Quote from: Carlton Banks on September 27, 2019, 09:47:06 AM

Quote from: pereira4 on September 27, 2019, 01:23:48 AM

Quote from: Carlton Banks on September 25, 2019, 03:12:29 PM

Quote from: camaund on September 25, 2019, 01:57:20 PM

Looks like hashrate is back up.

there is no evidence the hashrate ever fell

it cannot go "back up" if it did not even fall

What is the process like in order to manipulate every single hashrate chart out there without actually doing anything hashrate wise? seems a bit extreme to me.

it's not the graphs that were manipulated, it's the interpretation of them

The hashrate graph are averaged predictions

They are not "the" hashrate, there is no graph of "the" hashrate.

To plot these hashrate graphs we see for Bitcoin, you look backwards in time, at x number of blocks and the time at which those x blocks were announced to the Bitcoin network.

The obvious point is, that if you choose a small number for x as your window in which to calculate the average, you get a very "accurate" (but very erratic) graph line. And so people who publish hashrate graphs usually use 3 windows; say, average over 1 day, over 3 days, and over 7 days, and put all three average lines on the same graph in different colors (I hope this sounds familiar....)

The manipulation worked like this: a long gap between 2 blocks happened on September 23rd, 1 hour and 16 minutes I believe (much longer than the 10 minute average gap). If you calculate the average hashrate over the 6 hours that contained that 1 hour-long block interval, you'll get that very erratic measurement that implies a big drop in hashrate. But if you calculated it over 12 or 24 hours, still including that very long block interval, it's barely a blip on the charts, because the long block doesn't distort the average as much when you use more data points to calculate the average.

Conversely, if you calculate the average using just 2 hours, you could say the hashrate had dropped 90% or more, and technically it's not wrong!! But if you choose a short window, your average is increasingly meaningless. Averages are all about giving disparate data points a context, there's almost no context if you're using only 2 data points!!

I mean, fuck it, if you use only 1 data point to calculate the average change, the size of the series change is 0!!! If the window you're using moves zero data points ahead, then you can calculate infinity % change in hashrate, regardless of how much it went up or down!! It's not mathematically incorrect, but it is dumb as fuck.

I know those charts do not read realtime hashrate changes and do averages. However, what are the chances that, with absolutely no negative change in hashrate, we deviated as much as 1+ hour-long block interval away from the average of 10 minutes? Wouldn't it be more likely that there was indeed a temporarily decrease in hashrate which then caused this big gap? Probabilities aside.