User Authority - a personal look

in #ua6 years ago

Besides this being a very ugly post, I also thought it might be interesting to have a look at a few numbers that may effect the User Authority score and have a look at a comparison between those and the actual score. I have no idea how the algorithm works except that according to the top 100 that is visible, a vast majority of them are witnesses which means there is a heavy bias on witnessing, regardless of what the account actually does. This may or may not make sense depending on how you look at it.

Anyway, the User Authority is heavily influenced by the follows and who one is following and last night I went through and took my following count down a few by unfollowing accounts that hadn't posted in the last few months or, I haven't engaged with in the last few months. It won't change my feed much as some hadn't posted in the last 6 months. I went from 150 down to 133 but, this change isn't factored into the UA score yet I think. I don't recommend trying to change the UA score by unfollowing but, I do recommend cutting deadwood from one's feed.

https://steemd.com/@tarazkp

Ok, so my account creation date was 29th of January 2017 and my ID is: 129,866. People talk about retention/churn rate and if you consider how few active accounts there are now (guesstimates are around 60,000), you will see that back then with much fewer users, there must have been a great number of alts created by the first ones in. Interesting in my opinion. People talk about churn rates as if it is an issue but rarely factor in all of the alts that were created and fallen dormant as people often find it difficult to manage more than one or two. Also, there are a great number of parked names that skew the numbers.

Next, let's look at a few rankings which I took from https://mysteemitfriends.online which is a handy tool I came across a week or so back and it is a little bit of fun to have a look at what goes on. For example, have a look at the pending post payouts and how many are bidbot boosted accounts, look at the account join date and reputation and also, how many of the top 50 of those are powering down. There are very few there who are organic accounts.

My rep is 72.74 which is not a boosted rep and that puts me at 115. There are boosted accounts higher. Although the Reputation system is useless, I see my own as representative (to me) because I know what I have done to get there. In the larger scheme of things, the reputation is pointless which is something I recognised at the start of my Steem journey. These days it is the bidbots but, circle jerks of large accounts play a heavy roll in many people's account building too, and always has.

My Steem Power ranking puts me in 573rd position which is very sad considering how many people have earned a fair bit more than I. There re many accounts who have literally earned hundreds of thousands of Steem here but hold almost nothing. Their choice, perhaps it holds price down so distribution can happen. Perhaps. Pretty much though, if you have over 100 Steem Power, you are in the top 5% of holders on the platform and if you have 500 SP, you are a one percenter.

Next is the follower count and I ,don't have so many considering I get a great deal of engagement on my posts in comparison to some. Some large names with 5x the followers don't get anywhere near the same level and definitely not the depth. I have always taken the approach that followers are very important here but follower numbers are not. Who your followers are matters a great deal. I am fortunate that my work and content here has resulted in high quality follows and a fair number of people willing to engage in discussion. It took me six months to reach 500 followers.

So, User Authority (announcement post) is a system where:

In essence, UA can be seen as a new Reputation metric: it looks at the entire Steem Follower Graph and carefully tracks who follows who. It doesn't directly care about how "rich" an account is, or how old it is, or how much an account has posted, but please keep in mind that those factors do indeed indirectly influence any account's UA score.

This is mine.

https://steem-ua.com/

6.074 and 261

Good, bad, mediocre? I don't really know but it is lower than my Rep score (meaningless) higher than my SP holdings (good considering the ramifications of using SP as an indicator) and much higher than my follower count (again, good because of the resteem nonsense and Instagram f4f fools).

The UA is a half-measure in my opinion as even though it looks at the relationships, it seems to miss the engagement levels of users which is of course, very hard to quantify I will assume but, when it comes to interaction between users, it will have to be measured at some point. I am pretty sure @scipio shouldn't release the algorithms but at the same time, this is going to raise questions if this score is to be used to affect user experience and of course, opportunity. After all, it is positioning itself as an authority and with that comes many caveats and concerns.

What would be interesting to see @scipio, is the top 100 with the 'witness multiplier' removed to see how "normal" users rank in the system against each other. The reason is that it could help people identify accounts that are ranked highly so they can be identified, studied and perhaps it will have an effect on other accounts and the way they engage, post and behave. If this ranking is meant to improve the system, users should be able to use it practically to improve their own approaches and behaviors and a lead by example process can develop as it could bring more of the engaged users to the forefront.

I like the idea of it though and I think in time it and others will develop various methods to rank and sort users for various purposes. SMTs are likely to use these algorithms to affect user experience, filter and of course, scale their reward systems in a much more fine grained way than things like 'airdrops on reputation or SP'. It is going to get interesting because most likely, some accounts are going to appear near the top no matter how they are sorted while other will swing wildly. I figure that with these things like most others, consistency is a key.

The Steem environment is changing in many ways which I welcome in many ways but, it is going to cause new and unintended stresses in areas that we may not have experienced pressure before.

Lots of fun to come.

Taraz
[ a Steem original ]

Thank you to @magicmonk as your https://mysteemitfriends.online tool is fun. And, @scipio, @holger80 and the @steem-ua team.

Sort:  

These UA posts are tough to get folks involved in the comments :)

... it seems to miss the engagement levels of users which is of course, very hard to quantify I will assume but, when it comes to interaction between users, it will have to be measured at some point.

Hard to quantify and possibly more open to abuse, but an approach from this angle would be interesting to see.

I agree with the points re: witnesses, it's hard to see who's up there without 'just being a witness'.

I think in time it and others will develop various methods to rank and sort users for various purposes. SMTs are likely to use these algorithms to affect user experience...

Yup, and perhaps not one solution will fit all SMT - if they have a set of these to pick and choose to integrate, depending on the goals of the token then this could be beneficial.

Hard to quantify and possibly more open to abuse, but an approach from this angle would be interesting to see.

A lot of data points and would require profiling behaviours which means some behavioural psychologists, and behavioural economists and the like to build. Just like FB use.

Yup, and perhaps not one solution will fit all SMT

I think they will experiment based on their target market and see if they are able to drive behaviours that support their particular platform.

Some engagement, not a great deal :)

Hi taraz. The more I read ans see of this UA ranking system the more concerned I get. This is going to be abused by circle jerks. It seems that if someone has a high ranking and decides to follow you then your ranking goes up. Who says that this is now not a trade able item and can purchase a follower to increase your ranking. There just seems something so wrong with this ranking system.

It is hard to say at this point without watching it for a while but it is also going to develop overtime. Hivemind is also likely to bring options to tailor it further.

I hope it will be tailored as it may be the right way forward but I think it needs tweaking.

Hi Taraz I think it will and I see they are looking at engagement as well. It's early days still which is good.

Well according to mysteemfriends my rep is 57.73 which places me on rank 8236.
Now the interesting part is that ua rates my rep at 3.211 which does put me on 8214.

Surprised how close this is for my account! The difference is that with posting and commenting my rep will increase but as long as I don't get any new followers my UA rank will stays the same.

And yes, I am finally a 1 per center with own SP of 553, which put me on number 9664 in the list!

Surprised how close this is for my account! The difference is that with posting and commenting my rep will increase but as long as I don't get any new followers my UA rank will stays the same.

Yes, I am not sure how it goes with communities later considering that through there the chances of interaction with a narrower set of accounts and potential follows is likely.

And yes, I am finally a 1 per center with own SP of 553, which put me on number 9664 in the list!

Congrats :D

Thanks. The goal for this year is still to reach 750 own SP. Which will be a challenge. According to my calculation, the account needs to grown with 13 SP per week, which is a lot for a small minnow. But I do have a head start to the schedule.
We will see how things evolve!

Satirical Controversy:

Why do we need user authority when we have so few users? I thought we already knew (collectively) who the bad actors were.

Realistic Response:

People will engage with people that they like, that they're familiar with. The ones that are truly valuable are those that will try to persuade you to see things from a different perspective; not because they want to push their point of view upon you; but to enlighten others to the fact that there's more that exists outside the sheltered realm of their snow-globes.

Closing Thoughts:

I'm not sure which of my two paragraphs above is satirical or realistic anymore. :|

Why do we need user authority when we have so few users? I thought we already knew (collectively) who the bad actors were.

So few real users :P and yeah, we may know collectively but there are highlights that blind the midlevel (B-grade bad actors?) :)

I'm not sure which of my two paragraphs above is satirical or realistic anymore. :|

Well, I Try to increase options but essentially, people should just do what I say ;)

Interesting. I guess my ranks will be among the least. It is encouraging that your hardwork brought you to the level you are on steemit now. Your quality posts enticed me to follow your blog.

Thanks :)

You can check at the ua site using steemconnect.

Ok. Thank you.

Thanks for sharing this post about UA.

I believe that UA is better than RP..

Rep is useless. Not sure about UA yet as I need more time to observe it and how it is used. Is it possible perhaps that high ranking users can find and follow each other to boost their positions?

Yes I kind of agree that REP not the right measurement.

Also I need to look into UA as well and I think it’s possible to find high ranking users can find and follow each other to boost their UA... I definitely need to look into it.

I just commented because my friend @hashcash telling me about UA last night and I saw your post this morning..

@hashcash gets around doesn't he? :)

There is an idea here by @paulag which might add some other interesting wrinkles as wrinkles are ironed out:

https://steemit.com/steem/@paulag/steem-contribution-score-from-concept-to-model#@paulag/re-paulag-steem-contribution-score-from-concept-to-model-20180829t115821654z

@hashcash is awesome!!

I’ll have look into @paulag’s post.

Quick question.. will you be coming to SteemFest in November?

yep, he is.

I am doing my to make it to SF3 best but it depends on family challenges at this point. I hope so though and if you are, we can have a proper chat :)

The reason is that it could help people identify accounts that are ranked highly so they can be identified, studied and perhaps it will have an effect on other accounts and the way they engage, post and behave. If this ranking is meant to improve the system, users should be able to use it practically to improve their own approaches and behaviors and a lead by example process can develop as it could bring more of the engaged users to the forefront.
I like this idea a lot, being able to study success cases could be very helpful to enhance our accounts, results and experience here

being able to study success cases could be very helpful to enhance our accounts, results and experience here

I agree but the problem is that only the top 100 are visible. Perhaps a 'remove witness' view would show a more useful view for average users.

Of course, the witnesses are not normal accounts and don't count for study purposes

Hi @tarazkp! We are @steem-ua, a new Steem dApp, using UserAuthority for algorithmic post curation! Your post is eligible for our upvote! Thanks for your contribution, keep up the good work, and feel free to join our Discord server!

Coin Marketplace

STEEM 0.23
TRX 0.21
JST 0.035
BTC 97816.08
ETH 3416.41
USDT 1.00
SBD 3.21