I've been look at a method to use the wisdom of the crowd and perceived lack of interest in highly rewarded content to help with judgement on downvotes...
not original source but no reference at IBM.com sooo 🤷
At the time of writing and according to my data source (UTC), there are still 8.5 hours left of August 28th 2019. And the total count of downvotes so far today has surpassed the total of any day in the past week. Even more noticeable is the combined weight of these downvotes:
| Date | Sum Count | Sum Weight | Full downvote equivalent |
|---|---|---|---|
| 01/08/2019 | 3242 | -9720452 | 972 |
| 02/08/2019 | 3102 | -8684254 | 868 |
| 03/08/2019 | 2134 | -6704068 | 670 |
| 04/08/2019 | 2529 | -8214811 | 821 |
| 05/08/2019 | 1939 | -6854359 | 685 |
| 06/08/2019 | 1641 | -4175303 | 418 |
| 07/08/2019 | 2452 | -5814644 | 581 |
| 08/08/2019 | 2449 | -7163653 | 716 |
| 09/08/2019 | 2812 | -10839089 | 1084 |
| 10/08/2019 | 2790 | -7484883 | 748 |
| 11/08/2019 | 2383 | -8832999 | 883 |
| 12/08/2019 | 1823 | -7206883 | 721 |
| 13/08/2019 | 4534 | -8250523 | 825 |
| 14/08/2019 | 4467 | -7209649 | 721 |
| 15/08/2019 | 3655 | -7869205 | 787 |
| 16/08/2019 | 5524 | -8317027 | 832 |
| 17/08/2019 | 3431 | -6248539 | 625 |
| 18/08/2019 | 3353 | -5962928 | 596 |
| 19/08/2019 | 2863 | -6441590 | 644 |
| 20/08/2019 | 3158 | -3636864 | 364 |
| 21/08/2019 | 2841 | -5670519 | 567 |
| 22/08/2019 | 2331 | -3657605 | 366 |
| 23/08/2019 | 2316 | -4737647 | 474 |
| 24/08/2019 | 2081 | -7145175 | 715 |
| 25/08/2019 | 2522 | -7312728 | 731 |
| 26/08/2019 | 2618 | -8319849 | 832 |
| 27/08/2019 | 2235 | -9110369 | 911 |
| 28/08/2019 | 2698 | -16409791 | 1641 |
With east coast America still enjoying their morning coffee, the total weight of downvotes today has surpassed any other day this month with many hours to spare.
#newsteem has spoken.
Prior to the hard fork, there has been discussion on chain (and likely plenty off-chain) regarding where downvotes would be placed - Trending (highly 'valued' content), plagiarism/abuse, and personal vendettas 🤦♂️ to name but a few.
There are mechanics in place for the second and I'm not really interested in the third, but what about Trending? Something has to be there - even if every post was zero'd there would still be a list of content (as far as I am aware). And so, how to decide what should be there or not?
As an aside, I just wanted to mention the possibility of having a collection of say 200 accounts (the downvote Oracle?), if say 20 of which downvoted a piece of content, then (all respected) bid-bot votes would be removed auto-magically.
If the terms and conditions of the bots stated this so their users were aware of the risk, and if 200 trusted community members could be found somehow, that could work?
Back to the content with high pending rewards, the wisdom of the crowd, and a piece of script.
(for the links to work you will need to add an @ next to the username of the account in the address bar. I have not included it to avoid notifications going out to these accounts 🐥)
| link | comments | pending | $_per_comment |
|---|---|---|---|
| link | 1 | 62.807 | 62.807 |
| link | 1 | 57.079 | 57.079 |
| link | 1 | 52.516 | 52.516 |
| link | 1 | 50.744 | 50.744 |
| link | 1 | 47.335 | 47.335 |
| link | 0 | 47.255 | 47.255 |
| link | 1 | 46.967 | 46.967 |
| link | 1 | 46.447 | 46.447 |
| link | 0 | 46.041 | 46.041 |
| link | 1 | 44.473 | 44.473 |
| link | 1 | 43.876 | 43.876 |
| link | 0 | 43.093 | 43.093 |
| link | 2 | 83.557 | 41.7785 |
| link | 1 | 41.55 | 41.55 |
| link | 0 | 41.496 | 41.496 |
| link | 1 | 41.342 | 41.342 |
| link | 0 | 41.138 | 41.138 |
| link | 1 | 40.627 | 40.627 |
| link | 1 | 39.565 | 39.565 |
| link | 0 | 38.545 | 38.545 |
| link | 1 | 38.077 | 38.077 |
| link | 1 | 37.144 | 37.144 |
| link | 0 | 37.089 | 37.089 |
| link | 1 | 34.862 | 34.862 |
| link | 1 | 34.52 | 34.52 |
| link | 2 | 66.232 | 33.116 |
| link | 1 | 32.64 | 32.64 |
| link | 1 | 32.096 | 32.096 |
| link | 1 | 31.835 | 31.835 |
| link | 0 | 31.767 | 31.767 |
The above is a list of the top 30 pieces of content that have 2 comments or less, are at least a day old, ordered by their pending payout divided by number of comments.
(0 and 1 comment give the same value to avoid divide by 0 (NAN) errors which isn't ideal)
So the content is at least 24 hours old, and should have had some eyes / views in Hot or Trending, but there is little to no engagement? Why?
- They 'botted' late (post miles from Trending by now)
- They are just starting out and don't have an audience yet
- The content isn't interesting
It is likely to be one of these 3 things and so a manual assessment should be made - just starting out and promoting a couple of posts is fair game?
I looked at the recent posting history of the account and also grouped accounts along with their comment count and pending payouts before deciding on which content I would remove a small allocation of the pending rewards from.
One account had over $350 in pending rewards and < 10 comments across 7 posts....
Is this a reasonable approach if taking the additional manual steps prior to action?
Or in other words, is highly rewarded content without engagement which is being consistently(persistently) produced by the same account worthy of a downvote?
Anyway, that's how I've chosen my 2.5 a day - today and yesterday. (Watch the engagement rise from those 25 rep friends 😀)
Cheers
Asher