Less Wrong used to like Bitcoin before it was cool. Time for a revisit?
Less Wrong used to like Bitcoin before it was cool. Monthly threads popped up around the same time a pricing bubble brought mainstream attention last year. When the bubble popped, and price continued to deflate, discussion on this site stopped entirely. Was there a change of sign in the social status of the topic, is the topic fully explored, or has there simply happened nothing of interest over the last year?
If you are not familiar with Bitcoin, here is one intro I happen to like.
Kaj Sotala lists a number of previous threads on the topic:
There seems to be quite a bit of a Bitcoin interest around here, with several articles about it already: [1 2 3 4 5 6 7]
Less Wrong seems like a good place to discuss recent developments, if one does not want to suffer the inanity of the officially unofficial forum. If you are not longer interested in Bitcoin, perhaps send your remaining balance to the Singularity Institute?
Clippy noticed that Bitcoin seems to make it easier for software agents to earn money, convince humans to do jobs for them, and optimize the universe in a paper-clip friendly direction.
If Clippy is right, is it a problem?
Singularity Institute visiting fellow Thomas McCabe is running GetBitcoin, a notable money handling service.
Does he have any insights to share?
Gwern wrote an article arguing that Bitcoin is an ugly protocol: Bitcoin is Worse is Better
What does gwern think today?
How should one extend or rework the protocol to make Bitcoin, or a successor more appealing?
Wei Dai is quoted in the original whitepaper. He writes:
What does he think of of Bitcoin and cryptocurrency in general? (Still mining?)
Scott Aaronson recently published a technical paper on quantum money, unrelated to Bitcoin. Would implementing this scheme require keeping qubits stable for an indefinite timespan, for the money not to go poof?
I actually finished that a month or two ago, so it reflects my current sentiments. No events have substantially changed my opinion, and some reinforce them—for example, I thought it was very stupid for anyone to engage in finalization on Silk Road (releasing your money from escrow to the vendor before receiving anything), and indeed, a big vendor recently scammed buyers out of something like a hundred thousand plus dollars worth of bitcoins by requiring finalization and then absconding.
Damn fine writing in that essay. Some of the only intelligent criticism I’ve seen of the protocol.
Thanks.
As with quantum computing itself, storing quantum money would require low enough error rates that you could successfully apply quantum error-correction (if error rates are too high, you introduce more errors during error-correction than you can correct). Before you hit this point you can’t do the computations anyway, and after this point it is only a tiny bit harder to keep qubits around indefinitely. (Modern QKD works because you don’t have to do any computations on the qubits, in addition to not having to store them.)
Needless to say, this is what Wei Dai / Satoshi Nakamoto would say if they were the same person.
Needless to say, this is what Will_Newsome / Satoshi Nakamoto would say if they were the same person.
(Sorry. Played Resistance for the first time yesterday evening.)
Needless to say,
I still use it for some things, but there’s not really a whole lot to discuss; the price is stable, Silk Road is still running, people are still taking it for donations. Infrastructure isn’t always exciting.
One actionable topic that could be discussed: does the current price reflect what we expect Bitcoin’s value to be?
On a related note, has anyone else in the Less Wrong community looked into RipplePay? It’s a system that, instead of being based on a finite currency, is based on distributed transferable IOUs.
Paper fiat money can be interpreted as an IOU from the government, transferable, and valid for use on your taxes. Ripple payments are private IOUs, valid for canceling debts with the issuer. To make a ripple payment to someone who trusts you, you simply issue them an IOU (assuming you haven’t maxed out your credit with them...). To make a ripple payment to someone who doesn’t trust you, the system finds a route along which IOUs can be exchanged (eg Alice gives one of her IOUs to Bob, who gives one of his to Carol, and then Carol gives whatever it was that Alice bought to Alice).
My biggest critique is that there isn’t a distributed protocol, and the only implementation currently relies on a single central server. That, combined with the bootstrap problem, makes it unusable in practice. (BitCoin has a similar bootstrap problem, but less of a critical mass problem, since you don’t have to find a route among people who both trust each other and use BitCoin.)
Comments or critiques, on either the theory or practicality?
The concept of RipplePay seems to require people accepting significant default risk from many people, for no compensation. The size of the risk accepted and the carrying capacity of the network are proportional, such that making it useful requires making the default risks large. The proponents suggest setting up lines of credit with your friends, but this seems like courting disaster; not only would I be risking both money and friendships, I’d have to pay for expensive due diligence, I’d have to occasionally offend people by signaling mistrust, and I’d be setting things up so that if I did go bankrupt (perhaps because I lent too much money to someone who defaulted), my social network would transform into angry creditors.
In exchange for all this risk, I gain nothing: I’m still using an inflating currency, I can’t move large balances quickly or secretly, and I gain a constant chore of settling books with people. And after all that, I still need a traditional bank for value storage (since RipplePay is only attempting to solve the transfer). No thanks. RipplePay will never be anything more than a novelty used for tiny dollar amounts.
As a thought experiment: what if you simply considered RipplePay as a way to determine whose turn it was to pay for the pizza tonight? (This would be predicated on a really friendly UI for such things, obviously.) Would it be interesting in that context?
The idea is that you accept default risk only from people to whom you can apply social arm-twisting. This includes them arm-twisting you not to overextend credit or take risky debt.
Well, for start the single centralized server is a huge liability, previous digital currency schemes have collapsed when the government sued/arrested/raided the central server for enabling money laundering.
Also, unlike bitcoins, IOUs aren’t fungible; thus, there is a need to haggle over each transaction.
The IOUs are transferable, if not entirely fungible. The idea is that the per-transaction haggling is handled by the server (automated route-finding), some fungibility is achieved automatically (circular debts are canceled automatically), and people will act to balance income vs expense by settling large outstanding debts for cash (incentice provided by credit limits).
The single server problem is a huge liability, but distributed route finding and cryptographic chains to provide distributed record keeping seem to me a remarkably easy problem.
I don’t know much about RipplePay, are the IOUs denominated in some standard unit?
In the current implementation, they can be denominated in any of several typical currencies. I’ve never played with the exchange rate side of things, so I don’t know off hand how that works in practice. (I’ve only used ripplepay a tiny bit in total, for tracking a couple small debts between friends.)
I was wondering about this only the other day. If you’re wondering “what became of BitCoin after the price peaked last year”, here’s the price chart (log scale—see tick boxes to change)
Note: log scale.
Linear-scale version: http://bitcoincharts.com/charts/mtgoxUSD#igWeeklyztgSzm1g10zm2g25
Wow … that actually looks to be stabilizing.
Good point—have edited to make clear.
This story came up on Hacker News recently (article in question): a company is creating chips that apparently can do 1 terahash/s, and currently pricing them for around $30K.
As this comment points out, the total combined hash rate of the entire bitcoin network is about 12 TH/s (source). So buying just one of these chips gives you >7% of the current total hashing power. In theory this is about 21 BTC (or 130+ USD) per hour, which means you break even after 230 hours, or fewer than 10 days!
Assuming no one else is doing the same thing.
Of course.
Presumably the company in question could easily manufacture a whole bunch for itself and get a significant portion of the bitcoin market (although it would have to be careful about destroying the currency’s value).
They can get 100% of the mining and transaction fees market by snapping their fingers. Once they’ve made their initial investment into manufacturing the chip, the marginal cost of making more of them is minimal. Far below the $30k they’re selling the 1 TH solution for.
They can grab 50%+ of the mining market for themselves pretty easily. Then, they can increase their capacity to keep up with the growth of the network for cheap—all while they sell their processors to others, who pay much more for the mining capacity they get, than it costs for Butterfly Labs to make more.
It’s possible to crash the currency if you’re in a position where you can reliably mine more than 50% of the blocks. When you have more than 50% of the network’s computing power, you can exclude other miners—gain 100% of the market by simply ignoring everyone else’s blocks. The P2P network will respect your chain, because it’s longer. Then, you can impose any transaction fees you like, or refuse to process any transactions at all.
However, unless you intentionally do things like that to crash the currency, its value doesn’t come from the miners. It comes from people who use it for transactions and for storing value, which aren’t necessarily the same people who mine the currency.
Last year, MBlume suggested:
This now exists at BetsOfBitcoin. I signed up one exactly one month ago.
So far, I’ve won 13 out of 13 bets. Mostly small bets and low yields, but winning is fun.
I warn people against this site. Their adjudication of bets is highly dubious and their customer service just screams “sham”.
I no longer bet at the site.
Could you please explicate your complaint? I’ve had no disputes so far.
I found less than a handful of complaints in the announcement thread.
Bets (are supposed to) get rejected if the outcome is not easily decidable.
Except it’s not really a prediction market. You could know the exact probability of an event happening, which is different from the market’s opinion, and still not be able to guarantee profit (on average).
Bitcoin’s primary function in my life is to provide a constant temptation to say ‘screw this nonsense’ and earn significant sums of money very quickly, though bitcoin-enabled anonymous crime.
I have a strictly academic interest in this subject. What were you thinking of?
Well, game theory dictates that I shouldn’t tell you about my really good ideas, in case I decide to use them in the future.
But there’s some obvious low-hanging fruit. Prices on the silk road are significantly inflated, and the odds of drugs being intercepted in the mail are extremely low. A small marijuana grow operation (or the synthesis of more sophisticated drugs, if you have the chemical background), could turn a reasonable profit very quickly. If you can produce a high-quality product, revenue of even a very basic operation could potentially be tremendous.
There’s also identity theft. You can buy credit card information using bitcoins, use the compromised credit cards to buy more bitcoins, and then discard the cards immediately. This one strikes me as less palatable to the nominally moral man than the first one. Probably more profitable, though.
If the enterprise in question has a second-mover advantage, game theory dictates it could be in your best interests to tell me your really good ideas :-)
...but not if it has a first mover advantage. I mean, aside from the ‘second mouse gets the cheese’ factor.
Do you know a Bitcoin exchange that takes credit cards directly, or are you imagining an intermediate step like money orders?
It looks like the one I saw in the past may have folded, as I can’t find it. In which case, it might be necessary to use an intermediary, like moneygram. For that sort of large-scale credit card theft, you’d want something fast and automatable, to try to earn the maximum per-card return in the interval before the owners notice and disable them.
Looks like you may be too late. The US based Bitcoin exchange Tradehill had to shut down operations earlier this year because of this exact method of fraud. Here is a description of the modus operandi. The fraud was conducted via an intermediary payment processing service.
Interesting. Not that I was actually planning on committing large scale identity theft, mind.
Looks like you are may be too late. The US based Bitcoin exchange Tradehill had to shut down operations earlier this year because of this exact method of fraud. Here is a description of the modus operandi. The fraud was conducted via an intermediary payment processing service.
Looks like you are may be too late. The US based Bitcoin exchange Tradehill had to shut down operations earlier this year because of this exact method of fraud. Here is a description of the modus operandi. The fraud was conducted via an intermediary payment processing service.
For one thing, stealing bitcoins is far safer and easier than most other types of theft.
Thinking about Bitcoin always makes me think whether it could be profitable to invent a new “Bitcoin 2.0”—officially it would provide some improvements over the old protocol, but of course the real reason would be to have all those new free coins just for yourself.
In case of success, this has a potential to bring a lot of money. Therefore some money could be spent now, for example to weekly spam Slashdot with articles about the advantages of “Bitcoin 2.0”.
If there is a good chance of a successful takeover, this is what rationalists should be doing. :D
(As a side effect, such takeover, especially if repeated, could destroy the credibility of digital currency. Yes, there is a limited number of coins in a bitcoin protocol, but there is an unlimited number of protocols. At that moment, the next step would be to invent a new “Bitcoin Omega” to fix this problem. Etc.)
It’s actually a good idea to think about this problem. There must be some way to make the hashing contest useful for some other purpose; or if not, a better hash than SHA can be picked (like memory-bound hash functions).
Why would anyone switch? Well, SHA will be broken sooner or later. When that happens, people will be interested in switching to a Bitcoin 2.0 (eg. by destroying regular bitcoins in exchange for new bitcoin2.0s).
EDIT: and clean up other parts of the protocol: http://en.bitcoin.it/wiki/Hardfork_Wishlist
There have been several attempts at this. They are generally seen as scams. There are a few that attempt to have actual added value (I personally think Namecoin has potential), but most just tweak the constants.
Bitcoin’s mining system seems odd.
There’s a reason we don’t make coins out of precious metals. Mining them is a waste of resources. Mining bitcoins has the same problem. If they’re worried about people not liking them because of the profit they’d get, can’t they just donate the money to charity or something?
There are several reasons why we don’t make coins out of precious metals, and also a few why we should. The wastefulness of resources is relatively low down the list.
The point of a commodity-backed currency is to provide stability in the money supply. Provided the commodity is widely regarded as valuable, but isn’t likely to change in raw value in the long term, having x amount of money represent y amount of that commodity means that your money is a safe store of wealth. If you want to earn lots of money now and spend it in the future, you can be confident that inflation isn’t going to devalue your savings in the interim.
This is one of the reasons the price of gold is negatively correlated with equity performance. In Bear markets, people buy gold because they believe it’s less likely to lose value than holding your wealth in stock or currency, and generally they’re correct to hold this belief. In part, this then becomes a self-fulfilling trend, as gold itself becomes a good investment at the start of market decline. Partly because of this, you’ll find armchair investors are especially keen on the Gold Standard, because that’s what they think money is for.
There are problems with this, though, namely that inflation isn’t just a monetary phenomenon, and currency isn’t just a store of wealth. Money is also a medium of transaction. There has has to be enough money, in sensible units, to represent the value of all the goods and services that are traded with it, so as the economy grows, there has to be a proportional increase in the money supply, otherwise your currency will appreciate to unwieldy levels and you’ll suffer from effective deflation.
Gold is a bad candidate for fulfilling this function of money. There’s a finite amount of it. We’ve already mined most of what’s easily obtainable, and eventually any given good or service will be represented by an increasingly small quantity of gold. Any changes to the gold supply are going to be sudden shocks to the economy (say I irradiate a whole bunch of it like in Goldfinger, we discover a massive new supply of it, or a huge gold asteroid drifts into Earth’s orbit). Also, part of the usefulness of gold as a basis for currency was that it wasn’t used for a whole lot else. As various industrual uses were found for it, that impacted both those industries and the currency. If we found out tomorrow that eating gold cured cancer, increased penis/breast size, and gave people superpowers, that would be bad news for a gold-backed currency.
The purpose of the bitcoin mining process is to provide a stable supply of bitcoin. Since each block adjusts the difficulty of producing the next based on the efficacy of mining currently taking place, the production of bitcoin should be continual but stable, (at least until we mine them all).
I am still thinking about the ramifications of bitcoin, and what properties it has as a currency, but the method of bitcoin production is one of the things that makes it novel and special as far as currencies go. It can function as both a store of wealth and a medium of transaction without those functions necessarily working against each other as they do with traditional commodity-backed currencies.
A nit pick:
The key idea here is not that you need enough currency to complete all transactions (since money gets reused many many times), but that people’s desire to hold a currency is a key determinant of its value. If people’s desire to hold a currency increases in real terms (such as during , but the nominal amount of money does not increase then the currency has to appreciate. This can lead to unwieldy currency units as you mention, but it also has bad dynamics. An overall shortage of money will lead to a reduction of economic activity as people try to get more currency (this logic also works in the reverse direction).
Fiat (public or private) currencies have the advantage that they can increase or decrease the amount of their currency in circulation at will using open market operations to match people’s overall desire to hold their currency. Whether this advantage is actually an advantage depends on how good of a job you think central banks do of actually doing this.
Inflation talk like this always makes me antsy. I pretty much agree with what you say above, but a monetary interpretation of inflation, specifically as regards the money supply relative to total economic activity, strikes me as the most sensible way to talk about why we’d experience de facto deflation with an exhausted commodity-backed currency in a growing economy, which was the salient point as far as bitcoin and the Gold Standard are concerned.
I don’t think the desire-to-hold interpretation of currency value is incorrect (in a sense it’s almost tautologically true), but in this case I think it makes sense to view that as a repercussion (an increase in demand for finance with no corresponding increase in real loanable funds, for example), rather than as a driving factor. The money supply is opaque to most agents in an economy, and their desire to hold onto it is based on how much they can get away with sitting on at any given time. If the money supply is small in relation to the number of transactions they have to make, that number must, by necessity, decrease.
Edit: Wait a minute. That last bit doesn’t make any sense at all. If I’m an agent in an economy, my only evidence for the true value of the money supply is how much I can buy with it relative to what I could buy with it in the past. After a period of growth I should be able to purchase a relatively larger amount with the same quantity of currency, so in that sense I guess my desire to retain currency is a direct effect of growth with a fixed money supply without even having to think about the size of the money supply in relation to transactions. They are implicit in whatever background activity is responsible for the growth, however.
Hm. OK. I could be convinced that either angle is useful for addressing the issue. I guess it depends on where you think the action is. The example of number of borrowers in relation to loanable funds is definitely one for the money-supply-to-economic-activity ratio. It’s also possibly best dealt with in terms of interest rates, but it suggests a category of coarse, high-capital exchanges which are susceptible to policy intervention. On the level of individual economic agents, currency-retention is probably the easier idea to work with, but it’s dealing with very fine-grained aggregate processes that are less tractable in terms of interventionist policy.
I don’t understand what you mean. Are you just saying “the way I explained it is a quick way to communicate the sort of problem I have in mind”? Your mention of inflation confuses me because neither of us had mentioned it before.
I didn’t mean to say “the value of money = desire to hold it”, which I think would probably be wrong or ill specified. A price is the rate at which you can trade one good for another good. Money has lots of different prices (2$ per lb apples, $300 per oz gold etc., $50 per hour of massage etc.) since it participates in lots of different markets, whereas most goods only participate in one market (where they can be traded for money). You can talk about a “price level”, but coming up with a precise and useful definition is a little tricky.
I just meant that if for each agent, the marginal utility of money rises (or you add more agents) and the quantity of money does not rise then at equilibrium the price of money will have to rise across the board (deflation). I was saying that your example of economic growth leading to more transactions is a special case and trying to explain that broader framework.
There are many reasons why people’s general desire to hold money might rise or fall, for example if a checking mechanism is introduced then people will need to hold less money to conduct their transactions (reduced desire to hold money) or if there’s a global financial crisis, people might say “holy shit, I can’t trust any of these assets, I’d best just hold money” then people’s general desire to hold money would rise.
Are we communicating better?
Also I categorised this as “inflation talk”, because it was wheeling out a few of the concepts behind the question “what are the causes of inflation?”
I think there’s something about macro that makes me communicate badly.
Sorry...you might want to see my edit, which I started before reading this response, and which basically (although less concisely) repeats your explanation about marginal utility of money rising for each agent.
I have totally confused the issue, and you are quite right.
That makes more sense to me. Have you heard of Monetary Disequilibrium theory? It’s the most reductionist approach to monetary economics, and what I was describing more or less.
Not by name, but it seems highly compatible with many other accounts of macroeconomic fluctuation.
As you may have gathered, macro is not my strongest suit. I mostly hopped on board an econ syllabus for other reasons, and now find myself with a whole bunch of useful and explanatory macroeconomic concepts that most other people have never heard of.
You do understand why Bitcoin is analogous to making coins out of precious metal rather than fiat, and what purpose the mining system serves, right?
The mining system provides the block chain that verifies transactions; yes. And eventually the payoff to miners becomes dominated by transaction fees rather than by payouts for finding blocks.
One of my concerns with Bitcoin is that today, on the margin, choosing to accept BTC in exchange for goods means agreeing to give lots of goods, on demand, to the people who currently are sitting on a lot of BTC — who have not (on the whole) done anything particularly useful to earn that wealth. Even a merchant who thinks a Bitcoin-like system is a good idea may be reluctant to agree to participate in making a bunch of nerds with video cards very, very rich.
I don’t think the merchants really care about that. They care about make themselves rich. If making nerds with video cards also rich is a side effect, so be it. But if nerds with video cards can dilute the value of their currency, they may have a problem. If this dilution occurs on a known schedule and will slow down over time and eventually stop altogether, and competing currencies can be diluted at any time by central bankers, it may not be much of a problem.
A bit of a tangent, but the increase in supply of bitcoin is not the only thing that affects the value of bitcoin. A key determinant is how much people want to hold bitcoin. How much people want to hold a currency is not a very stable thing, meaning that the even with a fixed quantity of bitcoin in circulation, the future value of bitcoin will be highly uncertain. This will also tend to produce gluts and shortages of bitcoin, which have their own negative effects.
Doesn’t regular money have this same problem?
Sure, in a sense; but merchants don’t really have a meaningful choice of whether to accept it. BTC is new.
Well the problem is arguably worse with regular money since the number of bitcoins that will ever exist is limited, whereas with regular money governments can, and sometimes do, print arbitrarily large quantities.
Tangent:
Have you heard the criticism that bitcoin is significantly suboptimal because the quantity of bitcoin can’t adjust the quantity to fluctuating demand to hold bitcoin? Do you take it seriously?
I’ve heard it, and I don’t take it too seriously: adjustments are as adjustments do. A fiat currency like the US dollar seems to be getting the worst of both worlds—vulnerability to pathologies like hyperinflation, and yet, the Fed won’t loosen money despite inflation expectations that verge on zero or negative. What’s the point, then?
Gotcha, I guess it depends on what you’re comparing bitcoin to.
Though bitcoin will generally be in constant deflation at a bit less than the rate of economic growth. Some people think that would be bad, though I personally don’t have a strong opinion about that.
I’ve heard it argued that the predictability of the amount of inflation/deflation could be more important for stability than the actual amount. If so, that would be an advantage for bitcoin over other currencies.
Do you have any thoughts on that?
EDIT: just saw your other comment where you said, “A bit of a tangent, but the increase in supply of bitcoin is not the only thing that affects the value of bitcoin. A key determinant is how much people want to hold bitcoin. How much people want to hold a currency is not a very stable thing, meaning that the even with a fixed quantity of bitcoin in circulation, the future value of bitcoin will be highly uncertain. This will also tend to produce gluts and shortages of bitcoin, which have their own negative effects.”
I suppose that answers my question.
Yup, predictable inflation/deflation is going to be better than unpredictable inflation/deflation because people can plan for it. Bitcoin will have uncertainty about its future value. I don’t know whether this would be more or less than for USD.
As a side note: because money is an asset, expectations about its future value are very important for its present value so most of the time “predictable inflation/deflation” will mean “constant inflation/deflation”.
The following is just cause I misunderstood your post at first:
Predictable inflation/deflation can have real effects if it’s expensive for plan for it. For example if it’s not very practical to index things to inflation or costly to change prices (supermarket prices have frequent sales, but their reference prices only change about once a year).
Another way: if central banks don’t pay interest on cash, then people will have inappropriate incentives about how much money they should hold. If you have strong inflation and pay 0 interest on cash then it has a large negative real return and you punish people for holding it even if that doesn’t reflect real costs. If you have strong deflation and pay 0 interest then cash has large positive real return and you reward people for holding it even if this doesn’t reflect actual benefits. The interest rate on cash should probably be lower than short term real interest rates since otherwise people will choose to hold cash rather than other short term securities leading to an explosion in the demand for cash (and if it were higher, the currency issuer would probably be losing money). The Fed recently (2008) started paying interest on reserves (which is cash banks hold at the Fed), but they set that interest rate in a stupid way (it should probably be negative right now).
Some things I would like to know:
Clippy noticed that bitcoin could make it easier for software agents to earn money, convice meatspace agents to do jobs for them, and optimize the universe in a paper-clip friendly direction.
If Clippy is right, is this a problem?
Singularity Institute visiting fellow Thomas McCabe is running GetBitcoin, a notable money handling service.
Does he have any insights to share?
Gwern wrote an article arguing that Bitcoin is an ugly protocol : Bitcoin is Worse is Better
What does gwern think today? Could one extend or rework the protocol to make Bitcoin, or a successor more appealing?
Wei Dai is quoted in the original whitepaper. He writes:
What does Wei Dai think of the of Bitcoin and cryptocurrency today?
Scott Aaronson recently published a technical paper on quantum money, unrelated to Bitcoin. Would implementing this scheme require keeping qubits stable for an indefinite timespan, for the money not to go poof?
I still have more faith in the future of the Magic Online Event Ticket as a digital currency.
Big oof here.
I felt the same way. Was curious about what sentiment was back then, because I very often see people making predictions using the early years as some kind of first cycle. I really do not think it is analogous.
I have since lost some faith in the Magic Online Event Ticket. I do not have any more faith in Bitcoin than I did eight years ago. Furthermore, I’m fairly confident that Bitcoin is of net negative value to the world—a huge amount of energy is devoted to computing otherwise useless hashes so hackers can collect ransoms and some people can try to take money from greater fools.
(There are blockchain projects that might have actual value, but Bitcoin is not one of them.)