Image: shedquarters, CC0
There were a couple of posts lately about finding decent authors to follow and a lot of lists with various great author were filled (and hopefully followed). Let's take another approach here: why not check the blockchain data to find out who gained the most followers lately?
Wisdom of the crowd, what could possibly go wrong?
Boy, was I wrong...
Here's the list of the top 20 accounts that gained the most followers since Oct. 1st 2019, so within the last ~8 weeks:
| Author | new followers |
|---|---|
| steemyoda | 1594 |
| raise-me-up | 1106 |
| voteme | 735 |
| tipu | 701 |
| marcusmalone | 654 |
| sp-group | 642 |
| therising | 562 |
| dtube | 552 |
| anomaly | 534 |
| merlin7 | 527 |
| followforupvotes | 511 |
| haccolong | 489 |
| pasie15 | 481 |
| crypto.piotr | 471 |
| hoaithu | 470 |
| crypto.talk | 468 |
| aekraj | 464 |
| puncakbukit | 463 |
| tuanis | 459 |
| big-whale | 447 |
There we go, highly dominated by bots. Help me - Which is the highest ranked "human" account in this list?
- steemyoda has ~700k SP, hasn't posted for months and has a buggy bot script but seems to give out generous votes. but 1500+ follows based on this?
- raise-me-up/voteme/tipu/therising - maybe also follows due to votes given out?
- marcusmalone wasn't posting for months and seems to run a script to follow anybody, which may get them follow-backs
- sp-group: a polish curation group - is this the winner here?
- dtube: no need to explain I guess
- anomaly and merlin7 look like curation bots
- followforupvotes and haccolong: follow for upvotes...
- pasie15: their daily inspirational quote certainly isn't the reason for 400+ new followers...
...
OK, this approach didn't work very well - at least not to find authors worth to follow :D
This post was possible with the free hivemind database access given out by @emrebeyler:
Thanks a lot for this!
Hivemind stores all follow and un-follow operations in a table called hive_follows, with the state (follow/unfollow), the user ids of following and followed account as well as a timestamp. Here's the query used to get the top 20 list above:
select a2.name, count(a2.name) as following, hive_follows."following"
from hive_follows
join hive_accounts a1 on hive_follows.follower = a1.id
join hive_accounts a2 on hive_follows.following = a2.id
where hive_follows.created_at > '2019-10-01' and hive_follows.state = 1
group by a2."name", hive_follows."following"
order by count(a2.name) desc
limit 20;
Note that this query might double-count situations where users follow another account while already following it or unfollow->follow situations.
Any particular data you'd like to see from a hivemind DB access? Leave a comment below and I'll evaluate.
