You are viewing a single comment's thread from:

RE: People Rank - Using Page Rank Algorithm for Better Curation and Rewards

in #steem8 years ago

@dantheman I like you are recognizing the shortcomings of the current system (which is already darn impressive) and are willing to consider changing things for a more strategically valuable and sustainable system. I do have some concerns with the model you propose, but also have some ideas which may be worthy to evaluate.

The model you are recommending will create a distributed tree for whales to empower others to share in the distribution of rewards. The biggest and most un-resolvable problem will be in the inherent structure it will establish, which will be reinforced by expected self-serving behaviors. Not all will fall into this pit, but the system will definitely support it.

This model, as I think your describe it, basically sets in motion the creation of dynasty's, where the whale/Emperor empowers a second class who will always be loyal and the wealth will continue to remain mostly within these selective circles as the acts of self-interest will continue without barriers. In fact, in a worst case scenario, can you imagine a situation where whales connect to other whales and the power distribution remains very flat and within a relatively small group. There is no real motivation to vote outside the elite community as part of this system thus propagating classes with a great divide. I think in this system, the poor will continue to struggle for the outlaying scraps.

If the goals of change are:

  1. Retain the value principle and current structures of Steem Power as a mechanism of 'mass' to move rewards
  2. Encourage a steady stream of good content creation by ever more of the community
  3. Identify and reward the best of the content, regardless of topic, author, or self-interest reward
  4. Identify people as domain experts, trustworthy, and valuable contributors as part of a reputation system
  5. Not overly burden the system with new complexities or new measures of value
  6. Promote active and regular participation by all Steemit members (reading, voting, etc.) but not allow for dilution via too many votes by bots or mass-voting tactics. I.e. vote power controls must remain in place.
  7. Give no advantage to those who would create multiple accounts to vote. Kill-off new accounts, after a period of time, that do not contribute to content. (unused accounts or accounts just used as part of an upvote bot-army)
  8. Empower all accounts in good standing (even new accounts) to have some measurable meaning to voting and participation in a system which rewards good behaviors (the act of just voting) with more weight (outside of the SP and SD system - don't want to muck with that if at all possible) that users can witness benefits to their participation
  9. Tie downvoting to expertise of the voter and provide automated specific weights as to the reason
  10. Empower whales to more easily facilitate and oversee dispersion of their influence to trusted parties (ie. like @smooth) and give them the ability to designate topic experts
  11. Make allocation of steem power a voter-controlled variable to facilitate scale-ability of up-votes (from minnows to whales) Ex. so whales don't have to upvote hundreds every day to disperse their value to the system.

If these are the goals you are working towards, I think I might have an structure which will achieve these while making Steemit more extensible, user-friendly, and sustainable over time with respect to handling more content in a better organized and curated manner.

It boils down to using the current factors you already have (SP, Reputation, Vote Power), which don't need to fundamentally change in any way. The key will be using SP like 'mass', Vote Power like 'speed' and Reputation like 'direction' to create a model where Vectors are aligning to good content. Such model does allow minnows to pool for a collective vote which rivals whales, but at a cost of limiting the number of votes. A set of swim-lanes become a forcing function for better dispersion among the topic categories. Upvotes tie to reputation of categories which then power the weighting of down-votes.

...Okay, I will hold there without getting into the details or mechanics. If you are interested I can create a separate post and elaborate. I think it is doable without cratering the system or forcing a major redesign. Let me know if you find my ramblings interesting and I can create a presentation, video, something, to outline the changes I have drawn sloppily on my whiteboard.

Sort:  

Pics or the whiteboard didn't happen.

Yes, I didn't take a picture of it as, well, it is just plain messy. I have lots of different thing co-mingled, but it makes sense to me. I can clean it up, put it in a powerpoint presentation if people are interested. For right now I will sleep on it and see if I can poke holes in what I am thinking.

Honestly, I didn't think anyone would read my post. It makes sense to my eyes as I was scribbling in 4 colors. I was just doing it as an intellectual exercise then writing it up to get it out of my head. Speak up if anyone wants to see my ideas. I think I can hit all those goals I stated, but peer review is really the litmus test.

I'm about to start scanning and posting some of my concepts. What the hell? Maybe it could help spur ideas elsewhere or bootstrap a project totally.

I have drafted a more detailed proposal based on my ideas here: https://steemit.com/steemit/@mrosenquist/steemit-proposal-for-developer-and-community-evaluation
Take a look and let me know

he biggest and most un-resolvable problem will be in the inherent structure it will establish, which will be reinforced by expected self-serving behaviors....

This model, as I think your describe it, basically sets in motion the creation of dynasty's

I am not sure if it is the biggest vulnerability, as I explained another one in my comments, but I agree this could be a degenerate outcome. Afaics, we would incentivize a top-down rigid structure, which I also mentioned in my longish comment post.

Btw, I didn't upvote for your algorithm suggestion, which sounds to me like probably not correct.

In all fairness, I have yet to outline the application of any algorithm. I was just validating the goals and limiting parameters to the problem (technical, behavioral, and process). If you really want to know how I think the system can be modified to attain all the goals I stated (assumed), let me know. I can produce a strategic framework for review.

...and don't worry about not upvoting. Only vote for things you respect or agree with. :)

I did upvote. I just meant my upvote was for the part I responded to. Thanks.