There was virtually no loss of miners or hashpower after the July 3rd change in our PoW algorithm to Equihash-BTG (also known as Equihash<144,5> or Zhash.)
Many people are confused by the reduction in hashrate, which was a natural result of the change in PoW algorithm.
On GPUs mining Equihash, output is 12x faster when doing <200,9> (previous algo) compared to <144,5> (current algo). This means that the same amount of mining power that generates 100k of hashrate with <200,9> would produce 8.33k of hashrate with <144,5>. This would be a reduction in hashrate (measured as solutions per second), but not a reduction in hashpower (measured in the size of the mining network’s dedicated hardware.)
If the exact same number of miners remained after the PoW change, we would expect the hashrate to be 1/12th the prior figure.
As anyone can see on BitInfoCharts, our hashrate in late June was around 26Msol. If we lost zero miners, we would expect 26/12 = 2.16 MSol after the PoW change. The actual figure was about 2Msol, showing a very slight drop - very few miners lost, despite the software changes.
In early July, many makers of mining software were still improving their code to be more efficient with <144,5>; the necessary optimizations are dramatically different from <200,9>. Moreover, at the time of our upgrade, there were no efficient options for AMD miners. As a result, most AMD miners moved to other coins, but enough additional NVIDIA miners joined the BTG network to mostly make up the difference.
Summary: the hashrate declined to about 1/12 the prior rate, as expected, but the hashpower in terms of hardware dedicated to mining BTG remained roughly the same.
How has the network fared since then?
Since August, mining software has improved and become available for more hardware variants. In addition, confidence in BTG has improved, and the price of BTG appears to have been rising to reflect that confidence.
Our hashrate has been steadily improving. Here’s a chart of the past 60 days: