Bitcoin Forum

What is up with the bitcoin hashrate jumping between 125MM GH/s and 160MM GH/s???  Does anyone know what causes this?

Not to be too simplistic but the short answer is variance.

Just because the hashrate appears to be jumping around doesn't mean that it really is. Overall bad luck on the network can make it appear that the hashrate has dropped. And the inverse is true for good luck.

Further, network hashrate is just an estimate based on how quickly blocks are found. If a block isn't found every ten minutes, then the hashrate may appear to be dropping. And if blocks are found quickly, then the hashrate will appear to be rising.

I'm sure someone has a more eloquent and accurate explanation. I'm certainly no expert on the technical aspects of BTC.

You did a great job of explaining it.  Obviously there is new equipment added and old equipment removed, which has an effect on the total network hashing rate, but the swings of 40PH/s aren't caused by this.  Nobody, contrary to what the conspiracy theorists would have you believe, decides to turn on a bunch of hardware to then just turn it back off.

There is a very key statement that needs to be understood: the bitcoin protocol has no idea whatsoever the hash rate of the network.  It calculates an approximate hash rate based upon target difficulty and solved block time.  The formula is:
Since we know the difficulty and the time, we can solve for the hashes per second.  As you can see, as time goes up and down, hash rate must also go up and down since the difficulty remains constant, at least until the adjustment at the 2016th block when the difficulty adjusts itself to make time equal to 600 seconds.

Mining Pools on the other hand typically calculate their hash rate based on the number of shares turned in.  For example: https://www.btcguild.com/index.php?page=pool_stats and http://eligius.st/~gateway/stats/recent-blocks

Interestingly both of those pools have had an increase of their hash rate, and a drop in pool luck over the same time frame.

<TinFoilHatOn: ConspiracyTheory>

What if a major pool, or private miner (Ghash, CoinTerra, etc), mined with half of their hardware on a public pool but executed a withholding attack on the pool?  They would collect BTC rewards for the other blocks found by the public pool, but withhold all blocks found and transfer thier BTC rewards to their own pool.  This would make the public pool's luck look bad, and their own pool luck look good.  

They could even delay the block release until a time when their own pool luck was bad, or the public pool's luck was good, to prevent it from looking too obvious.  This delay would cause the entire combined network hash speed to appear much lower than it actually is, at least until they release the found blocks.  They would only need to release them before the difficulty increased when the blocks might become invalid.

</TinFoilHatOn: ConspiracyTheory>


Disclaimer: I have done no research beyond the links listed above, and this is pure speculation.

Not sure to be honest, but it has been jumping up and down like mad over the past few weeks. Difficulty predilections have been all over the place too, was at 7% the other day, at 5% today.  Not complaining though  ;D 

If you are talking about the hashrate you get from pool then its variance, but from your program it should be stable....

As the total hashrate increases the estimated rate will have numerically higher swings but percentage wise I think 15% swings +/- are reasonable and that is what we are seeing over the 3 day bands.

Take a look at https://blockchain.info/charts/hash-rate?timespan=30days&showDataPoints=false&daysAverageString=1&show_header=true&scale=0&address=

Why is the recent hash rate so crazy?  Who is adding tons of hardware and then removing it???

The Variance Fairy  ;D

I see a swing between 123 for the lower band and 162 for upper, so average of 140 roughly, 20/140 swing ~ 15% swing in each direction - that's crazy?

Look back at the old graphs in 2010 and you will see cycles where people turned on and off rigs based on the payday and you had even wilder swings (CPU days).

Zoom the scale out and you will see it's nothing new.

I might be a first day attack  >:(

Attack? What attack are your referring to?  It is simply the variance of the network - it will never consistently find a block/10 minute and so the ESTIMATED rate will bounce around. It's always an estimate.

I'm quoting myself here because apparently people who continue to claim crazy theories don't read the few posts above theirs.

The answer is not the boogeyman, mysterious manufacturers powering up new hardware then turning it off again, aliens or anything other than VARIANCE.  Read my answer quoted above.  Then read it again if you are still asking yourself why the hash rate swings like it does.

I agree with above saying it's just variance. Looks pretty insignificant.

and once again op has touched the weakness of BTC.  So here goes.


Software does exist that will not report a block.  Not a rumor not fake it is as real as mining itself.

It was created  by accident and truly hurt BTC Guild before it was discovered. Most likley cost me around .2-.3btc personally as I was underpaid due to the accidental with holding. Along with many other btcguild miners.

The pools have not resolved this issue no solve has been found.  The only solve is solo mine since the only person hurt by with holding is the one that found the block.

Here is the simple math  a big pool with deep pockets that has that software  can attack a small pool and hurt its luck.

Most likely this will kill bit coin in the long run.

  So to the op this could be simple bad luck/variance or this could be a deep pockets pool attacking a weaker small pool.

Your only prevention is to solo mine or don't mine. Or do what I do put mining power in five pools.

While block withholding attacks are indeed real, they don't account for the swings in reported hash rate.  If as you say large pools have it in for the little guys, then you wouldn't see the sizable swings in hash rate of 40PH/s either way.  Yes, there would be some impact, but that impact would be a negative swing in the rate since the larger pool would have dedicated a portion of its own hash rate to perform the attack.

This falls under the category of conspiracy theory.  Did it happen in the past?  It did indeed, on both BTCGuild and Eligius (at least those are the only pools to have documented it having occurred).  It was one particular miner using an incorrectly written custom version of mining software that did not submit shares that would solve blocks. 

Stating that a block withholding attack will be the demise of bitcoin is, while possible, not a very likely scenario in even the most conspiratorial camp.  Remember, the attack has to come from the MINER not the pool in general.  It's not like GHash.io could suddenly take their entire 50PH/s (or whatever they have mining there now) and decide to perform the attack on Slush, for example.  They could potentially take their OWN hardware and alter it in such a way.  However, this brings up the next question: why would they?  Performing such an attack undercuts their own bottom line at the same time it tries to undermine the luck of another pool.  It's just plain stupidity to purposefully execute this kind of attack.  All you gain by it is the *hope* that people mining on the other pool will jump ship.  That's it.

Your last sentence is great advice: spread out your hashes.  Put some on multiple pools.  It should help decrease the variance for you as a miner.

jonnybravo, so if I understand correctly (pardon me, I am not a math wizz), the quoted hash rates are basically reflecting the network "luck" to solve a block?  Would that be correct?

The quoted hashrates are based on calculations of how quickly blocks are solved.  The only person who knows how fast each miner is actually going is the miner who can view the actual MH/GH/TH/PH on their end.  The pools guess your hashrate based on how many shares are submitted.  The network guesses the hashrate based on the difficulty and how often blocks are solved.

If I told you I flipped a dime a number of times and got 5 heads, you would guess I tossed the coin 10 times most likely - similarly the calculations are done "backwards"

We are under attack! The network is under attack!

i doubt it is an attack, more likely just a variance.
Time will tell.

Maybe it's that professor hacking a supercomputer to mine bitcoin again
http://banksworstfear.com/professor-banned-mining-bitcoin/

Please.  That professor's supercomputer would be green with envy sitting next to a KNC Neptune.  The swing is only about 15% from the midline.  That's been the norm since 2009.

so there's plenty of ASIC companies making products; could be any one of em testing their fleet; right? (along with normal variance)

Just where does a company get temporary access to a datacenter capable of testing dozens of PHs of equipment?

It would take weeks to setup, cost a fortune and all just to test a massive pile of equipment before shipping.

What we are seeing with the network hashrate is normal growth which includes home miners and massive, permanent datacenter all adding new equipment. And normal variance swings.

However, when the network was much, much smaller, it may have been possible to notice new waves of equipment being tested. But those days are long gone.

Some hard numbers on variance in hash rates.

If you take the average hash rate over one day you can expect to see 10% swings from one day to the next.  Even swings of plus and minus 20% of the true value are reasonably likely.  For example if the real network hash rate is 200 Petahash it is statistically reasonably for the daily average to read 160 Petahash one day and 240 Petahash the next even if nothing changed.

If you take more samples (blocks) the variance (error) declines.  Over 2016 blocks (difficulty change) 2% errors in variation are possible even if the number of miners never changed.  If the real network growth over 2016 blocks is exactly 2% the calculated difficulty change could be anywhere between -2% and 6%.  

So expect to see negative difficulty changes in the future due to calculation error even if the network is still gradually growing.  

Mathematically there is a 10 minute block average solve time with a 10 minute variance (poisson distribution).  So the standard deviation percentage error in solve time is 100 / sqrt(number blocks).  This gives a standard deviation error of 8.3% for 144 blocks (1 day), 4.5% for 504 blocks, and 2.2% for 2016 blocks (difficulty change).  The variation will be within 2 Standard Deviations of the true value 95% of the time.  

It looks like a 0-day attack with multiple switch bypass, there is no doubt now, the network is under serious attack, take all the coins away from the blockchain and butin trevor or paper wallet else you cannot be sure and sleep bad there is little time. HODL!

You keep rambling on about an attack.  Show me one credible article discussing this attack.

Might not be an techinical expert, but it is an attack for sure! Look at the exhange rate it going down down down... its under attack! HODL!

Expect increases to jump in blocks as each DC comes online. My guess is when the cost of mining on a world average across data centres equals about 85% of the mining profit ( around about a diff of 80⁹ give or take.... ), then these large jumps will be long over.

The coming year will be dominated by the war of the data centres.

+1

Bitcoin will go up

Shit.... The hash rate may break 180MM GH/s. A 40 PH/S swing in a matter of 12 hours is some pretty serious variance. I doubt anyone is attacking the network, but something is going on.

The price is down due to all the news from the US Consumer Protection Agency and all the regulation talk.

I am getting serious worried too  :-\  (but still convinst that it is some kind of new attack, that we do not know of yet, maybe NSA or KGB has build in a trojan or something)

Still goes to show that the majority of traders are emotionally driven amateurs.

NSA?  KGB?

If this is an attack, then the protocol has been under attack more or less for the entirety of its existence.

Brian has obviously let his prescriptions lapse.  ;)

Haha...  Good call!

Aliens attacking our currency  ;D

but yeah there are alot of miners out there turning on and out their hw

It appears the estimated hashrate is coming back to earth. Bad luck last round + Good luck this round = Crazy spike.

Yeah, it kinda missed them the other day + I was looking to deep in the research of CannabisCoin, that is abad combination. I realized that KGB is closed down, so they have a new name, but I think the same people is involved just a new brand.