Who said anything about having just a few validating nodes? If Bitcoin ever gets as big as the major card networks there will probably be tens of thousands of nodes, probably even more. The internet scaled to a billion people using backbone routers that use a floodfill network and a single global routing table. Maybe the guys who built the backbone should have just given up and said the masses can stick with AOL?
You probably should apologize, because you haven't understood what happened at all.
Of course the Bitcoin protocol is versioned. Did you really think it wasn't? The issue that happened here is that there was a previously unknown bug in older versions that meant they failed to process data that they should have been able to process. The nature of Bitcoin is very unusual - everyone in the system has to independently perform complex calculations and arrive at exactly the same result. In this case, the newest version of the software arrived at the correct result and the older version puked and died, because of an issue with the Berkeley DB it used (the new version uses LevelDB).
If that problem had been known about before, then it would have been versioned away. However, it wasn't.
There isn't really anything directly comparable to Bitcoin in this regard, but I suppose your rant is a bit similar to saying, "stupid Opera developers, their crap doesn't render my web page. Don't they understand versioning?" when the problem is that it's not emulating a bug in Internet Explorer.
Is that you Peter?:) I don't know anyone else who would describe the system like that.
Bitcoin is not an O(n^2) network. The total work done by the core nodes is "number of nodes multipled by number of transactions" - that is two very different quantities. It doesn't help the argument to abuse formal notation like that.
Pieter reported today that his optimized secp256k1 implementation can reach 20,000 signature checks per second on a single core. That is a lot better than the generic OpenSSL code was able to manage. Bear in mind that although it sounds scary, all of VISA is only about 10,000 transactions per second. So even if we're generous about our assumptions on number of inputs, CPU load would barely stress a single core. Today. With existing hardware. Disk IO will be the bottleneck for the forseeable future, but we're talking about a working set of a few hundred megabytes today. Even with an order of magnitude growth the entire thing fits in RAM on my puny laptop. And with two orders of magnitude growth it still fits on a pretty average dedicated server.
But Bitcoin will not reach VISA traffic loads today, or this year, or even this decade. By the time it does, dedicating one single computer to being a node will seem a completely trivial investment in order to have full security on the worlds financial system, because let's face it, if Bitcoin is processing tens of thousands of transactions per second then it's well on its way to being exactly that.
Ultimately, Bitcoin is growing fast and will scale up. That is what Satoshi wanted it to be, that is what most of the people actually creating businesses and writing lots of code also want - so if you think that'll make your coins worthless in a few years, I will happily buy them off you.
It's worth noting, that Berkeley DB (the one with the weird limits) was already replaced. The problem is that not enough users have upgraded to the replacement yet, and it's better to match their brokenness than diverge unexpectedly. They'll have to upgrade sooner or later though.
No, they are not saying that. If you look at the announcements, people are being told to use 0.7 if they are miners, and for everyone else it doesn't matter what version they're on (so you might as well be on the latest version). At some point miners will need to switch to 0.8 too. It's a matter of co-ordination.
Yes. At some point everyone will have to upgrade to 0.8 and people who don't will effectively be disconnected from the system. However that needs some kind of co-ordination. The fact that there was a chain split is not the emergency, the fact that it came unexpectedly was the problem.
The Bitcoin protocol has support for dispute mediation in it (actually, 2-of-3 signing for coins). Unfortunately the surrounding ecosystem does not exist... the features aren't exposed via GUIs and there are no dispute mediators who support it. But probably it will come in future. Right now there doesn't seem to be much demand, many sellers have been able to build a trustworthy reputation.
The DNS registrar actually spoke about this incident publicly - it turns out that there was no social engineering, BitInstant just selected dumb security questions/answers when they registered the domain name. It's poor security on BitInstants part, no more or less.
That's pretty much what all hedge funds do, isn't it? Pick some asset they think will grow in value, buy it up (often using leverage), and then wait to see if their bet works out. Often they wait long periods of time. The fund is being targeted at people with lots of money and enormous appetite for risk - for these people, there aren't enough direct investment targets (like startups) so the easiest way to invest in the future success or failure of Bitcoin is indeed, buy and hold.
More to the point, if they're at the level of doing secret sharing on private keys they are quite capable of implementing their own fund management software. For instance bitcoinj makes it easy to implement watch-only wallets, there's even a command line tool for it. I really doubt they need the help of GUI tools to set that up.
Yes, you could do that, and in fact they did. However it doesn't really help because there's no supporting infrastructure for people to download lists of tainted outputs and trigger alerts when a transaction rooted on that output shows up in your wallet, and if there was, it isn't clear what you would do about it, and even if it was, it isn't clear how you'd stop people gaming the system by claiming coins as stolen when they actually were not. But it's technically feasible.
Given that the only people who knew what was in the files before BM were the people in the government, it's hard to see why it'd cause any changes in government policy. It did embarrass a bunch of politicians in other parts of the world (and revealed a US spy in the German government), but mostly these days those other parts are too focused on domestic economic problems to think much about foreign policy.
I think the impact of what Manning did is real, but it'll be a slow burn, long term kind of thing. People read day after day in the news about the US drone strike program, but whilst ink on paper is one thing, seeing a video of a bunch of journalists get nuked from the air is something quite different. It really brings it home to people in ways other mediums just can't. The other thing it did is expose to what extent much of the rest of the world had become servile to US interests, for instance, the al-Masri story was quite shocking and I think the cables were really the first time hard evidence surfaced that it was true. Before that the best evidence that the story was correct was isotopic testing of his hair. It also revealed that the German government had basically been penetrated and owned by the CIA at several levels.
Just generally there's a ton of useful background on so many issues in those cables that people will be using them as evidence to back up positions for years to come.
You realize that most European countries are both cutting spending AND raising taxes, right? That's normally how one tries to balance a budget - increase revenue and decrease costs. All "austerity" really means is reducing government spending, and America will have to do that sooner or later, even the Dems don't disagree about that...
How many data points would you need to accept the outcome? Let me flip it around - what evidence is there that the theory is correct? It's not only historical data that contradicts this theory. Smaller markets also do, like consumer electronics.
Economics as a discipline isn't invalid, but you do have to take it with a huge pinch of salt especially at the macro level.
Here are a couple of facts that should make you think twice about macro-economic wisdom. One of the most basic tenets of science is to compare your theories against observable reality to see if they match, and if they don't, you come up with a better theory. But the theory of the deflationary spiral was not actually studied to see if it matched historical data until 2004, and then when it finally was analyzed, it was discovered that the theory was wrong. Should have been a pretty huge event, but no, Bernanke and his colleagues continue to act as if the study was simply never done.
If that theory is wrong, what other conventional wisdom might be wrong? Well, it turns out that economists have models of how the economy works. Of course they do. The state of the art in macro-economic modelling, widely used by central banks, are dynamic stochastic general equilibrium models (DSGE). Sounds sophisticated, right? Wrong. These models are so crude they do not include banks at all because the people who designed them thought that banks had no impact on economies. Worse still, as their name implies the models predict equilibrium rather than the boom/bust cycles that typify real economies. It might seem obvious to the man on the street that if your explanation of how the economy works ignores banks and predicts stability, then you have a pretty bad explanation... but this is the quality of science on which central banks base their decisions.
Is it any wonder the world got so messed up? Maybe you should indeed exercise more skepticism towards the so-called "dismal science", and consider whether us Bitcoiners have got it right after all.
The good thing of inflation is that it keeps people from hoarding cash, as hoarding cash means you're losing value.
How is "hoarding cash".... also known as saving.... equivalent to losing value? In a stable financial system saving money should be value neutral. In the financial systems we actually use, it's the opposite - inflation is the reason that savings lose value. This is not a good thing!
The design of Bitcoin is correct. It's the design of existing currencies that's hosed, though possibly calling the result of a few centuries of evolution "design" is overly generous.
Let me sum up our existing currencies for you in two paragraphs. Because pre-Bitcoin we lacked a way to globally synchronize everyones view of each others balances, we have to delegate this job to organisations called banks. We trust them to decrement one accounts balance when another is incremented and this is why people have to, you know, go to work and more or less live within their means. Naturally the trust we place in them for that is constantly abused, because the banks routinely increment peoples accounts without decrementing anyone elses. They charge interest for this "service". They especially love incrementing the governments accounts in this way, but the free money doesn't just go there, it sprays all over the place. This eventually leads to inflation.
Inflation is extremely convenient for governments because it lets them buy nice things for voters and pay for it using an invisible tax on savings. Because people would be naturally very angry at this taxation if it was done directly, economists and politicans came up with a brilliant theory as to why this is actually a good thing! Zero inflation, says them, would be terribly bad because if peoples savings aren't being constantly stolen at 2% per year or more then they lose all interest in profit. Given the choice of sitting on their money and getting no return, or making a safe 2%-per-year investment, those stupid citizens will choose to sit on it, and that will lead to economic collapse and anarchy and cats living together with dogs, which would never do.
There's a couple of problems with this theory. One is that it is contradicted by actual studies of historical data. The other is that it makes no sense. Capitalism is based on the assumption that people like profit. Normally, if someone is offered a low or even zero risk way to profit, they'll go for it unless they need the money right now. But when it comes to the (non existent, it turns out) "deflationary spiral" we're expected to suddenly stop believing that.... we're told that people will pass up good investment opportunities unless they're forced into it. That flatly contradicts how people actually work. The only kind of investment that becomes interesting solely due to inflation is the bad kind. You know, the kind that tends to be involved with asset bubbles.
By the way, Bitcoin is not intended to be deflationary. You assert that people will lose Bitcoins and sure, a tiny amount might disappear that way. But you'd better expect that easy and automatic backup will become common in future. All the incentives to properly solve that problem are there.
Encryption is not in any way "free". The calculations might be fast but the complexity of the key distribution and management infrastructures are never free.
Secure microkernels in ROM with mandatory enforced security models did go mainstream, actually. Hundreds of millions of people have one in their living room, it's called a games console and they have a pretty enviable track record of being difficult to hack even by extremely competent attackers who have physical access to the machine. The sad thing is that despite Microsoft having a lot of in-house expertise in how to build these kinds of system, they deployed exactly none of it to build secure business workstations, instead that's been left to Google with ChromeOS which (you guessed it) is made up of layers of sandboxes with a secure ROM that integrity checks the next level up before running it. If you disable extensions, which I think admins can do remotely for managed domains, it's proven extremely difficult to penetrate.
Posting even though it'll undo my mods. Just had to point out that the linked story is about a kill of a person. Drones are much larger than people and follow much more predictable paths. Presumably if you can make the bullet go far enough, it might be possible with luck and on a still day to get a bullet into the machine somewhere, and cause some damage. Also drones can't easily run away or perhaps even notice if somebody is shooting at them, so you'd get more attempts than with a person.
Slashdot Mirror
The irony level of that post almost makes my head explode. Tough Love indeed!
Who said anything about having just a few validating nodes? If Bitcoin ever gets as big as the major card networks there will probably be tens of thousands of nodes, probably even more. The internet scaled to a billion people using backbone routers that use a floodfill network and a single global routing table. Maybe the guys who built the backbone should have just given up and said the masses can stick with AOL?
You probably should apologize, because you haven't understood what happened at all.
Of course the Bitcoin protocol is versioned. Did you really think it wasn't? The issue that happened here is that there was a previously unknown bug in older versions that meant they failed to process data that they should have been able to process. The nature of Bitcoin is very unusual - everyone in the system has to independently perform complex calculations and arrive at exactly the same result. In this case, the newest version of the software arrived at the correct result and the older version puked and died, because of an issue with the Berkeley DB it used (the new version uses LevelDB).
If that problem had been known about before, then it would have been versioned away. However, it wasn't.
There isn't really anything directly comparable to Bitcoin in this regard, but I suppose your rant is a bit similar to saying, "stupid Opera developers, their crap doesn't render my web page. Don't they understand versioning?" when the problem is that it's not emulating a bug in Internet Explorer.
That explanation is nonsense. The hardware has no knowledge of what's in the blocks. It deals only with the headers.
Is that you Peter? :) I don't know anyone else who would describe the system like that.
Bitcoin is not an O(n^2) network. The total work done by the core nodes is "number of nodes multipled by number of transactions" - that is two very different quantities. It doesn't help the argument to abuse formal notation like that.
Pieter reported today that his optimized secp256k1 implementation can reach 20,000 signature checks per second on a single core. That is a lot better than the generic OpenSSL code was able to manage. Bear in mind that although it sounds scary, all of VISA is only about 10,000 transactions per second. So even if we're generous about our assumptions on number of inputs, CPU load would barely stress a single core. Today. With existing hardware. Disk IO will be the bottleneck for the forseeable future, but we're talking about a working set of a few hundred megabytes today. Even with an order of magnitude growth the entire thing fits in RAM on my puny laptop. And with two orders of magnitude growth it still fits on a pretty average dedicated server.
But Bitcoin will not reach VISA traffic loads today, or this year, or even this decade. By the time it does, dedicating one single computer to being a node will seem a completely trivial investment in order to have full security on the worlds financial system, because let's face it, if Bitcoin is processing tens of thousands of transactions per second then it's well on its way to being exactly that.
Ultimately, Bitcoin is growing fast and will scale up. That is what Satoshi wanted it to be, that is what most of the people actually creating businesses and writing lots of code also want - so if you think that'll make your coins worthless in a few years, I will happily buy them off you.
-Mike Hearn
It's worth noting, that Berkeley DB (the one with the weird limits) was already replaced. The problem is that not enough users have upgraded to the replacement yet, and it's better to match their brokenness than diverge unexpectedly. They'll have to upgrade sooner or later though.
No, they are not saying that. If you look at the announcements, people are being told to use 0.7 if they are miners, and for everyone else it doesn't matter what version they're on (so you might as well be on the latest version). At some point miners will need to switch to 0.8 too. It's a matter of co-ordination.
Yes. At some point everyone will have to upgrade to 0.8 and people who don't will effectively be disconnected from the system. However that needs some kind of co-ordination. The fact that there was a chain split is not the emergency, the fact that it came unexpectedly was the problem.
The Bitcoin protocol has support for dispute mediation in it (actually, 2-of-3 signing for coins). Unfortunately the surrounding ecosystem does not exist ... the features aren't exposed via GUIs and there are no dispute mediators who support it. But probably it will come in future. Right now there doesn't seem to be much demand, many sellers have been able to build a trustworthy reputation.
The DNS registrar actually spoke about this incident publicly - it turns out that there was no social engineering, BitInstant just selected dumb security questions/answers when they registered the domain name. It's poor security on BitInstants part, no more or less.
The article says that they've already thought of that, and taken steps to prevent it - hence the use of secret sharing and other threshold schemes.
That's pretty much what all hedge funds do, isn't it? Pick some asset they think will grow in value, buy it up (often using leverage), and then wait to see if their bet works out. Often they wait long periods of time. The fund is being targeted at people with lots of money and enormous appetite for risk - for these people, there aren't enough direct investment targets (like startups) so the easiest way to invest in the future success or failure of Bitcoin is indeed, buy and hold.
More to the point, if they're at the level of doing secret sharing on private keys they are quite capable of implementing their own fund management software. For instance bitcoinj makes it easy to implement watch-only wallets, there's even a command line tool for it. I really doubt they need the help of GUI tools to set that up.
Yes, you could do that, and in fact they did. However it doesn't really help because there's no supporting infrastructure for people to download lists of tainted outputs and trigger alerts when a transaction rooted on that output shows up in your wallet, and if there was, it isn't clear what you would do about it, and even if it was, it isn't clear how you'd stop people gaming the system by claiming coins as stolen when they actually were not. But it's technically feasible.
Right, banking websites are only ever made by experts, no coderz or buttards there.
Given that the only people who knew what was in the files before BM were the people in the government, it's hard to see why it'd cause any changes in government policy. It did embarrass a bunch of politicians in other parts of the world (and revealed a US spy in the German government), but mostly these days those other parts are too focused on domestic economic problems to think much about foreign policy.
I think the impact of what Manning did is real, but it'll be a slow burn, long term kind of thing. People read day after day in the news about the US drone strike program, but whilst ink on paper is one thing, seeing a video of a bunch of journalists get nuked from the air is something quite different. It really brings it home to people in ways other mediums just can't. The other thing it did is expose to what extent much of the rest of the world had become servile to US interests, for instance, the al-Masri story was quite shocking and I think the cables were really the first time hard evidence surfaced that it was true. Before that the best evidence that the story was correct was isotopic testing of his hair. It also revealed that the German government had basically been penetrated and owned by the CIA at several levels.
Just generally there's a ton of useful background on so many issues in those cables that people will be using them as evidence to back up positions for years to come.
You realize that most European countries are both cutting spending AND raising taxes, right? That's normally how one tries to balance a budget - increase revenue and decrease costs. All "austerity" really means is reducing government spending, and America will have to do that sooner or later, even the Dems don't disagree about that ...
How many data points would you need to accept the outcome? Let me flip it around - what evidence is there that the theory is correct? It's not only historical data that contradicts this theory. Smaller markets also do, like consumer electronics.
Do you know of a way to create a new currency and associated economy from scratch without volatility?
Economics as a discipline isn't invalid, but you do have to take it with a huge pinch of salt especially at the macro level.
Here are a couple of facts that should make you think twice about macro-economic wisdom. One of the most basic tenets of science is to compare your theories against observable reality to see if they match, and if they don't, you come up with a better theory. But the theory of the deflationary spiral was not actually studied to see if it matched historical data until 2004, and then when it finally was analyzed, it was discovered that the theory was wrong. Should have been a pretty huge event, but no, Bernanke and his colleagues continue to act as if the study was simply never done.
If that theory is wrong, what other conventional wisdom might be wrong? Well, it turns out that economists have models of how the economy works. Of course they do. The state of the art in macro-economic modelling, widely used by central banks, are dynamic stochastic general equilibrium models (DSGE). Sounds sophisticated, right? Wrong. These models are so crude they do not include banks at all because the people who designed them thought that banks had no impact on economies. Worse still, as their name implies the models predict equilibrium rather than the boom/bust cycles that typify real economies. It might seem obvious to the man on the street that if your explanation of how the economy works ignores banks and predicts stability, then you have a pretty bad explanation ... but this is the quality of science on which central banks base their decisions.
Is it any wonder the world got so messed up? Maybe you should indeed exercise more skepticism towards the so-called "dismal science", and consider whether us Bitcoiners have got it right after all.
How is "hoarding cash" .... also known as saving .... equivalent to losing value? In a stable financial system saving money should be value neutral. In the financial systems we actually use, it's the opposite - inflation is the reason that savings lose value. This is not a good thing!
The design of Bitcoin is correct. It's the design of existing currencies that's hosed, though possibly calling the result of a few centuries of evolution "design" is overly generous.
Let me sum up our existing currencies for you in two paragraphs. Because pre-Bitcoin we lacked a way to globally synchronize everyones view of each others balances, we have to delegate this job to organisations called banks. We trust them to decrement one accounts balance when another is incremented and this is why people have to, you know, go to work and more or less live within their means. Naturally the trust we place in them for that is constantly abused, because the banks routinely increment peoples accounts without decrementing anyone elses. They charge interest for this "service". They especially love incrementing the governments accounts in this way, but the free money doesn't just go there, it sprays all over the place. This eventually leads to inflation.
Inflation is extremely convenient for governments because it lets them buy nice things for voters and pay for it using an invisible tax on savings. Because people would be naturally very angry at this taxation if it was done directly, economists and politicans came up with a brilliant theory as to why this is actually a good thing! Zero inflation, says them, would be terribly bad because if peoples savings aren't being constantly stolen at 2% per year or more then they lose all interest in profit. Given the choice of sitting on their money and getting no return, or making a safe 2%-per-year investment, those stupid citizens will choose to sit on it, and that will lead to economic collapse and anarchy and cats living together with dogs, which would never do.
There's a couple of problems with this theory. One is that it is contradicted by actual studies of historical data. The other is that it makes no sense. Capitalism is based on the assumption that people like profit. Normally, if someone is offered a low or even zero risk way to profit, they'll go for it unless they need the money right now. But when it comes to the (non existent, it turns out) "deflationary spiral" we're expected to suddenly stop believing that .... we're told that people will pass up good investment opportunities unless they're forced into it. That flatly contradicts how people actually work. The only kind of investment that becomes interesting solely due to inflation is the bad kind. You know, the kind that tends to be involved with asset bubbles.
By the way, Bitcoin is not intended to be deflationary. You assert that people will lose Bitcoins and sure, a tiny amount might disappear that way. But you'd better expect that easy and automatic backup will become common in future. All the incentives to properly solve that problem are there.
Except right now it inflates at the rate of 25 coins every 10 minutes. So your statement seems to be wrong.
Encryption is not in any way "free". The calculations might be fast but the complexity of the key distribution and management infrastructures are never free.
Secure microkernels in ROM with mandatory enforced security models did go mainstream, actually. Hundreds of millions of people have one in their living room, it's called a games console and they have a pretty enviable track record of being difficult to hack even by extremely competent attackers who have physical access to the machine. The sad thing is that despite Microsoft having a lot of in-house expertise in how to build these kinds of system, they deployed exactly none of it to build secure business workstations, instead that's been left to Google with ChromeOS which (you guessed it) is made up of layers of sandboxes with a secure ROM that integrity checks the next level up before running it. If you disable extensions, which I think admins can do remotely for managed domains, it's proven extremely difficult to penetrate.
Posting even though it'll undo my mods. Just had to point out that the linked story is about a kill of a person. Drones are much larger than people and follow much more predictable paths. Presumably if you can make the bullet go far enough, it might be possible with luck and on a still day to get a bullet into the machine somewhere, and cause some damage. Also drones can't easily run away or perhaps even notice if somebody is shooting at them, so you'd get more attempts than with a person.