Bitcoin took a bit over 2 years to develop before it was released, Satoshi started some time in 2007. He has said that much more of the work was design than coding.
For comparison, I'm around 1.5 years into re-implementing Bitcoin in "simplified payment verification" mode and still far from complete, based on around one day a weeks worth of programming (in practice it's been a bit more). So my belief is that Satoshi was remarkably efficient, remarkably dedicated or both.
Judging from the type of code you find inside, sometimes I wonder if he may have been retired or semi-retired (consulting work, etc). Bitcoin is very clearly the work of one man, yet the amount of effort taken to build it is quite extreme, and the amount of forward planning involved (script) is astonishing.
Topic: Initial Bitcoin client development cost (Read 1449 times)
The first github commit (25 kLOC) would suggest that it took one person 70 months to develop, or 6 years, using the simple COMOCO model. This is entirely possible for one person to do, provided they are in some stage of their life where they can work on a project like this for an extended period.
The characterization that the client is missing security exploits is not true, nor does it remain true. I don't know why this security researcher would make this representation, but in my experience the statements of computer security people deserve scrutiny.
For example, an early significant exploit in the code allowed an attacker to generate 184 billion bitcoins.
It's clearly not perfect, but its better than anything that existed before it -- I'm sure we'll all agree to that.
The characterization that the client is missing security exploits is not true, nor does it remain true. I don't know why this security researcher would make this representation, but in my experience the statements of computer security people deserve scrutiny.
For example, an early significant exploit in the code allowed an attacker to generate 184 billion bitcoins.
It's clearly not perfect, but its better than anything that existed before it -- I'm sure we'll all agree to that.
Sorry, I thought everyone would have already watched his presentation by now and understand what "entire classes of bugs missing" means. It's not that the code doesn't (or can't ever) have exploits; rather, it's that the code looks like lines have been removed/replaced which may initially have caused exploits, suggesting that there had already been a methodical and rigorous security audit.
Here's a vid of his presentation (the ascii art he sneaks into the chain is pretty cool-- if you're running the client on your machine then its on your hd right now):
http://www.youtube.com/watch?v=bifnDbPdCTI
Anyway I just emailed him directly with my query which is what I should have done in the first place
The first github commit (25 kLOC) would suggest that it took one person 70 months to develop, or 6 years, using the simple COMOCO model. This is entirely possible for one person to do, provided they are in some stage of their life where they can work on a project like this for an extended period.
The characterization that the client is missing security exploits is not true, nor does it remain true. I don't know why this security researcher would make this representation, but in my experience the statements of computer security people deserve scrutiny.
For example, an early significant exploit in the code allowed an attacker to generate 184 billion bitcoins.
It's clearly not perfect, but its better than anything that existed before it -- I'm sure we'll all agree to that.
The characterization that the client is missing security exploits is not true, nor does it remain true. I don't know why this security researcher would make this representation, but in my experience the statements of computer security people deserve scrutiny.
For example, an early significant exploit in the code allowed an attacker to generate 184 billion bitcoins.
It's clearly not perfect, but its better than anything that existed before it -- I'm sure we'll all agree to that.
I'm just interested in ballpark-- would the total cost be within reach of a few dedicated financial experts/cryptographers, or are we talking nation state level funding?
Maybe you should put your tin foil hat off --
Look at other open source software. Frequently you find codebases which would cost several Million $$ development cost in a commercial setup. And obviously, the corresponding commercial environment would never be able to earn that money. This doesn't mean that these open source software must have been finnaced hiddenly by [enter whatever you like here]. It just means that there is a whole world beyond standard commerce and ecconomics.
Please read the OP more carefully. I'm asking about the cost of developing the initial client code which was not known to the public until Satoshi made his first post about it.
I'm just interested in ballpark-- would the total cost be within reach of a few dedicated financial experts/cryptographers, or are we talking nation state level funding?
Maybe you should put your tin foil hat off --
Look at other open source software. Frequently you find codebases which would cost several Million $$ development cost in a commercial setup. And obviously, the corresponding commercial environment would never be able to earn that money. This doesn't mean that these open source software must have been finnaced hiddenly by [enter whatever you like here]. It just means that there is a whole world beyond standard commerce and ecconomics.
I'm thinking not only about the cost of developing the idea and writing the initial code but also Dan Kaminsky's statement that there are entire classes of bugs missing from it.
Well, you can do things like run sloccount on it— which actually gives reasonable figures for 'typical' software... though bitcoin isn't especially typical, there is a whole novel distributed algorithm at its center, its own network protocol, and as you mention more than typical care was put in to making it secure.
Feeding the earliest commit in git (easier than finding the first release zip file) to SLOCCount gives me
Total Estimated Cost to Develop = $ 464,573
Thanks, but as you say, this isn't typical software. Even if we attribute the novel use of cryptography to a programmer's epiphany we're still left with the security audits.
I'm not sure where to look to compare-- I guess Tor's annual expenses which are a few hundred thousand (just glanced at the 2007 financial report), but there's a lot more than just security audits going on there and it's an open process. The work done on the original Bitcoin client wasn't done in the open, so it seems like they would have had to spend a considerable amount paying some person(s) to do that work.
I'm just interested in ballpark-- would the total cost be within reach of a few dedicated financial experts/cryptographers, or are we talking nation state level funding?
One thing is for sure, they made a great software. I was reading about a week before about bitcoin, cryptography and I was thinking this sounds complicated. When I finally got courage to try it, I installed the client and I was surprised how simple and self-explanatory it was.
Estimates don't make any sense for this kind of software, since it had something really distruptive & innovative inside it.
However now everyone could make their own bitcoin fork easily, since the algorithms and the idea is now public knowledge. So you could have bitcoin fork really cheaply, since it is open source.
However now everyone could make their own bitcoin fork easily, since the algorithms and the idea is now public knowledge. So you could have bitcoin fork really cheaply, since it is open source.
I'm thinking not only about the cost of developing the idea and writing the initial code but also Dan Kaminsky's statement that there are entire classes of bugs missing from it.
Well, you can do things like run sloccount on it— which actually gives reasonable figures for 'typical' software... though bitcoin isn't especially typical, there is a whole novel distributed algorithm at its center, its own network protocol, and as you mention more than typical care was put in to making it secure.
Feeding the earliest commit in git (easier than finding the first release zip file) to SLOCCount gives me
Total Estimated Cost to Develop = $ 464,573
Do you mean how much was paid to have it developed? Or how much would something similar cost to develop if done with a normal for-hire team?
The latter.
Do you mean how much was paid to have it developed? Or how much would something similar cost to develop if done with a normal for-hire team?
Much of the development work has obviously been pro-bono, and I hope the devs have acquired lots of coin so they all get rich as they appreciate
Much of the development work has obviously been pro-bono, and I hope the devs have acquired lots of coin so they all get rich as they appreciate
Hello,
Does anyone have a link to any research done on the approximate cost of developing something like the initial Bitcoin client? (Don't remember what the version # is but I'm referring to one Satoshi initially made available after the first public posting about the technology.)
I'm thinking not only about the cost of developing the idea and writing the initial code but also Dan Kaminsky's statement that there are entire classes of bugs missing from it.
Thanks
Does anyone have a link to any research done on the approximate cost of developing something like the initial Bitcoin client? (Don't remember what the version # is but I'm referring to one Satoshi initially made available after the first public posting about the technology.)
I'm thinking not only about the cost of developing the idea and writing the initial code but also Dan Kaminsky's statement that there are entire classes of bugs missing from it.
Thanks
Jump to: