Presenting evidence of increasing mining centralization on the Ravencoin network
Background
This article is published in the context of the current (July 2019) debate on the existence of ASIC miners on the Ravencoin blockchain. It should be read alongside a previous article published by @hardman regarding the possible existence of ASIC hardware for the x16r algorithm.
Discovery
In July 2019 a large miner was discovered on the Ravencoin network. This miner was initially identified by kawwbot, a transaction monitoring and identification bot I developed for the Ravencoin network. The bot detected several large transfers with input from multiple addresses to a single address. This address is associated with the Binance exchange and is believed to be a user deposit address.
A sample transaction can be viewed on the ravencoin.org blockbook explorer here. The majority of inputs for this transaction are newly generated coins from multiple addresses. As the main output is an exchange address (with a small output to a change address), this appears to be a single miner or pool depositing coins on the exchange.
To determine the extent of mining centralization, a custom tool was developed to scan the blockchain for the output of coinbase transactions (i.e. miner addresses). For each block mined, the address was matched to a pool (where known). The input addresses identified as belonging to the large miner are given the name “Unknown Pool A”. The resulting data spans the six months from Jan-July 2019 and is made available as a csv file at the end of this article.
Plotting the data against time shows that the miner first appeared in January. The share of mined blocks grew from 10–20% in February then 30–40% during March. This coincides with the spike in the price of RVN at the beginning of March and an increase of hashrate from 4 Th/s to 12+ Th/s through April.
As of July 11th, this unknown entity is mining approximately 45% of Ravencoin blocks (around 3.25 million RVN/day). With a current hashrate of 10 Th/s we can say the miner has around 4.5 Th.
Assuming a hashrate of around 40 Mh/s for a top end GPU card¹, this would imply a farm of 112,500 GPUs.
For FPGAs with a proposed² hashrate of 500MH/s a farm of 9,000 cards would be required.
This hashrate could also be achieved with 2500–5000 of the FusionSilicon ASIC miners described in X16R ASICs in the fictional world of Westeros, a similar size to the Monero farms in the article.
Conclusion
A single miner or pool controls a significant proportion of Ravencoin hashrate. This is either an extremely large GPU farm, or an FPGA or ASIC miner. If the latter, then it may already be too late to change the block hashing algorithm via a voting process with a high threshold, as an ASIC miner would be unlikely to vote in favor of their obsolescence. It would however still be possible to pass a vote with a lower threshold, or to fork without a vote.
Acknowledgements
My thanks to Ravencoin community members Hardman and Russ for reviewing this article prior to publication.
