Steemit Daily Stats (In Development)

in #steemit6 years ago

In Development - Steemit Daily Stats


Stats Steemit

Source : SteemKR

Everything is currently in development.
I would love to see some comments about what statistics you would like to have !

If you know :

  1. some circle voters,
  2. some people using upvote bots heavily,
  3. some people who are manually curating,
  4. some people just using Steemit as it was intended,

Please tell me !

That way I can cross-check with my algorithm, and make it better !

Line

Typology Of Accounts (estimated)

Typology Of Accounts
Based on 2799 accounts analyzed by the algorithm

Which accounts are classified as Fake :

  1. Every account that receives a delegation before any real activity (except the Steem ones),
  2. Every account that only receives votes, or only gives votes,
  3. Every account for which self-upvotes are systematic and predominant,
  4. Upvote bots.

Which accounts are classified as Suspicious :

  1. Circle voters (some of them are classified as Fake thanks to points 1 and 2),
  2. People whose majority of upvotes come from upvote bots,
  3. Everything that deviates from the normal use of Steemit.

For bettering this classification, I need your help. If you know for certain some accounts are circle voters, upvote abusers, genuine manual accounts or any particular type, please tell me so I can improve my classification algorithm. I can also tell you how YOU are classified and why, so we can discuss it, always with the goal of improving the classification made by the algorithm.

Line

This first step is primordial, as it will be used to make other statistics.

The statistics that will be displayed :

  • First of all, the typology of the accounts (as just shown).
  • Posts per day -> Classified by nature (real posts VS bot posts) and by application.
  • Comments per day -> Same classification : nature and application.
  • Votes per day -> In Rshares : classified by Normal / Suspicious / Fake to see who's taking the biggest part of the rewards' pool.
  • Transfers per day : Classified by - with upvote bots / with exchanges / with curation bots / with game bots / between accounts.
  • The biggest transfers of the day (between accounts only).
  • The amount of Steem powered up, down and delegated for the day.

If you would like to see other statistics, please let me know !

For the graphics, here it's done with matplotlib but I will surely take a more pleasant-for-the-eyes library.

The statistics will take some time before being really representative of Steemit, as it takes time for the algorithm to make an assessment of each account, but even before it assessed all accounts, it will give a better understanding that raw statistics with no analysis of the nature of accounts / posts / comments / votes.

Thanks for reading,
And thanks in advance for your comments,



To show your support : please upvote, follow, resteem or comment !

Or all of the above

And please check my previous post :
Curation Bot : The choice of a "business model"

Sort:  

I don't think statistics about circle voting and people using upvote bots heavily go in the right direction, for the same reason that I did not support the "downvoting" proposal of @heimindanger.

For me it's the wrong approach as it generates a lot of negative energy and distracts us from the only thing that will make the success of Steemit: great content.

You don't need rewards to have great content on this platform. Take for instance Unenumerated, the blog of Nick Szabo: he didn't need rewards to produce outstanding content for many years blogging. On his blog, for which I bet he never received a dime, he made history.

Like him there are many bloggers with outstanding content - they would produce that content regardless of the rewards. However what they do need is serenity, a creative and stimulating atmosphere and community. Nobody likes negativity, acrymony, flame wars.

So instead of producing statistics that display the dark side of steemit, why not turn around and look at the positive figures?

I would like to see statistics for the people with the most cumulative words posted (as daily average), whether as posts or as comments (a split between posts and comments would be good too).

And if haejin is among the top producers (not sure since his articles are not very long) then it will go to show that even if many people do not appreciate his writing (I don't) he still does work in the sense of the Steem whitepaper

Another stat I would like to see is a ranking of the people who "spread the wealth", meaning who allocate the biggest part of their voting power (proportionally, but above the "dust" threshold) to as many accounts as possible.

A ranking of the people who "spread the wealth" <<< is published by @boddhisattva. He has all kinds of information over there in his feed that talks about who spreads the wealth in various tags.

Also you can join the curation league of @abh12345 and learn a lot about who is actively curating.

Thanks a lot for these discoveries @fitinfun !

I'm gonna have a lot of reading before finding a way to make something more useful than this, it won't be easy.
The only thing I see (right now) is encouraging "smaller" authors. The idea of a daily words' mean of @sorin.cristescu (posts / comments) would allow active lightweights to get more visibility.

For the curation part, the League 2 (the n°1 is good too, but more oriented on ROI) of @abh12345 is really interesting, and for @boddhisattva it's great but it could be "improved" by implementing some sort of a relative formula, to allow people with limited Steem Power but who also spread the wealth (according to their means) to still be displayed.

Thanks again for pointing out the great contributions of these authors !

You're welcome. Another good place to look is in the curation contest by @thedarkhorse. This is going on for 90 days and is going into the fourth week now. You might find some insight there too. Good luck with your project.

I didn't think about the consequences of this negative approach, and I think you're right.

I haven't been here on Steemit for long, so I don't understand yet all the intricacies of the ecosystem, but I think some practices need to be pointed out (Because for me, that could be categorized as abuse, and it's damageable to Steemit, as it can discourage a lot of newcomers - whether they were here for money or for writing good content -, I've lived it and I know others who feel the same). If we don't speak about it, it's sure we won't act to stop it - even if a lot of people know it exists. And with more people knowing about it, the more chance we have to find a good compromise, satisfying everyone (investors, curators, authors). I don't have a solution (I discovered all this too recently).

So taking your comment into account, I think a way to proceed would be :

  1. We need some more transparency - we have the blockchain but not anybody can produce their own statistics -, to see exactly what's going on. I'm not aiming at citing names or such, just getting some orders of magnitude about the different phenomenons. Just food for thought.

  2. For your proposal, it would be a perfect way to counter-balance this negative approach - even if I consider it more like a transparent approach, for information purposes only, no blaming, no pointing any finger to anyone - by having statistics about genuine authors and voters, and in this part, why not cite names ? I think they won't mind.

  3. And for the order of presentation, let's put the good news before the bad ones...

Thanks once again for your insight @sorin.cristescu

For your information, all statistics I cite are ready to be published, but I need to be sure that my classification algorithm is good before posting statistics that might not reflect the reality.

and then your aim if to sell your algos to enemies of mankind and france for them to target and attacks their opponents? anyway, what ever your success it will be buffed by plaoc :).

Hi @sweecee !
What's the meaning of plaoc ? I couldn't find anything, maybe there's a typo in there.
Before I can make a real answer, I need that.
But it's not a missile guiding algorithm, and nobody will want to buy it :D So no problem here...

no, i have high hope about your talent, I don't hope to be wrong...

plaoc? hehehehe... they are the friends of the mss... they protect china since day one a few thousand years ago... people liberation army of china ,ministry of state security... both kill the enemy, domestic, foreign or aliens :).

Thanks @sweecee, I'll try not to disappoint you, but it will take time to have something reliable, algorithms take a lot of time to develop. Especially since I'm still new to Steemit, there's still a lot for me to discover.

For PLAOC, shit, I would prefer to have them on my side than opposing me ! That's mere common sense ! :D
Emmanuel "The Gigolo" Macron (joke, no offense, in fact, no, you can take offense) won't be of any help and his puppeteers have a Nation to ruin and bring to its knees, a whole social security system to deconstruct, no time for real politics and international affairs - so we'll just follow the US foreign politics, which is not at all in our interest...
I love France for its people, not its government. Is betraying a traitor a betrayal ? PLAOC, MSS or whatever your name is, if you hear me, I'm open to suggestions ;-)

and the food!!!! Not to forget the food that those terrorists from america always try to destroy and hollowed out.

Is betraying a traitor a betrayal

NOOOOOOOO !!!! it's the only path to survival specially when the traitors is about (white) genocide...

I know, I know, I wanted to have a bot for steem... one that does autovote... but still a work in progress even if I just need to setup it as it is already coded ;)

ps you don't need to put the @sweecee if you reply directly to a post or a reply as it goes directly to my reply tab :)

Coin Marketplace

STEEM 0.30
TRX 0.12
JST 0.034
BTC 63688.35
ETH 3125.30
USDT 1.00
SBD 3.97