Magic Communal Trees: Content Validation Trees, Part One

in #steem5 years ago

So how do we hack the blockchain to make people want to use it? Today, I propose an idea called a "Content Validation Tree". It's a vain attempt to separate content into two sets: Really 'Good' Content and Not Really 'Good' Content.


christmastree1808558_640.png
A crazy idea just in time for the holiday season

But I give you a fair warning. This is one of those creatively crazy ideas that will never actually get implemented. But it's a crazy idea I can add to my ever growing collection of crazy ideas focused on improving user experience, user behavior, and incentivization mechanics.

So, what is a content validation tree? It is essentially a subset of pieces of content that are linked together by specific mechanisms and in specific patterns. If you understand the tree data structure, it looks something like that. If you have seen a tree before, it resembles something like that.

You have a single piece of content and from that content different content is linked to each other and branches out. Of course, there are specific rules, but that's the general overview.

So, let's build a content validation tree. Let's learn the general idea and discover the incentives behind building this magical yet potentially powerful content structure.

Let O be the origin piece of content. O is the foundation of our tree. It really doesn't matter what piece of content is O. All that really matters is that everyone participating in the tree agrees on O.

And if they disagree, guess what? There's nothing saying we can't have multiple trees? Hell, you could have community-specific trees. Actually, scratch that. You should have community-specific trees. But before we get ahead of ourselves, let's go back to O.

O at the beginning of everything, is the only piece of content in the tree. Seems kind of lonely, right? Well, luckily for us, we aren't going to leave O alone forever. We're going to add another piece of content C1 to our tree.

But we're not going to randomly select any old piece of content. Because that is boring. We're going to deliberate select certain pieces of content. And we are going to do it in a decentralized and fair way that encourages positive user behavior (I hope).

First, we layout some requirements for a piece of content to be eligible for selection. The first requirement is that any hopeful author needs to vote on an eligible piece of content, E within the tree.

Right now, O is the only piece of content in subset E. So, C1 must come from an author that has voted on O. Next, an author has to explicitly link their content to O. The specifics here aren't important for now, but the author needs to indicate on the blockchain that their specific piece of content is linked to O.

The eligible pool of content to be selected as C1 therefore must be explicitly linked via content and voting on O in this case. If the pool only had a single piece of content, we would simply select that piece of content, but we'll make the growth of the tree slow enough such that there should always be multiple pieces of content available for selection.

So how do we select a piece of content from this pool of eligibility? We have a selector, S pick from this eligible pool of content by finding the piece of content that maximizes the selection filter equation (SFE). The SFE could be any equation in reality, but I'll give an example to give a little context.

One SFE that I could use would involve maximizing the square of the positive voting weight (by the author) on the tree multiplied by the Stake-Based Q-Filter Score squared. The Q-Filter Score is simply a content rating algorithm I invented in prior experimentation and could be replaced with whatever rating equations that tree designers would find appealing.

But how do we select S? And how do we validate the S's pick? The answer to both of these questions is that we need a set of validators, V, whose job is to verify that S has the best known maximum number using the SFE.

Note that we say 'best known' rather than 'best'. Because as these trees grow and as more people interact with these trees, complex SFEs will not only be hard to compute, but given the three second block times, continually changing.

But having a hard problem like this to solve isn't the worst problem in the world. It makes correct validation ahead of time less predictable and creates an interesting prediction competition dynamic for any account trying to become the official S.

How does any selector end up selecting a piece of content? They simply vote for that piece of content and they tell the network of validators the explicit link to the piece of content. The selector can only perform this action once during a selection period. This means selectors can only vote once per selection period or their selection is ignored. This prevents gaming of the system via tactical voting or downvoting.

Once a selector has selected a piece of content and registered that decision with the validators, no other selector can select that piece of content. This creates a race to find a piece of content that maximizes a currently uncertain value. Good luck winning that on a consistent basis.

So selectors basically optimize a math equation and have to make their best guess on which piece of content is the best according to that equation. Validators go through the list of selector predictions and using a specific block time, calculate the SFEs and then announce their support by voting on the piece of content they feel satisfies the criteria.

The piece of content with the most validation stake is then the next node placed into the tree. It's really that simple. To be honest none of that is really simple at all, but at this point hopefully you're getting a taste of the mechanisms here.

But who gets to be a validator anyway? Well, accounts eligible to becoming validators must be either a successful selector at least once or have content within the content validation tree.

This means that you have to either have selected "good" content or have produced "good" content prior gaining the ability to validate "good" content. When we apply this to the local/specialized community level, members contribute to the tree before gaining trust to join the tree and select new content worthy of their tree.

Essentially these trees are community groomed content playlists that use some blockchain mechanisms in order to confirm and validate relevant and 'somewhat-quality' material for the rest of that community surrounding the tree.

So, let's turn to back to our example now that we have loosely defined some of these mechanisms. Several selectors select different pieces of content that they believe belong in the tree as C1. Who picks who is S and C1 in our case? The author of the origin post, O.

But it seems kind of unfair to give all that power to one person at the beginning, right? But think about what they are doing. They are advertising the start of a community tree. If the community doesn't think they're worthy of starting a tree and being a validator, they simply won't participate in that tree and the tree ends up starving out or growing anemically until it dies.

With stronger trees, there is more participation in authoring posts, finding good posts, and validating which of those good posts is the best for a particular round. With all of that voting, content in a healthy tree gets a nice bonus on their content because voting is used to opt-in to the system.

The goal we are trying to achieve with these trees is to reward content for being content rather than some vehicle to cryptocurrency by creating an artificial structure that requires communal earning of 'trust' in order to move into a position of power (validation) rather than simply having the fattest pockets.

In Part Two, we dig deeper into the possible implementation issues, discuss incentives, and maybe address some feedback on this post. Thanks for reading to this post. I apologize if your brain is fried. It has taken a week of brainstorming and being confused about certain aspects before getting to this point. And I'm not even sure how feasible this whole thing is yet. But at least I'm contributing to the conversion. I think.

Sort:  

complex.jpg

It's really that simple.

Now it all makes sense :P

Loading...

Interesting idea, although it means linking to previous posts in order to be associated with a tree? And if you don't link, then you don't get into the new schema unless you get validated by someone who upvotes the post?

In order to be included into the tree, you both have to link with another node in the tree and be validated by a group of validators who retrieve your post from a group of posts chosen by selectors who vote on a post to select it.

Coin Marketplace

STEEM 0.35
TRX 0.12
JST 0.040
BTC 70638.80
ETH 3565.34
USDT 1.00
SBD 4.73