16.5 Estimating the number of bitcoin miners

Previous posts in series
Complete data for this post
Summarised data for this post

0. Introduction
Today I read a really well written post by DrHaribo of BitMinter. If you're a novice miner - or even just interested in mining, I urge you to read it. One part of the post caught my eye:

"To experienced miners: yes, it is true, fewer and fewer people are mining. Mining is becoming more and more centralized. Small miners give up when their hashpower becomes insignificant and buying significant hashpower is out of their budgets. It is an unfortunate development. This may be important for the future of bitcoin."

Eleuthria of BTC Guild wrote later on the same page:

 "Just backing these statements up.  BTC Guild today is using less bandwidth per month today than ever, even going back to June 2011.  For reference, there are single unit ASICs today with more power than the entire Bitcoin network had back in June 2011.  The load is also down across the board.  Fewer miners in total with higher average speeds than what we've seen in the past."

As you can imagine, I was a bit concerned about this and I hadn't really seen it coming. However, I do have everything required to make an estimate of the number of miners in the network, so I decided to do just that.

1. Hashrate per miner, by Pool
Regular readers will know that I provide a weekly estimate of hashrate density of miners by pool, illustrated in the plot below.

I obtain this data from the few pools that provide it, plus the known solomining companies such as Megabigpower (which account for one miner each). Unfortunately some of the largest pools don't provide hashrate per miner data so the proportion of the network for which I can track hashrate per user is declining:

(Raw data is here, and summarised data here)

So I'm recording the number of miners for 40 to 60% of the bitcoin network. This means that there will be some error in the estimate I provide - and since the distribution of hashrate per miner tends to have quite fat tails, the error could be significant.

However, even if the numerical estimate of miners in the bitcoin network has significant errors, right now I'm interested in a trend - are there fewer miners now than previously? Any systemic errors in the data should not change significantly over time, so any trend in the number of miners in the network should be reflected in the data.

With that in mind, here's my weekly estimate of the number of miners in the network since April 2013:

A reasonably constant number of miners until October last year, then a spike and another dip.

What can we use to check the accuracy of the results, or the validity of the trend (or both)? Maybe changes in the network hashrate will reflect an increase in miners.

The network hashrate:

and the percentage change in the network hashrate:

don't really show a significant change when the peaks and troughs in miner numbers occur. However, newer miners won't make up the bulk of the hashrate - they'll mostly be starting out with lower hashrate devices.

So, what is another source of data that can be used to check the miner estimates? New miners often head to bitcointalk.org, and contribute new posts and topics as they learn about mining. Bitcointalk.org doesn't provide statistics for the mining section alone, but the overall statistics are provided. Compare the "Estimated number of miners" chart with the "Bitcointalk.org statistics" chart. I'm actually amazed at how similar they are, especially since the mining board is not the largest part of the site.

2. Summary

• It appears that lots of miners joined the network at the end of last year,  and after a drop seem to have stabilised at an estimated 55 000 miners, perhaps slowly decreasing. While DrHaribo's comments and concerns are quite valid and something we need to take seriously, at least it seems that miners are not yet leaving in droves.

• Why the correlation between Bitcointalk.org statistics and number of miners in the network? Perhaps it is simply that new miners miners entering the network and new members joining Bitcointalk.org did so due to the same events  (probably increase in the BTCUSD exchange rate).

organofcorti.blogspot.com is a reader supported blog:

1QC2KE4GZ4SZ8AnpwVT483D2E97SLHTGCG



Find a typo or spelling error? Email me with the details at organofcorti@organofcorti.org and if you're the first to email me I'll pay you 0.01 btc per ten errors.

Please refer to the most recent blog post for current rates or rule changes.

I'm terrible at proofreading, so some of these posts may be worth quite a bit to the keen reader.
Exceptions:

• Errors in text repeated across multiple posts: I will only pay for the most recent errors rather every single occurrence.
• Errors in chart texts: Since I can't fix the chart texts (since I don't keep the data that generated them) I can't pay for them. Still, they would be nice to know about!

I write in British English.