Lowest Difficulty increase in a while!

[DannyHamilton]

is it possible that there's no lack of added hashing power, and it's just getting exponentially harder to find such low hash numbers?

I'm not sure I understand your question.

That sounds a bit like asking if the reason that heads comes up 50% of the time when flipping a coin is because it's getting exponentially harder for it to land on tails.

It gets harder to find the hash because the target value is lowered.
The target value is lowered because miners are finding 2016 blocks in less than 20160 minutes.
Miners find 2016 blocks in less than 20160 minutes because hashing power has been added since the previous difficulty adjustment.

If hashing power isn't added, then miners don't find 2016 blocks in significantly less than 20160 minutes.
If miners don't find 2016 blocks in significantly less than 20160 minutes then the difficulty isn't adjusted very much.
If the difficulty isn't adjusted very much, then it doesn't get much harder to find low enough hash values.

[jonald_fyookball]

is it possible that there's no lack of added hashing power, and it's just getting exponentially harder to find such low hash numbers?

I'm not sure I understand your question.

That sounds a bit like asking if the reason that heads comes up 50% of the time when flipping a coin is because it's getting exponentially harder for it to land on tails.

It gets harder to find the hash because the target value is lowered.
The target value is lowered because miners are finding 2016 blocks in less than 20160 minutes.
Miners find 2016 blocks in less than 20160 minutes because hashing power has been added since the previous difficulty adjustment.

If hashing power isn't added, then miners don't find 2016 blocks in significantly less than 20160 minutes.
If miners don't find 2016 blocks in significantly less than 20160 minutes then the difficulty isn't adjusted very much.
If the difficulty isn't adjusted very much, then it doesn't get much harder to find low enough hash values.

I understand that the difficulty adjustment is directly proportional to the speed of the previous 2016 blocks.
What I'm asking is:   is the relationship between network hashing power and difficulty linear, or is it
sublinear or logarithmic?  Or is that relationship mathematically unknown because of the nature of
hash functions?

[FlyForFun]

I don't think the low difficulty increase is going to last, a lot of ASIC are in the pipeline..

[Harley997]

is it possible that there's no lack of added hashing power, and it's just getting exponentially harder to find such low hash numbers?

I'm not sure I understand your question.

That sounds a bit like asking if the reason that heads comes up 50% of the time when flipping a coin is because it's getting exponentially harder for it to land on tails.

It gets harder to find the hash because the target value is lowered.
The target value is lowered because miners are finding 2016 blocks in less than 20160 minutes.
Miners find 2016 blocks in less than 20160 minutes because hashing power has been added since the previous difficulty adjustment.

If hashing power isn't added, then miners don't find 2016 blocks in significantly less than 20160 minutes.
If miners don't find 2016 blocks in significantly less than 20160 minutes then the difficulty isn't adjusted very much.
If the difficulty isn't adjusted very much, then it doesn't get much harder to find low enough hash values.

I understand that the difficulty adjustment is directly proportional to the speed of the previous 2016 blocks.
What I'm asking is:   is the relationship between network hashing power and difficulty linear, or is it
sublinear or logarithmic?  Or is that relationship mathematically unknown because of the nature of
hash functions?

Assuming that the network has 100% luck (meaning it does not have overall good, nor bad luck) then the relationship will be linear.

For example if the current network hashrate was 100 GHs at the beginning of a difficulty change, and a miner started a 10 GHs miner as soon as the difficulty changed then the difficulty would increase by 10% after 2016 blocks (the hashrate increased by 10%), assuming that there are no other changes to the network.

[jonald_fyookball]

Assuming that the network has 100% luck (meaning it does not have overall good, nor bad luck) then the relationship will be linear.

For example if the current network hashrate was 100 GHs at the beginning of a difficulty change, and a miner started a 10 GHs miner as soon as the difficulty changed then the difficulty would increase by 10% after 2016 blocks (the hashrate increased by 10%), assuming that there are no other changes to the network.

What I'm hypothesizing is that low block
header hashes may start to become exponentially rare
at some point.  (Or at least superlinearly rare).  Iow,
a huge amount  of hashing power added to the
network will have an effect of diminishing
returns of finding progressively lower hashes.

I don't know if this is true or not, so if anyone
has any mathematical information, even if
the answer is that is definitively mathematically unknown,
would be helpful/interesting.
 

[mwizard]

As has been mentioned previously you are only looking at statistical averages.  There is no accurate way to tell the total network hash rate.

You should assume an error of 5% is likely in difficulty changes (2016 blocks).  For example saying the difficulty increased by 4% means the real network may have actually changed by anywhere between -1% or 9% , or more rarely even further from the given figure.

Displays of network hash rate over 504 blocks are even more doubtful.  There you can assume a plus or minus 10% error.

The most recent 20% fall in difficulty change is rare but similar changes have occurred in the past.  The June 29th difficulty change (25%) was probably above the real network hash rate change, the July 12 change (3%) below the real value.

In summary the long term trend is clear but there is a lot of statistical noise in individual difficulty changes, as in any network hash rate calculations particularly over short periods.  

If you really want some math details, the block solve time averages 10 minutes with a 10 minute standard deviation (Poisson distribution).  So the standard deviation in the error over 2016 blocks is 100/sqrt(2016) or 2.2%.  Most of the time the difficulty change will be within 2 standard deviations which means a plus or minus 5% error in the quoted difficulty.  Over 504 blocks the errors are twice as large, typically plus or minus 10%.

[revilo]

Or more simply. If mining ASICs dont improve their GH/Watt quick enough then they become unprofitable.
There are currently no ASICS that come anywhere near close to giving a solid ROI.

[Syke]

The two best mining manufacturers, Bitmain and Spondoolies, are in between generations. By the end of July they should both be pumping out huge amounts of hashpower. Enjoy a couple small difficulty increases. It's a temporary respite.

[freedomno1]

A mere 3.08% increase in difficulty this time round
Is this the future of things to come?
Or is the increase going to continue to grow exponentially?

Thoughts...

Enjoy it while it lasts the next rise of more ASIC units is incoming
When though is another question but the price will probably go up as well causing the next spurt.

[ALToids]

Or more simply. If mining ASICs dont improve their GH/Watt quick enough then they become unprofitable.
There are currently no ASICS that come anywhere near close to giving a solid ROI.

Electricity is becoming more of a factor but for most miners in cheaper areas the main cost is still the hardware cost up front.  If it costs the manufacturer $200 to design and make an ASIC but they want to sell it for $500 then they will do so as long as they have sucke...err customers.

The question is what hardware manufacturer is willing to sell miners at near minimal profit margins but in bulk?

[CryptoCrane]

I don't think we've seen the last of the 25% increases in hash rate - especially if the price spikes. The previous efficiency gains in Gh/watt were largely due to smaller and smaller lithography. The first ASICs started with really old 110nm tech and the huge spikes in hash rate over the last year have been mostly attributable to manufacturer's catching up to modern 28nm tech. 20nm is bleeding edge and 14nm is reserved for the likes of Intel for the next couple years. The improvement in Gh/watt from 28nm tech to 20nm tech is ~29%. Going from 20nm to 14nm will be another 30% on top of the last efficiency gain.

Given current bitcoin prices at ~$630, electric cost per kWh @ $0.1326 (my rate), and best GH/watt for 28nm = 2Gh (0.5W/Gh) - one can see that 14nm tech will give us about double the efficiency of today's 28nm tech. If every miner in the world were running 14nm @ 0.25w/Gh, the equilibrium point in which electricity costs as much as mined revenue will be when the diff rate = ~370b or 21.3x the current rate (17,336,316,979 x 21.3 = ~370b).

Assuming 14nm tech becomes widely available in 2016, our average increase per diff adjustment would be around 30% between now and then. Of course, I don't expect to reach equilibrium with 14nm gear and therefore this whole exercise is pointless - but it's fun to imagine anyways!

[philipma1957]

The two best mining manufacturers, Bitmain and Spondoolies, are in between generations. By the end of July they should both be pumping out huge amounts of hashpower. Enjoy a couple small difficulty increases. It's a temporary respite.

well yes and no.  the gear that does 2 watts a hash is slowing dropping off the market. soon s-1s will not pay for the power they spend.

if they all turn into 140 gh at 180 watts on lower volts/clock  there will be  more then 20000  x 40 gh drop in hash.


 that is just some of the older gear not hashing or reduced hashing.

So my guess is the 200 day trend will continue its drop.  

june 2013 to dec 2013 200 day diff avg = 24%

jan 2014  to july 2014  200 day diff avg = 18%

july 2014 to feb 2015  200 day diff   avg =  13.5%     that is my best guess for the next 19 jumps or 200 days.


Disclaimer I am rounding a bit, but in the ballpark with the last two   200 day  periods.


 The diff has shown a down trend.  the exact dates are:

  Dec 21 , 2013  to July 12 , 2014       about 18% avg  1,180,923,195  to  17,336,316,979  14x the hash

  June 5 , 2013  to  Dec 21 , 2013      about 24% avg      15,605,633   to    1,180,923,195  75x the hash



that is a huge drop in diff and hash increase.

this will continue baring a new magic chip  that does .01 watt per hash or .02 watt per hash.

second way hash  and diff may do better is  price goes to 6 or 7 k a btc.

[ThomasCrowne]

Enjoy it while you can.  I'm guessing the next difficulty increase will be somewhere in the neighborhood of 7-8% and then it gets worse from there.

[philipma1957]

Enjoy it while you can.  I'm guessing the next difficulty increase will be somewhere in the neighborhood of 7-8% and then it gets worse from there.

I figure the next 200 days are about 13-15% avg.

Now that does come up with interesting numbers.

At 13%   200 days from now we are at  jan 30th 2015  diff would be 108g  or  6.25x the diff now and  the hash power would be

 about 750ph
_______________________________________________________________________________ _________________________
At 15%  200 days from now we are at jan 30th 2015 diff would be 141g or  8.15x the diff now and the hash power would be about

978ph

  at 15%  diff your  gear  does not pay for power if it burns .8 watts a hash at 10 cents a kwatt with btc at 623.


Now the s-1 took about dec 2013 to july 2014 to drop from 4,000+ to 150+  in usd  and not worth running for a lot of miners.




If you have an sp30 and it gets .4 a watt at the best watt to hash it stops earning at day 271.

 but it is due to mine in 60 days while the s3 is due to mine it 3 days.

    I don't see avg growth over 15 %.   but I am just guessing here.