CURATION and SBD are BROKEN
Curation:
The selection, organization, and presentation of online content, merchandise, information, etc., typically using professional or expert knowledge.
What does our expert knowledge consist of?
How are posts on Steemit being curated?
Oh well that's easy: by payout value.
How is that working for us?
Durp.
Our whitepaper is embarrassingly outdated. Proof-of-brain is not a thing. It is a faraway pipe-dream. What we have is optional stake weighted tipping. First we had circle-jerks, now we have bid-bots. If I'm being honest, I feel like this is an improvement and we are heading the right direction. However, this is off-topic.
Why curation makes no sense on a fundamental level.
Curation has been embedded directly into the consensus algorithm. Does that make any sense to you? If you try to control inflation on the Steem blockchain you are forced to use this mechanic. Let's explore this concept a bit deeper.
You'll notice I said:
How are posts on Steemit being curated?
and not
How are posts on Steem being curated?
This is a huge deal and is the foundation of my argument. Curation was created and embedded into the consensus algorithm with Steemit.com in mind. It should not be there. It is a huge mistake.
You see, it is impossible to curate something on the Steem blockchain because the Steem blockchain is simply a chain of blocks in sequential order. From this perspective, it would make more sense if Steemit was a forum that showed things in sequential order. However, that is boring and not innovative at all, so curation was born in an attempt to appear unique and intelligent. It has been an abysmal failure.
As previously discussed, Steem is a LEGO set for developers. As also discussed this LEGO set should be as simple as possible while also providing maximum functionality. The more complex the LEGO set, the more it can be exploited by the people who understand it best. This is why I write things like Steem Problems Can't Be Fixed By Hard Forks. Steem's problems are not created by its foundational roots. However, curation is adding zero functionality to the platform because it can be gamed from the outside.
Wait, so how does curation work again?
Steem nodes are not only forced to keep track of which posts/comments got upvoted, but they are also required to keep track of the order of those upvotes. Upvotes get run through an algorithm that awards a higher percentage of curation to the ones who upvoted first.
Fatal flaw
The foolish assumption of this mechanic implies that higher payout is equivalent to higher visibility, which is not a feature of the Steem blockchain, but of Steemit.com, a Podunk website that displays blockchain information in an extremely flawed fashion. You don't base the mechanics of a consensus algorithm on some shitty app you are planning on making in the future. What the actual fuck. It has taken me a full year to realize how stupid this is and put it into words. I always had that gut feeling but never knew where it came from.
As a failed lottery
Curation on Steem is a rigged lottery. The idea is for whales to put themselves out there instead of self-upvoting. The idea is to create a system where the whale can actually get paid more than their full upvote back in curation if they upvote the best content.
The problem with this is that the system is transparent, so no whale that wants curation rewards is ever going to upvote a post that's paying out $5 or more. Anyone with their own bests interests in mind is going to either simply upvote themselves (or auction via bid-bot) or upvote someone else with the vague hope that more altruistic people are going to pump the curation tab while they scoop the majority. This lottery idea would only actually ever work if the system wasn't transparent and you couldn't see how much posts were getting paid out until it was too late, the logistics of which are seemingly impossible.
Wait, who controls what users see again?
The concept of curation itself is what caused bid bots to be developed. We assumed high payout equals quality content deserving of a wider audience. We assumed wrong. When it really comes down to it, Steemit and the other front ends can choose to show you whatever content they feel like. They are in full control of curation, yet we cling to the fantasy that stake holders have this power. This is an illusion. So far, this illusion coincides with reality, but it won't forever.
No! We can still make it work!
Many users here want to take this failure and double down; make the lottery even more extravagant by doubling curation from 25% to 50%. Yeah, that's a terrible idea for so many reasons; the first of which being vote buying, which is already the established way to game curation. A whale upvotes literally anything, scoops the majority of curation reward, and then buys a bunch of votes and scoops all the curation. That's a near limitless money loop right there that's only capped by the number of users willing to sell their vote, and we've already been conditioned to sell our votes at a loss in mass.
Guess what else we doubled down on? SBD production. @timcliff wrote something up and actually had it implemented (pretty cool) so we would start printing SBD up to the 10% haircut limit. The idea here is create more SBD to push the price down and then it will finally get liquidated for Steem. Yeah, ultimate backfire, now we've been haircutting the liquidation payout for a while now and SBD is still significantly under $1. Because of this "feature" our consensus algorithm is now printing way more inflation than the platform ever should.
Slowly but surely, SBD is being converted into Steem. I think we capped out at about 15M SBD and we are down to 11M, so 4M has been burned for significantly more Steem than it was worth when it was created. Oops. That's the price you pay when you peg your asset to $1 and let users claim that $1 whenever they want, even if the price crashes 95%. Unfortunately, SBD can't even do what it's supposed to (maintain the peg) so in my opinion is value is less than zero when concerning is usefulness to this platform.
For a while I was screaming from the rooftops that we need 100% of our inflation to be printed in SBD to bring the value down. I'm glad no one took me seriously because that would have been a disaster. This is just another example of how the complexities of our economy are not really fully understood by anyone. Macro economics creates network effects involving hundreds of variables, creating equations that will never be solved.
Unfortunately, we can not rewrite the past. We can not remove curation or SBD from the history books, but we can change them in the future as we see fit. For example, we could just stop printing SBD and the developers could engage in massive insider trading, buying it all up and then announcing that fun fact afterwords. Wouldn't that be fun? We could even turn it into a Bitcoin clone with mining and half-lives and everything.
I actually don't want to get rid of SBD. I've talked about how it can be fixed. We can implement the MakerDAO self-loaning smart contract to add an additional option for creating and burning SBD. This would basically hard-peg it to $1 compared to its current volatility. It's stability would be several orders of magnitude more predictable. The weakest part of the system is that it relies on Oracles to feed it information about how much a dollar is worth in comparison to Steem, but the witnesses are already doing that with price feeds so whatever.
Curation, on the other hand, will almost certainly be deleted from our consensus algorithm. How do I know this? Because there are already ways to circumvent it, and on a foundational level it is nonsense.
Opting-Out
How can one opt-out of being forced to participate in curation? Simple:
- Upvote yourself and tip SBD instead.
If you self-upvote comments that no one else has voted on after 15 minutes, you'll scoop 100% of the curation reward (more if someone does end up voting after you). Take this money and use it to tip out users you support. I imagine such a system will eventually be featured in some kind of rogue front-end.
Speaking of front-ends, what happens when one gets created where the order of posts has nothing (or very little) to do with total payout? What if this front-end becomes the most popular? Now we're in a situation where curation is embedded in our consensus algorithm and that makes absolutely no damn sense whatsoever because the most popular front-end is curating using a superior method. People will ask themselves: "WTF are we doing? Kill curation now."
Curation through Resteems
I've been pushing this idea for a while, and I'm gonna keep on truckin. Curation should be granted through the process of resteeming. This makes perfect sense because resteeming is the literal definition of curation (not total payout). The post is being displayed to a new audience and is backed by the reputation of the resteemer. The owner of the content should get to choose how much they are willing to share. One person should be able to create a post with an option saying:
I will share 0% curation with anyone who resteems my post.
and on the flip side:
I will share 100% curation with anyone who resteems my post.
It would work like an affiliate program. Posts that were upvoted through the resteem would be credited to the resteemer.
This would be great because anyone who just wanted to get their message out there would be able to slide the curation bar up to 100% and everyone would have a huge financial incentive to share that post. By resteeming, it would basically become your post and you would get 100% of the reward from people who upvoted it because it was reblogged by your account.
Ad Revenue
Speaking of affiliate marketing, I'm sure many of you were disappointed that Steemit has been talking about putting ads on the site. Why? If Steemit is crap just use another front-end. I think it's a great idea. LOL, are they already here? Ad-blocker rules.
Adding commercials to the site is a great way to face the platform outward and get people promoting their own content EVERYWHERE instead of just using a bid-bot to jack up the payout value. It's a self-propelling form of curation.
Again, I think that it needs to be done very carefully. I think we should initiate decentralized ad contracts.
Decentralized ad contracts.
I believe that users should own their own page. If someone comments on your blog and you don't like what they said, you should be able to bury that post without even using a flag. The same is true for the comment the OP wants at the top. This can and will happen, if not on Steemit then somewhere else.
Inline with owning your own blog page, I believe Steemit and its users should be able to bid on ad contracts. Users are already weaving ads and referral links into their blogs, why not forge a contract with Steemit for something more effective? All the contract would do is modify the percentage of returns given to Steemit vs the OP. If the OP doesn't give Steemit a high enough % they can decline. If Steemit doesn't give the OP a high enough % they can decline. However, considering that Steemit is fully in control of curation (who sees what) they could make some very interesting deals.
Think about that for a second: Steemit controls the vast majority of who sees what (because most people are still using Steemit.com), yet curation is directly embedded within our consensus algorithm. Sorry for sounding like a broken record, but this revelation is new to me and I just can't get over it.
Simpler math
The way curation works right now there is a lot of unnecessary math being computed on the nodes. Nodes are tracking the order of every upvote. After seven days passes and the payout is imminent, an algorithm is run to figure out exactly how much curation every upvote receives based on the order of the upvote and how many rshares that upvote provided based on Steem Power (or more accurately Vesting Power based on Steem Power).
It should be obvious that next time we face scaling issues and the witnesses/node operators are like:
HOLY FUCK MY SERVER IS MELTING.
We are going to be looking for ways to simplify the process of consensus. And if you think that isn't going to happen you have no idea what I (and many other developers) have planned for this platform. Developers always push the boundaries of what is possible.
Eliminating curation will be a great start, because even in the case of the resteeming solution being embedded into the consensus algorithm, curation is calculated as a flat fee instead of a weird algorithm based on timing and vote weight.
However, because my resteeming solution is totally optional and doesn't need to be forced on anyone, adding it to the consensus algorithm again makes little sense. This feature should exist outside our basic LEGO set in the form of an outside layer smart-contract.
Conclusion
Plain and simple: less is more.
Simplicity is the answer.
Curation is not simple and can be removed from the consensus algorithm.
SBD is not simple and can be removed from the consensus algorithm.
I don't want to see these things disappear. I want them to work as intended.
Agreed that the curation mechanic is not something that should be in the consensus layer, or at least not as currently constituted. I like the resteem idea. There are lots of front ends already doing what you suggest and disregarding blockchain level curation mechanic when it comes to serving up content / visibility.
Steeve.app uses AI / machine learning based on your interactions with the platform to serve you up content it thinks you will enjoy.
Steempeak.com has super simple solution of displaying posts curated by curation groups, rotating through a selection of curation trails as you visit the website. See e.g. below where it is showing the @c-cubed curation trail (full disclosure I am one of the cofounders of @c-squared / @c-cubed LOL). Even though this is ridiculously simple and relies on the curation group selecting good posts, the result WORKS. The posts displayed on steempeak.com are ~ one million times better than posts displayed on steemit.com trending and hot pages, if my quick mental math is correct ;)
There's a lot to like here. Curation via resteems seems like a really elegant alternative.
The idea an author owns their posts is a big one for me.
I wrote this idea (Authors should be able to set minimum reputation required to comment) back when reputation was the only real metric, but UA score or SP balance would work too.
Even better, would be if an author could manually set the RC cost to comment.
Front ends would leave a little indicator in the comment bar showing how much of your available RC it would take to leave a comment, and it'd keep a whole lot of spam off the chain.
Nice. As long as we have a good group of people thinking about these solutions they will eventually happen.
Without even looking at the open source UA can be gamed like a cheap con. Since my early activity it had been my habit to follow all active witnesses. Now that my 30 choices are fairly stable that practice seemed no longer necessary. The first one unfollowed dropped me 8 places in the UA score. Since then my dive from a 4500-ish placement has probably been drastic. The reason for saying probably is because my delegation has been withdrawn once it was realized what a flaky system it is.
So if you want a gpod UA score simply start following all active witnesses. Study the open sourced algorithm and you are likely to be able to game the UA system even further.
UA doesn't care who you follow. It only cares who follows you.
If that is the case my assumption was wrong. Will dig through their code when time permits. Seems a great coincidence that just as my culling of witnesses began that my ranking started falling off at the same time.
Really like the "curation by sharing" (I think the term resteeming is ridiculous, and try to avoid saying it) idea -- it makes a lot of sense.
I also agree with what you suggest about the way content is organized on Steemit / Busy / Steempeak -- and have thought for a while that organizing by value or by number of votes is broken, considering how easily votes are bought and sold...
I've been working on my own Steem app (fitness content), and currently, all the posts are ordered chronologically. I've been considering for some time now to add an option so that it could be sorted as "Under-appreciated Activities" -- effectively, "various runs/bike rides that are impressive (long distances, etc...) that have a low payout value". After it hits a certain threshhold, it would more-or-less be removed from that list, and could only be found chronologically, or by browsing that users account.
Regardless, it's true that things could be improved. Big believer in keeping things as simple as possible.
I've always been incredibly confused about alternate front-ends. They copy the way Steemit.com does things way too much. Thankfully, things are starting to change, even if very slowly.
Not sure if I am sold on all of it but I like that you are thinking outside of the box.
The ad contracts idea piqued my interest as I have been thinking of a similar idea but unrelated to the front end.
Resteemed for visibility. On another note, what's up with the PoB DAO? Any updates?
There's a vote going into effect on the dolphin counsel at the end of the month to secure the charter... I see you are still in there.
@lyndsaybowes is chatting up the "passionate-ideas" section ready to smite the wicked. Your idea of a defense strategy is right up my alley. I want to create a decentralized script that we can use to completely nullify an account; every upvote flaged to zero; every downvote reversed. The idea is to demoralize the enemy and then add them as front line conscripts into the army.
Personally I've been doing way to much brainstorming and way to little developing. I still have so much to learn javascript-wise it's very daunting. However, once the dolphin counsel is up and running and we actually have the power to do stuff I'll focus more effort there.
I really like how Steemhunt implemented it's own curation system based on how much someone interacts with their platform... but it still doesn't quite work properly... Influencers get a much larger scare (x3 or x5) in order help cool but unvoted hunts up the ladder, but most influencers use their votes on their friends first...
When my coding skills are up to it, I'm keen to create niche-specific Steem front ends that curates based on quality. I'm sure it can be done, but I think it takes a lot of moderation to lead the group.
I think the moderation can be decentralized and incentivized, so that people like you and me will just do it automatically and our efforts can be combined to create an emergent network effect.
Yes... but people will always try and maximise... so I guess the trick is to find the behaviour that you like (ie, people upvoting the best content) and to figure out a way that that behaviour provides the maximum rewards.
In the case of Steemhunt, people were voting hardcore in the first minute because if you were one of the first to recognise a hunt that became popular you got a better score... then the algorithm changed so that unrecognized hunts gave a better score, so people looked for those... in neither case were the 'best' hunts found... and that's the tricky part.
Maybe some people will always maximise and there's nothing you can do about it, and maybe if it's good enough for most people than you'll get the desired effect. It's just tricky when most games, etc teach us to maximise above our own natural instincts.
I think the trick is to find accounts that don't maximize and give them the higher reputation scores.
Resteemed let’s negotiate a reasonable percentage of the curation 😝 but seriously I like your logic, it’s not something I encounter much here...
Posted using Partiko iOS
You're pretty smart there fella! I was against pegging until I was stupified by Tim's explanation that convinced me it was a great idea and saw it slowly break. Some people just don't listen. But what do I know, I'm no master dev. I do know you have nailed down some masterful arguments up there. Followed you and if ever you're in the area, find #thealliance discord server😎
I've been ranting for a while now that curation rewards should be 0.
Ends a lot of gaming for rewards, and that was my only concern, as I am incompetent to reckon the programmatic functions.
Thanks!
yep, I think it's pretty sure they should be optional an exist in a smart-contract outside of the consensus algorithm.
Great concept on the Resteeming as it is almost useless though it represents a valid effort to distribute the information. I had never thought about it being part of the curation!
Posted using Partiko iOS
The interesting thing about all of this is that we don't have to ask permission to make it happen. A front-end can develop these ideas without the green light from the witnesses or Steemit Inc.
The one thing, that has been outdated in the white paper... or maybe it was never so, is that nice graph of post payouts.
Showing a long tail, and a gradual rise.
The reality was a very steep rise, and the long tail was cut short by making the minimum payout almost impossible to get to. 100 minnow votes didn't do the job.
Even now with the flatter algorithms, it is nothing like that nice graph.
Yes, the algorithm should be easily accessible to everyone. The very nature of our consensus algorithm is hidden in a jumble of undocumented code. It's very unprofessional.