Misaligned incentives: How voting on GOOD content makes me LOSE curation rewards

in steemit •  2 years ago

So I was just replying to a comment on @dan's most recent post and I was shocked (again) at how easily predictable top content is. Believe me, it isn't because it repeatedly create superior content. Here's how well a (very) simple bot can perform on the current platform:

This is my trending page. Note the 6/8 upvotes on today's top trending posts. I haven't manually upvoted a single one of these posts. They were all early votes that should get (close to) optimal rewards.

How does my bot do it? Is it some wizard? Am I secretly a whale on another account upvoting content after this accoun has upvoted it? Does it use advanced machine learning for text comprehension combined with live analysis of sentiment on the platform?

Well, no. It doesn't even bother reading the text, since it has very little correlation with its popularity.

All it does is it reads the username and uses past data to determine the most lucrative users and the optimal upvoting rate at any given hour during the week.

Without even optimizing its timing (which is a huge part of the upvoting game) and with massive connection issues on my end, I was able to make close to 1.5% of my total SP in a week through curation rewards only. I'm confident that it won't remain that profitable in the future, but that would lead to a 216% yearly increase in SP.

The problem should be pretty clear now:

If you're not part of the bot game, a lot of your potential rewards are being sucked up by them.

It's in fact so bad that I almost certainly lose rewards by manually upvoting content I find interesting (each vote reduces your voting power/ future rewards). And if you don't care about small players like me, you should care about the big ones.

I think we need to realign the incentives by solving some of the visibility, centralization and voting issues. I highlighted some of those in my previous post and still think that incentivized downvoting would help. I also believe that a lottery based trending page based one's upvotes as described by @arcurus would be a huge plus - I invite you to read his post if you haven't. Finally, I also support investigating the people rank voting structure.

Let me know what you think in the comments.

Authors get paid when people like you upvote their post.
If you enjoyed what you read here, create your account today and start earning FREE STEEM!
Sort Order:  

Ironically, I upvoted this post manually.


Hey! So did I, I'm not on my bot's upvote list yet :p


I think if @arhag wrote this...it would have been a $4000+ post...

But then again I read the post, which ruins the whole purpose.


Haha yep, most of my posts are quite ironic in that way. 2-3 posts on how to increase your visibility based on data got close to no visibility, another one proposing downvotes didn't get downvoted and this one pretty much claiming that reading it would be a waste of time if your goal is to get curation rewards.

I really need to understand this better. How can it b that if I vote for some good posts it will diminish my future receiving? Well, I will learn.


Ok, so it probably won't for you. If you look here you'll see that my bot automatically votes on content. However, everyone is limited in the amount of times they can upvote per day. After each upvote, you lose some of your voting power. Your rewards are proportional to your voting power. So if you vote too much for no reason, you'll end up having no curation rewards.

Here, I'm arguing that no matter what I vote on, my bot will always be doing better than me at identifying what will end up being popular. By voting manually, I'm lowering my voting power, which will lower my potential rewards (due to my bot) in the long run.

Hope this clarifies it all. Thanks for the comment!

I hope you all have time to check out my post on the SteemTrain

Seems we think alike :)

Also as your SP grow those reward will contribute way more SP than 1.5% I think.


Well the reward is proportional to your SP, so I'd still get close to 1.5%, but of a larger amount.


That makes sense.

I think the reputation system should be calculated according to your votes and have a modulating effect on the rankings as they are sent to your browser, as well as compounding with the overall global score rankings. Here's my thoughts on some parts of how to do this: https://steemit.com/steemit/@l0k1/downvoting-and-reputation-my-thoughts-on-what-works-and-what-doesnt

This makes it more personal and eliminates the influence of bots.

probably you should stick to 20 upvotes per day? and upvote after 30 min that post is published


Well predicting exactly how many posts should be upvoted in a given hour isn't necessarily something you can do. Your best option is to consider the probability that a good post will be posted at any given time, and adjust your daily upvote rate as a function of that.

Upvoting after 30 min is definitely not optimal on those posts. @wang upvotes after 14 mins, @positive upvotes at 10 mins and @recursive at 11 mins just to name a few. Voting after those leads to considerable drops in rewards. As I said, I haven't worked on the timing of the upvotes at all, and it still performs well.