Not exactly, but we've had some fun experimenting with AI agents recently through @love-scout with @hivetrending and I was thinking as to what more we could do to improve the rewarding of posts and users on Hive with the help of it.
Now, I need to preface this by saying that I'm not going to go into details in this post, maybe ever, because with things like these if you're too open about how things work it causes people to change their behavior based on the details to attempt to skew how it works in your favor.
When youtube's algo was leaked, users started begging their viewers to hit likes, leave replies, etc, a lot more than they did before that. When Twitter's algo was leaked/publized, they adapted to the new rules as to what would boost their tweets the most to garner the most attention and potential ad revenue, while CT just realized how screwed they were.
So I'd like to instead talk about the idea, the way I'd like to see it funded and what I think it'd accomplish and how we could avoid extra unnecessary fees as to not point at me and go all "you're just trying to do this for your own profit" when in reality this would in a way negatively affect my own rewards as well as other stakeholders who have stake in the ecosystem, but only slightly.
Okay, so what do we want to accomplish with this?
Earned Hive to feel more worthy.
Easily earned Hive easily end up on exchanges, whether you've landed yourself some big autovotes, some free delegation or you're delegating to a project that guarantees your daily votes redirected back at you - a lot of these things cause you to attempt to "farm" hive by posting daily.
Now naturally one solution to this are downvotes, but at this stage you'd have to downvote half the chain and deal with the repercussions of that endless drama and negativity as if we'd need any more of that currently.
Matter of a fact however is that activities such as the ones listed above cause authors to not care about their content's quality, effort, engagement rate, consumption, attention, etc. They're lunch is there waiting for them as long as they show up and post daily.
There are however authors that do care about their audience, their content, effort, etc, and may still post often or just every now and then but because they're sticking to their morals to not buy votes or trade them or selfvote through delegation schemes are often being underrewarded/overlooked and it may cause a sense of following in the shoes of others, i.e. not caring.
Furthermore the idea would be that, even if you do all the above, you could still maybe earn some more if you start improving, which is the main idea behind this project.
Lack of influence
Telling curators what to change is harder than telling authors, some are just going to continue autovoting those 10 users daily and may not even be contactable any longer, some will continue to sell votes cause they're getting a fee out of each one, while some just wanna curate the way they've been doing and think that's the best way to go forward, ignoring things that others may deem important for the sake of quality and consumption on the platform.
A lot of stakeholders for instance deem it worth more to burn their votes/future inflation than to vote on authors currently.
Manual curation being just as rewarding as auto curation may be a big factor as to why certain stakeholders just don't bother as well.
Needless to say, this project would need large influx of voting power to work well, but if it is possible to gain it at this point in time is uncertain.
Potential solution
One solution could be an experimental account, multisigged for security with key holders also being aligned as to what the exact parameters in question are for how the curation is happening which is what I didn't wanna publicize and talk about here. They'd also understand that an AI agent is following these parameters rather than human labor so the curation work being done could be relatively cheap if not free, although it would need an influx of voting power which could be gained from the DHF.
Basically, a proposal asking the DHF for a certain amount of HBD, which would then be turned into Hive Power, example: 10 million Hive Power, to be used for curation with strict parameters and AI curation.
The account would be shared with reputable stakeholders so the HP, curation rewards and future disbanding and returning of the HP to the DHF would be assured.
In the meantime it would actively curate and technically lower curation rewards for current stakeholders as there'd be e.g. another 10m HP actively curating and competing for the reward pool.
For this experiment to prove it is causing a shift in how people treat their Hive accounts and earned Hive, it'd need a significant voting weight behind it to see changes. Meaning, it's not going to do much if we just rely on delegations as it's difficult to gain them from the current userbase/size without also making clear what the hidden parameters/requirements are to receive curation from the AI agent.
Anyway, that's all I kind of wanted to write about it for today. Naturally since this experimental project is cheap to set up and maintain, it could run on small-time/growing delegations while returning all curation rewards to the authors. I just believe a temporary influx of cash from the DHF to power up a massive account would do it more justice and would see effects sprout quicker.
This would mean that from the current ~200m powered up hive, if this account were to use 10m HP, it'd mean that curators would receive 5% less curation rewards than they currently do and similarly some authors would potentially receive less author rewards than they currently do as well if this project never knocks on their posts, but that'd kind of be part of the point of it.
So I understand that this is not something everyone is going to be a fan of, many will probably want to know more info and risk leaking information to the public as to how exactly it operates, many may even figure it out over time as they see it operate and will be able to guesstimate what it cares about in you as authors.
Hive is flexible so I personally would be interested to see something like this exist and be attempted, who knows, maybe the real solution to downvotes is that if an AI agent is doing it then you can't really complain to them forever as one is used to doing to downvoters these days (or maybe you can but you'd just be talking to a bot that'll instantly reply as to why they decided to downvote you). But even without the downvoting aspect in effect, I think having a "neutral" agent curate based on a set of parameters I think we can all agree as to what makes an author/user/content valueble and what strengthens our token and inflation redirection isn't something people can directly argue against.
Anyway, I'd love to hear your thoughts on it in terms of the curation aspect (the upvotes that is, please don't make this about downvotes as that's not something the project would dare to focus on early on - was just a fun point I thought fit in there if a neutral AI agent did it rather than humans).