a buyer.

if there's one thing we've learned about these doom and gloom papers is that talk is cheap.

We have a lot of "experts"willing to throw around theories but not willing to put up the money, time, our effort to actually demonstrate their hypotheses.

  If there was anything to what they have to say then they'd just take  those millions and run

What the hell are you talking about? What would you have the experts do to demonstrate their hypothesis?

Private pool?

Still I doubt even then botnets can replace honest miners.  Botnet nodes are just so weak.  CPU can't compete with GPU farms and soon even GPU won't be able to compete with FPGAs.

ok, let me spell it out for you.

if i were smart enough to develop a way exploit Bitcoin's code, i would either sell it for millions to someone who had the resources to implement the exploit or i would do it myself.  anyone can talk but the only way to prove it would be to do it.

You could also try to patch the exploit, would attempting to do that indicate stupidity?

Moreover, how is this relevant to the article? They are not even talking about an exploit.
 
Moreover, why would an exploit be worth millions? How could you extract that much of bitcoin's value before the bitcoin price collapsed. The most straightforward way of extracting value would be shorting on bitcoinica. It would be difficult to recover millions doing this. Do you picture Bitcoinica/Mt.Gox releasing money to people who place million dollar short bets right before an attack? Even if they were willing, do they even have that much liquidity lying around?

I don't agree with these guys, but your comments seem to be off in lala land.

you're right in the sense that its impractical to extracts millions from the chain or exchanges.  but i do think there are entities out there that would pay millions to destroy Bitcoin.

Can you give an example of proof-of-stake? Not clear on what you mean.

Neither, it is enjoyable for us "asshats" to use our time to read your rubbish. Do everyone a favor, stop posting anything at all.

Yes pretty please enlighten us ignorant posters what proof-of-stake means.

It means those with the highest coin holdings (wealthiest) are the only ones able to sign blocks and transactions, being able to do it without requiring heavy work. The reasoning is that since they have the most coin wealth, they have the most to lose, and thus they won't try to screw the network, lest they crash the price of the coin and lose everything. The idea is also that they will have the most incentive to protect it.

Prof-of-stake sounds like a good idea only until you realize that it already exists, and is called USD and Federal Reserve.

From a quick read of the paper, it seems that the authors consider price of electricity (in USD) and exchange rate of the bitcoins (BTC/USD) to be independent variables.
This hypothesis is most certainly wrong.
However the price of electricity over time becomes a very small driver of the BTC price as the main driver is the growing demand as the bitcoin economy grows.
If bitcoin serves only a small fraction of today's undergound economy, it can be shown that the price of bitcoins should be in the hundreds of dollars, far above the electricity cost component of the price.

until a shitload more people start mining realizing that there is profit to be had... it's not like we haven't seen this already

until that shitload of people then realize that bitcoin is a great investment because it's worth hundreds of dollars and decide to hoard instead.

because it's a great investment people are going to stop mining? I don't follow your logic one bit

Hashrate has fallen since July. Bitcoin isn't worth mining right now, but it is worth buying.

That all depends on electricity and equipment costs.

and the sell price in july vs the sell price now? I mean really...

It's worth mining for some people who have the right equipment. I agree that it's worth buying for most people more than it's worth mining, right now. Esp. if you don't have any decent mining equipment beforehand.

Hashrate declined sharply right after the Big Crash but has been catching up ever since and it's close to historical maximums already.

The sell price is lower now than in July. Are you saying that makes mining more worthwhile? I fail to see your logic here.  :P

I suppose for some very efficient mining rigs it may be marginally profitable, especially if the heat is valued as well. My mining rig will be shut down very soon if the price does not increase with the outside temperature.

looking at bitcoincharts, the lowest the price dipped in july was around $12. the lowest in august was around $7.

and either way, your post still was completely non sequitur to mine.

Just for reference.

Difficulty and hashrate, linear:

http://bitcoin.sipa.be/speed-small-lin-ever.png

exponential axis:

http://bitcoin.sipa.be/speed-small-ever.png


Gartner hype graph:

http://www.gartner.com/press_releases/images/169368_0001.gif


One has to understand that right after the GPU breakthrough in mining there hasn't really been any game changer (maybe FPGA's if they enter in sufficient quantity and when they do) and also that the Radeon 5xxx HD series hasn't really been beaten in terms of bang for the buck, as one would expect years after their release. Basically you cannot get a much better card than a high end 5xxx series for mining, as one would expect 2 generations later.

I'm glad you think that way. I hope miners keep selling their bitcoin in the $4-5 range. I'll keep buying them and we're all happy.  ;D

this.

it seems that many ppl around here fail to realize that at least 50% of the population is willing to think counterintuitively, aka, buy the dips.

It sounds like a bad idea.  I heard of another one recently called ripple, ripple-project.org. A trust-based system but doesn't seem like a lasting idea.

I read the white paper, I haven't read this entire thread.


My initial problem with the paper was the following:

1 - I don't know of any active botnets working over the gpu.  If they did I suspect that pc users would note the high amounts of heat and active cooling fans and request a fix from however supports their hardware
   a - almost all of the mining is don by gpu now (soon to be fpgas), in other words you would need about 50 standards botnet pc's using their cpu to equal one miner with a gpu
   b - for a perspective of the current cpu power in bitcoin the current fastest networked government owned super computer produces about 4.7 pflops where as the bitcoin community is beyond 53.5 pflops (factor of 10 larger in processing).  to achieve 50% of this with cpu based botnets one would need to acquire 100's of thousands of systems
2 - There is no economical analysis of other values botnets can have.  ie. it may very well be more profitable to operate a botnet to brute force a bank for account access
3 - There is no analysis of the declining rate of bitcoin production.  This is not a never ending game of producing bitcoins.  To date about 40% of bitcoins have been released.  If we were flooded with mass amounts of botnet gpu's it would help to secure the block chain, and also automatically enable competition/incentive for other botnet owners to move into the space (again securing the block chain further), it would cause normal users to exit the market (non-botnet operators), and create a renaissance of new mining innovation