We need a better system of voting on Steemit. Here are some of my thoughts on how to improve things...steemCreated with Sketch.

in #steemit7 years ago (edited)


These are some of the factors we supposedly quantify about a blog post with just a vote or flag:

  1. Post was entertaining
  2. Post was informative
  3. Post was useful
  4. Post was offensive, but not totally
  5. Post was obnoxious and needs to be flagged
  6. People who voted for this post are part of a clique that self-promotes
  7. Post was upvoted by bots with no aesthetic appreciation skills
  8. Post was incendiary to certain parties, but within the rights of free speech as guaranteed by the U.S. Bill of Rights
  9. Post was blasphemous and the poster should be dealt with according to sharia.
  10. Poster is making too much money and needs the wind knocked out of their sails.
  11. I'm in a bad mood and need to lash out at someone.
  12. Post is upvoted because it advances the values of the community
  13. etc. etc.

What I'm getting at is that voting is not a simple yes or no switch. It is really a large panel of switches that need some critical thinking and earnestness to effect a meaningful evaluation of a post. Put simply, a simple vote just doesn't cut it. We need something more.

questionaire.png

A questionaire evaluating each blog post may be a solution. The science of psychology can devise metrics that would evaluate responses to the questionaire in the area of how steemians perceived the level of creativity, entertainment, informativeness and other important vectors after they read a post.

The next problem is getting people to fill out the survey questionaire on the blog post. Who wants to take 20 mintutes to fill out boring questions? One answer is to pay them for their feedback. Nonetheless, this might seem to be counter to the Steemit objective of fostering fun, easy, worthwhile and productive social media.

Brainstorming the "voting" problem, perhaps machine learning (ML) will provide a solution. Assuming we can get a consensus on the writing qualities we want to see on Steemit - grammatically correct, correct spelling, making sense, originality, creativity etc. - we could develope ML algorithms that evaluate posts and dispense rewards based upon ML analysis of the data. This could even be done in conjunction with standard voting by allocating a percentage of the reward pool to the results of ML analysis of posts.

However, I suspect that most steemers are not acquainted with ML and it's capabilities. Suffice it to say that they are quite impressive. For example, scientists have created ML algorithms* that are effective in spotting trolls. In addition, the algorithms Google uses to detect spam, while not perfect, are formidable in their ability to block email spam. Furthermore, Google algorithms do a great job of blocking spam and off topic posts on Google+ Communities as I have witnessed over and over again in the communities I moderate there.

A sliding scale for votes, proof of stake, plagiarism bots and the voter reputation factor are steps in the right direction, but we need a wider scope evaluation algorithm than vote or flag.

Another possibility is taking Steemit beyond the stereotypical user experience of one person viewing images on a laptop, tablet or mobile and then responding on a qwerty keyboard. If the VR headset technoogy takes off, their would be the opportunity to measure human response to Steemit posts. Sensors would measure eye movements as steemers read articles, looked at pictures and listened to music/talk. Galvanic skin response and ultimately brain wave patterns could provide data as to whether a post had the kind of content we desire. Advertisers will be using these techniques to sell soap and toilet paper, why not use these techniques to foster good content?

In conclusion, my main objective here is more to think outside the box than to provide concrete proven solutions to the "voting" problem on Steemit. I hope it inpired some creative thinking, because we need something bigger and better than the current methodology.

References

*How a Troll-Spotting Algorithm Learned Its Anti-antisocial Trade


Thank you for your attention and please upvote, resteem and comment below...

Sort:  

I always like to read about Steemit voting system and how to change it for the better. As a newcomer I dont really find it bad, but it is so interesting that people are trying to find better solutions and write about it. Keep it up!

I think you may discover some weaknesses in the voting system as time goes by. It is very important here because voting is at the heart of the reward system, both curation rewards and author rewards.

You take it a bit far, but i've been thinking along the same lines. At the moment i have serious doubts about the curation process. There are far too many inconcistencies.

I like having ML doing at least some curation because it will be consistent and fair because the algorithms treat each individual the same. Furthermore, ML algorithms can evaluate an unlimited number of parameters and they dont need to be paid to do it.

that's interesting. But I guess the ML algorithm would be cheated by smart bots. :)

Would it be cheating if the smart bot analyzed the ML data and wrote a blog post that titillated Steemians so intensely as to actuate their clicking of a maximal number of votes for the article or comment? The Steemians would be tickled pink at the quality of posts as demonstrated by their votes.

Coin Marketplace

STEEM 0.19
TRX 0.15
JST 0.029
BTC 64349.20
ETH 2673.53
USDT 1.00
SBD 2.83