6 ms·

Bitcoin Block Time Halved To Five Minutes Amid Exponential Network Growth

63 points by CrunchyJams 13y ago

applecore 13y ago
Given the exponential growth in the network hash rate, how much sooner than 2140 can we expect the last bitcoins to be mined?
- ihsw 13y ago
  Once equilibrium is reached, we can expect hashing rates to bottom out. A similar rush will occur over time as the cadence of Bitcoin hashing technologies approaches more regularity, not unlike how new generations of CPUs and GPUs are released on a timely and predictable basis. This will be the pattern of Bitcoin hashing -- waxing and waning with each new generation of ASICs and the lull between each release. One has to wonder what will happen to all the surplus ASIC hardware as well-financed miners regularly upgrade their hashing infrastructure.
  - mithras 13y ago
    Perhaps sold for very little money to people who have free power. Soon it will be about power efficiency more than raw hashing power like what happened during the switch form CPU to GPU mining.
- CrunchyJams 13y ago
  It's not an exact science since no one knows future network increases, but you can see a detailed view of the thought process around that question here: http://thegenesisblock.com/at-this-rate-the-last-new-btc-will-be-issued-55-years-ahead-of-schedule/ http://thegenesisblock.com/at-this-rate-the-last-new-btc-wil...
  - sneak 13y ago
    This rate of growth is temporary.
- ufmace 13y ago
  Short answer - it isn't likely to change much. In order to significantly affect the overall bitcoin issuing rate long-term, the hash rate would have to continue to grow exponentially for an extended amount of time - years. In reality, the hash rate can only grow exponentially while an older mining technology is being supplanted by a newer one. It happened already with CPU to GPU/FPGA, and now it's happening again with GPU/FPGA to ASIC. It will grow exponentially while everyone hashing upgrades to ASICs, but once the hashing market is saturated with them, it will drop back to modest linear growth, as the only growth will be either new miners coming online or modest, linear improvements in ASIC technology. Haven't looked at this part, but I have a feeling that the difficulty itself wouldn't be able to go high enough to support long-term exponential growth in hashrate.
PaperclipTaken 13y ago
"A common misconception is that transactional confidence is based on time" Transactional confidence is based on the probability of someone compromising the network for enough blocks to undo a spend that you've already honored (IE you gave them what they paid for, then they undo the transaction). At 6 blocks, you are assuming that an attacker won't have the computational power to reverse 6 blocks in a row. When blocks come out every 2 hours, that means you need to find every single block for 12 hours, or essentially control the blockchain for a long time. When the block rate is down at 5 minutes, you only need to control the network for 30 minutes - you have more chances to get lucky and undo 6 transactions in a row in your 12 hours of purchased computing time. A halved block rate does not mean you need less computational power, but it does mean that computational power is probably cheaper and 6 blocks worth of power is going to be more affordable.
- makomk 13y ago
  Unless your purchased compute time is more than 50% of the network your odds depend mostly on the number of confirmations and not the time they take. The cost of a sub-50% attack increases exponentially with more confirmations but only linearly with longer confirmations, so the number of confirmations is more important.
- MacsHeadroom 13y ago
  A successful attack requires more than simply reversing any 6 consecutive blocks. It requires reversing a specific set of blocks containing a transaction one wishes to be reversed. This is why 5 minute blocks are virtually every bit as secure as 10 minute blocks.
- muyuu 13y ago
  5 minutes is still a perfectly acceptable block time. It only gets problematic as network latency can be significant in block discovery chances. This happens under a few seconds per block. There are alt-coins working with 1 minute blocks and they behave just fine. Provided hash power is spread among the network, there is absolutely no problem with this, and it won't be long until we go back to 10 minute-ish blocks.
anologwintermut 13y ago
Inaccurate title. The network still sets the difficulty every 2016 blocks ( i.e. 2 weeks) so that it takes 10 minutes on average to create a block. Right now, however, people are adding so much computational power in that intervening two weeks, that by the end of those two weeks it takes 6 minutes to create a block, not 10. This is presumably due to custom hardware mining rigs coming online since they just shipped.
- cgi_man 13y ago
  The title was accurate. Blocks are being found at a rate of 1 every 5 minutes. There's no way the next difficulty jump will be enough to bring the speed to 10 minutes since it takes the average time over the last 2016 blocks. It's likely in a couple weeks the time to find blocks will be even lower than it is now if the growth rate continues like this.
  - nwh 13y ago
    There's also a ceiling on the percentage increase and decrease possible.
- sneak 13y ago
  It only averages to 10 minutes if growth is flat. Growth isn't flat. This means that lots of ASICs are out there in the wild. This is a good thing for bitcoin and the bitcoin network.
  - beefxq 13y ago
    The only thing this mean is that the ASIC companies are making a killing off of this.
mikevm 13y ago
If only people were this determined to run things like Folding@Home....
- lightbritefight 13y ago
  So, introduce a coin variant where 1/1000 or 1/10000 of cycles is donated to folding@home or seti@home or just a random Boinc project. That way, you still make money, but now you get to donate your cycles to charity as well.
  - haakon 13y ago
    It's not easy to do something like that without sacrificing decentralization, which is a cryptocurrency holy cow. It would easily become a competitive advantage to remove the charity cycles from the mining client.
    - betterunix 13y ago
      "It's not easy to do something like that without sacrificing decentralization, which is a cryptocurrency holy cow." Only among the Bitcoin crowd. In the crypto research community people are fine with having a bank issue the money and have published a mountain of research on how to create secure digital cash.
      - gojomo 13y ago
        The people who are fine with centralized bank-issued cryptocurrencies haven't launched anything successful at the scale of Bitcoin. There may yet be hope for one, especially if it has government backing... but right now "the Bitcoin crowd" has a good existence-based case that decentralized-issue is important for adoption.
      - wmf 13y ago
        And all those systems were never deployed. Perception (f the banks, you can print money at home, and the money will grow in value over time) trumps research apparently.
batgaijin 13y ago
If only the bitcoin algorithm could have accounted for incredibly various technical and sociological issues. Good thing Moore's Law somehow exists and has maintained the curve perfectly despite not being a law, despite somehow not making any reference to the consumer activity that seemingly drives it, and a heavily innovation-constraining duopoly for every chip that is used in desktops/laptops. Some times I feel like the computer industry really is driven by magic smoke.
- EthanHeilman 13y ago
  >If only the bitcoin algorithm could have accounted for incredibly various technical and sociological issues. I think they did a pretty decent job given that "incredibly various technical and sociological issues" is the definition of a wicked problem (http://en.wikipedia.org/wiki/Wicked_problem http://en.wikipedia.org/wiki/Wicked_problem).
gojomo 13y ago
Some alt-coins have been destroyed by the in-rush of massive hashing power, which caused a difficulty-adjustment up, followed by an exit of that hashing power, leaving block-progress so slow that it would take 'forever' (impractically long) for the next normal compensating adjustment-down to happen. Since one of the joys of Bitcoin is using it to exercise one's paranoid imagination, consider scenarios where somehow, an enemy of Bitcoin (or promoter of an alternative) can somehow disable or redirect all that ASIC power in a short time frame. Perhaps, in cahoots with the ASIC makers, some magic input can cause them to begin malfunctioning (either temporarily or permanently). Or, a cartel controls enough SHA256 power to withhold capacity, and switch to a NewCoin needing the same hashing resources, in rapid coordinated succession. Or both. It might be hard for Bitcoin-prime to adapt in time, while NewCoin's deep-pocketed backers from the USG, CCP, or evil lectroids of Planet-X tempt miners and Bitcoin balance holders to switch to NewCoin, perhaps via an initial endowment mechanism which requires the irreversible destruction/relinquishment of old Bitcoin balances. Ah, good times. Good Stephensonian-science-fictional times.

63 points by CrunchyJams 13y ago

Given the exponential growth in the network hash rate, how much sooner than 2140 can we expect the last bitcoins to be mined?

ihsw 13y ago
Once equilibrium is reached, we can expect hashing rates to bottom out. A similar rush will occur over time as the cadence of Bitcoin hashing technologies approaches more regularity, not unlike how new generations of CPUs and GPUs are released on a timely and predictable basis. This will be the pattern of Bitcoin hashing -- waxing and waning with each new generation of ASICs and the lull between each release. One has to wonder what will happen to all the surplus ASIC hardware as well-financed miners regularly upgrade their hashing infrastructure.
- mithras 13y ago
  Perhaps sold for very little money to people who have free power. Soon it will be about power efficiency more than raw hashing power like what happened during the switch form CPU to GPU mining.
CrunchyJams 13y ago
It's not an exact science since no one knows future network increases, but you can see a detailed view of the thought process around that question here: http://thegenesisblock.com/at-this-rate-the-last-new-btc-will-be-issued-55-years-ahead-of-schedule/ http://thegenesisblock.com/at-this-rate-the-last-new-btc-wil...
- sneak 13y ago
  This rate of growth is temporary.
ufmace 13y ago
Short answer - it isn't likely to change much. In order to significantly affect the overall bitcoin issuing rate long-term, the hash rate would have to continue to grow exponentially for an extended amount of time - years. In reality, the hash rate can only grow exponentially while an older mining technology is being supplanted by a newer one. It happened already with CPU to GPU/FPGA, and now it's happening again with GPU/FPGA to ASIC. It will grow exponentially while everyone hashing upgrades to ASICs, but once the hashing market is saturated with them, it will drop back to modest linear growth, as the only growth will be either new miners coming online or modest, linear improvements in ASIC technology. Haven't looked at this part, but I have a feeling that the difficulty itself wouldn't be able to go high enough to support long-term exponential growth in hashrate.

PaperclipTaken 13y ago

"A common misconception is that transactional confidence is based on time" Transactional confidence is based on the probability of someone compromising the network for enough blocks to undo a spend that you've already honored (IE you gave them what they paid for, then they undo the transaction). At 6 blocks, you are assuming that an attacker won't have the computational power to reverse 6 blocks in a row. When blocks come out every 2 hours, that means you need to find every single block for 12 hours, or essentially control the blockchain for a long time. When the block rate is down at 5 minutes, you only need to control the network for 30 minutes - you have more chances to get lucky and undo 6 transactions in a row in your 12 hours of purchased computing time. A halved block rate does not mean you need less computational power, but it does mean that computational power is probably cheaper and 6 blocks worth of power is going to be more affordable.

makomk 13y ago
Unless your purchased compute time is more than 50% of the network your odds depend mostly on the number of confirmations and not the time they take. The cost of a sub-50% attack increases exponentially with more confirmations but only linearly with longer confirmations, so the number of confirmations is more important.
MacsHeadroom 13y ago
A successful attack requires more than simply reversing any 6 consecutive blocks. It requires reversing a specific set of blocks containing a transaction one wishes to be reversed. This is why 5 minute blocks are virtually every bit as secure as 10 minute blocks.
muyuu 13y ago
5 minutes is still a perfectly acceptable block time. It only gets problematic as network latency can be significant in block discovery chances. This happens under a few seconds per block. There are alt-coins working with 1 minute blocks and they behave just fine. Provided hash power is spread among the network, there is absolutely no problem with this, and it won't be long until we go back to 10 minute-ish blocks.

anologwintermut 13y ago

Inaccurate title. The network still sets the difficulty every 2016 blocks ( i.e. 2 weeks) so that it takes 10 minutes on average to create a block. Right now, however, people are adding so much computational power in that intervening two weeks, that by the end of those two weeks it takes 6 minutes to create a block, not 10. This is presumably due to custom hardware mining rigs coming online since they just shipped.

cgi_man 13y ago
The title was accurate. Blocks are being found at a rate of 1 every 5 minutes. There's no way the next difficulty jump will be enough to bring the speed to 10 minutes since it takes the average time over the last 2016 blocks. It's likely in a couple weeks the time to find blocks will be even lower than it is now if the growth rate continues like this.
- nwh 13y ago
  There's also a ceiling on the percentage increase and decrease possible.
sneak 13y ago
It only averages to 10 minutes if growth is flat. Growth isn't flat. This means that lots of ASICs are out there in the wild. This is a good thing for bitcoin and the bitcoin network.
- beefxq 13y ago
  The only thing this mean is that the ASIC companies are making a killing off of this.

mikevm 13y ago

If only people were this determined to run things like Folding@Home....

lightbritefight 13y ago
So, introduce a coin variant where 1/1000 or 1/10000 of cycles is donated to folding@home or seti@home or just a random Boinc project. That way, you still make money, but now you get to donate your cycles to charity as well.
- haakon 13y ago
  It's not easy to do something like that without sacrificing decentralization, which is a cryptocurrency holy cow. It would easily become a competitive advantage to remove the charity cycles from the mining client.
  - betterunix 13y ago
    "It's not easy to do something like that without sacrificing decentralization, which is a cryptocurrency holy cow." Only among the Bitcoin crowd. In the crypto research community people are fine with having a bank issue the money and have published a mountain of research on how to create secure digital cash.
    - gojomo 13y ago
      The people who are fine with centralized bank-issued cryptocurrencies haven't launched anything successful at the scale of Bitcoin. There may yet be hope for one, especially if it has government backing... but right now "the Bitcoin crowd" has a good existence-based case that decentralized-issue is important for adoption.
    - wmf 13y ago
      And all those systems were never deployed. Perception (f the banks, you can print money at home, and the money will grow in value over time) trumps research apparently.

batgaijin 13y ago

If only the bitcoin algorithm could have accounted for incredibly various technical and sociological issues. Good thing Moore's Law somehow exists and has maintained the curve perfectly despite not being a law, despite somehow not making any reference to the consumer activity that seemingly drives it, and a heavily innovation-constraining duopoly for every chip that is used in desktops/laptops. Some times I feel like the computer industry really is driven by magic smoke.

EthanHeilman 13y ago
>If only the bitcoin algorithm could have accounted for incredibly various technical and sociological issues. I think they did a pretty decent job given that "incredibly various technical and sociological issues" is the definition of a wicked problem (http://en.wikipedia.org/wiki/Wicked_problem http://en.wikipedia.org/wiki/Wicked_problem).

gojomo 13y ago

Some alt-coins have been destroyed by the in-rush of massive hashing power, which caused a difficulty-adjustment up, followed by an exit of that hashing power, leaving block-progress so slow that it would take 'forever' (impractically long) for the next normal compensating adjustment-down to happen. Since one of the joys of Bitcoin is using it to exercise one's paranoid imagination, consider scenarios where somehow, an enemy of Bitcoin (or promoter of an alternative) can somehow disable or redirect all that ASIC power in a short time frame. Perhaps, in cahoots with the ASIC makers, some magic input can cause them to begin malfunctioning (either temporarily or permanently). Or, a cartel controls enough SHA256 power to withhold capacity, and switch to a NewCoin needing the same hashing resources, in rapid coordinated succession. Or both. It might be hard for Bitcoin-prime to adapt in time, while NewCoin's deep-pocketed backers from the USG, CCP, or evil lectroids of Planet-X tempt miners and Bitcoin balance holders to switch to NewCoin, perhaps via an initial endowment mechanism which requires the irreversible destruction/relinquishment of old Bitcoin balances. Ah, good times. Good Stephensonian-science-fictional times.