We have seen with several data analyses in the past weeks and months that the overall activity in terms of posts, comment and authors on the blockchain is steadily decreasing. What about those stopping to post, are they still voting? And how big is the amount of SP that is sitting idle at 100% VP and is not voting (anymore?). This contribution tries to analyze how much of the total SP on the chain is actually used to vote.
- Scope and tools
- Context: Current posting/voting activity on the chain
- Analyzing the last voting time of all accounts
- Excluding Steemit's SP
- Results & Conclusion
Scope and tools
This analysis is based on all post, comments and votes done on the blockchain between Sept. 1st until Nov. 22nd, 2018. A snapshot of all Steem accounts including their last voting time stamp was extracted on Nov. 23rd around 18:00 UTC. All data was queried from @arcange's SteemSQL database using
python. The graphs were created with
matplotlib. All query and processing scripts to create the results and graphs shown here are on my GitHub.
Context: Current posting/voting activity on the chain
These graphs show the number of posts and comments created per day on the left and the number of distinct authors per day on the right. The time range is from beginning of September, across the hardfork, until Nov. 22nd. The effects of the hardfork are clearly visible: the few days immediately after the fork where most users were suffering from low RC and the overall reduced number of comments afterwards. In general, both the number of posts and comments per day are steadily decreasing and are currently slightly above 10k root posts and slightly less than 40k comments per day. The figure on the right shows the number of distinct authors in total (green) and for root posts (blue) and comments (orange) separately for the same time range. Apart from the hardfork effect, we can see also here a steady decrease in the number of authors. We're currently at around 10k distinct authors per day. What's remarkable is that we have more root posters than commenting authors - I would have actually expected this the other way around?! I at least write more comments than posts, but this may figure may be biased from bot accounts.
A seconds aspect is the voting activity:
The figure on the left shows the number of distinct voter per day. Apart from the hardfork effects, this number is also slowly decreasing and is currently at around 47k voting accounts per day. The only graph with a slight upwards trend is the sum of the vote values (in rshares) per day on the right. This means that the total SteemPower that is voting per day is slightly increasing, despite the number of voters decreasing. I didn't do the math here since this is only a side effect in the context of this work, but correlating this with the network inflation could give additional insights here. Also, there's a "dip" around Oct. 24th - does anybody know what happened there? The number of voters did not change, but the overall vote value was significantly reduced?!
Analyzing the last voting time of all accounts
The amount of powered-up but inactive SP can be found by going through the list of all Steem accounts. Each account holds a timestamp with the last voting time and the number of own and delegated VESTs. I have grouped the accounts into 8 categories, depending on when their last vote was done:
The graph on the left shows the distribution of all Steem accounts by number based on their last vote. We can see that around 50k accounts voted within the last 3 days. This is consistent with the voting activity graph from above.There are only a few (8k/7k/12k) accounts which last voted at least 1/2/4 weeks ago. A considerable number of accounts, 175k accounts, voted the last time between 4 weeks and half a year ago. Half a million accounts last voted up to a year ago, around 350k accounts more than a year ago. 56k accounts never voted. Note that this histogram does not take the account creation time into account. Especially the last number with accounts that never voted includes accounts that were created at any time since Steem exists.
More interesting for the given topic of this post is the graph on the right, showing the same categories but this time summing up the effective SP of the corresponding accounts. We can see here that accounts with around 108 Million SP in total voted within the last 3 days. Another 1, 2, and 3 million SP voted within the last 1, 2 and 4 weeks respectively. The peaks visible in the distribution on the left between 4 weeks and one year is not that strong here, confirming that the half million of inactive accounts has only little SP. What's strongly visible in the SP distribution is around 49 million SP which last voted more than a year ago and another 23 million SP that never voted. The obvious question is: who is holding that much SP and is not voting?
The result is probably not very surprising - the largest SP holders who haven't voted within a year from now are strongly dominated by Steemit accounts which do not vote at all. For a more meaningful result, these accounts should be excluded:
Excluding Steemit's SP
The distribution by account numbers on the left doesn't change by just excluding two accounts, but the SP distribution on the right has the share of idle SP that hasn't vote ever of within the last year considerably.
Results & Conclusion
- The overall activity on the blockchain is still decreasing. This applies both to the number of posts and comments, as well as the number of authors and voters per day
- Nevertheless, the total vested stake that votes increases slightly
- Analyzing the last voting times of all accounts shows that Steemit accounts make a big share of inactive stake.
Taking the final SP distribution graph that excludes two Steemit accounts with 50M SP in total, we can say:
- 72% of the remaining vested Steem has voted within the last 3 days, 74% has voted within the last two weeks
- 10% of the vested Steem hasn't voted within the last year and 5% hasn't voted ever
Hardfork 20 clearly changed the comment activity on the chain. The overall activity in terms of posts and voters is continuing a downward trend, and the low price probably isn't helping very much to change this. But did we lose large stake holders? At least from these numbers, this cannot be confirmed. The majority of inactive stake was inactive already for quite a while.