Return of Mystery Miner?

[marcus_of_augustus]

What the hell is going on with network hashrate?!

http://www.bitcoinwatch.com/

Climbed madly to peak at 4.4 Thash ... and now crashed down to 3.2 Thash ... outside statistical variability. Someone or something had a big push and now has backed off ... first attack? Someone trialling FPGAs?

Anyone got the skinny?

[PabloW]

Just noticed this too, whats going on?

[allinvain]

What the hell is going on with network hashrate?!

http://www.bitcoinwatch.com/

Climbed madly to peak at 4.4 Thash ... and now crashed down to 3.2 Thash ... outside statistical variability. Someone or something had a big push and now has backed off ... first attack? Someone trialling FPGAs?

Anyone got the skinny?

That I think is related to the flawed way in which the hash rate is calculated after a difficulty increase. In other words the problem is with bitcoinwatch.com

[Jaime Frontero]

8 blocks generated in half an hour...

[pwnyboy]

8 blocks generated in half an hour...

At 8pm EST there were 14 blocks generated that hour, between 8pm-9pm.  That'd easily account for a large spike, as the average at this moment according to bitcoincharts.com is 6 per hour.  Not sure how much of a window, time-wise, that they use to make this calculation.

Closely related, I'm seeing wild variations in the amount of time it takes, within each hour, to find blocks (whereas in times-past the finding of blocks seemed to happen at much more regular and predictable intervals at the lower difficulties).  Earlier in the day there were hours when only two or three were found.  A far cry from 14.  If anyone cares for the raw data i'll include it below, all times EST beginning at 8pm.  Also is there some site that lists all, or a large set of the blocks, historically?  I'm only seeing the last few on blockexplorer.com.

---

20:07
20:10
20:17
20:17
20:20
20:24
20:43
20:48
20:49
20:49
20:52
20:53
20:57
20:58
21:00
21:10
21:21
21:22
21:47
21:57
22:01
22:36
22:38
22:48
23:03
23:07
23:09
23:14
23:57
0:28
0:29
0:41
0:47
1:04
1:12
1:13
1:30
1:31
1:36
2:08
2:26
2:28
2:31
2:33
2:34
2:47
2:48

[marcus_of_augustus]

What the hell is going on with network hashrate?!

http://www.bitcoinwatch.com/

Climbed madly to peak at 4.4 Thash ... and now crashed down to 3.2 Thash ... outside statistical variability. Someone or something had a big push and now has backed off ... first attack? Someone trialling FPGAs?

Anyone got the skinny?

That I think is related to the flawed way in which the hash rate is calculated after a difficulty increase. In other words the problem is with bitcoinwatch.com

So just artifact of the averaging ... how boring.

Edit: Would be good if we got some kind of measure for how extreme the variance of the current block generation rate is ...

[KnuttyD]

Oops, sorry guys!
Accidently started up a few my ASIC based miners....
Ill try and keep em below 1TH/s from now on 

[marcus_of_augustus]

looks like more than just averaging blip ...
http://bitcoin.sipa.be/

[mathx]

So just artifact of the averaging ... how boring.

Edit: Would be good if we got some kind of measure for how extreme the variance of the current block generation rate is ...

This is actually troubling. Law of averages say that random wild excursions from the mean without any other input (like someone turning off the NSA's rig ) should be very rare, and the larger the excursion, the more rare.

Im sure someone is doing some stats analysis to see if this the fluctuations are expected or a very rare anomaly - if they're happenign too frequently, then either ther'es very large flux in computation rate for a reason (NSA, or a major blackout somewhere, or internet outrage), or someone's not aggregating hashrates properly.

as you say, tracking the variance and assigning a statistical value to its probability would be useful. cept i failed stats class.

[Chucksta]

ALIENS !! it be aliens 

[Vladimir]

By looking at sipa's charts I can say that if you disregard anything but the purple and red lines than there is no mystery. Sensitive hearts of some OCN script kiddies could not take difficulty change and their GPU's dropped off.

[marcus_of_augustus]

By looking at sipa's charts I can say that if you disregard anything but the purple and red lines than there is no mystery. Sensitive hearts of some OCN script kiddies could not take difficulty change and their GPU's dropped off.

Valdimir;
I wouldn't be so sure about that because the portion of the power that has gone off-line was largely in the section "Other" on the bitcoinwatch page from what I noticed (icbw) .... I think the OCNers would have been in with deepbit or one of the other pools, can't imagine they are going solo or private pooling ... or?

[Vladimir]

Good point.

Than I suppose from time to time when employers see electricity bills some rogue IT people get fired.

But note how decline of hash power coincides with difficulty change. It was almost the same before the last time we had drop in difficulty.

[marcus_of_augustus]

Good point.

Than I suppose from time to time when employers see electricity bills some rogue IT people get fired.

Ha! Could be ... there was some interview with IT guy at some school ... can't believe how cavalier he was about using school's power and hardware to belt out some BTC on CPU's ... yeah it is profitable on CPU's if you can steal the power and hardware, eh?
http://www.motherboard.tv/2011/5/27/a-bitcoin-lesson-from-a-system-administrator-who-s-growing-them-on-his-school-s-computers

But note how decline of hash power coincides with difficulty change. It was almost the same before the last time we had drop in difficulty.

yeah, it has the same feel as when last time difficulty dropped .... if it keeps going down I might schedule some maintenance down-time to get things back to scratch ...

[Vladimir]

Ohh you mean this one...

You work at a school, and your job is IT Director essentially?

Yeah, so I have access to the root account of a lot of the computers in a lot of the labs. I figure this summer, when the computers aren’t being used, I’d basically have my own little server farm.

Now explain what you are doing with the computers and Bitcoins?

On the computers there’s a Daemon that runs at startup that basically calls a specialized script; each of the computers has its unique username and password, logs into a server, gets a chunk of mathematical work to do, and reports back when it has a solution.

And how many computers do you have running it?

Right now I have it on about 15. I haven’t decided yet whether or not to to scale it up to the maximum I could do, which is about 60.

Is that illegal?

Well, Bitcoins are legal. Using the computers to do this? I am pretty sure that’s a grey area. I’d be hard-pressed to tell you which law I was breaking. I am pretty sure my employers wouldn’t be thrilled if they found it though.

Do you think they would even understand what was going on?

Most likely not.

I am sure that school's legal counsel will be less hard-pressed to find a few ways to press criminal charges on felony level (in US). Ignorance is rather bad defence.

[Sukrim]

15 school computers accounting for 1.5 TH/s?

We can be lucky he didn't vring all 60 online then... 

[pwnyboy]

I again saw 12 blocks generated in an hour yesterday, during a time when Deepbit was offline.  This is way outside the normal statistical variance, given the loss of hashing power.  It doesn't just rain packets out of the sky; it's pretty clear to me that the two are directly interconnected.

[ffe]

See

I bet someone controlling even 1% of capacity could, through the amplification of other miners joining him once they notice, cause hard/easy cycles where production in the easy cycle is more than twice desired production.

[Sjalq]

Is the fear security? I'm not following why folks are concerned?

[rezin777]

See

I bet someone controlling even 1% of capacity could, through the amplification of other miners joining him once they notice, cause hard/easy cycles where production in the easy cycle is more than twice desired production.

The fluctuations are occurring inside difficulty periods, not coinciding with difficulty changes.