RE: SteemWorld ~ Newsletter #2 / January, 2018

You are viewing a single comment's thread from:

SteemWorld ~ Newsletter #2 / January, 2018

in steem •  7 years ago 

I think your CSI statistic overweights number of votes, because one can really raise the number by just giving out a bunch of 1% votes. It results in someone being able to game the stat if they wanted, but not Really be helping the community as much as say someone who was giving out 100-10% votes or something.

Authors get paid when people like you upvote their post.
If you enjoyed what you read here, create your account today and start earning FREE STEEM!
Sort Order:  

That should not be the case and I think it is not. If one is giving himself 1 x 100% and 100 x 1% to different other accounts, the value should equal = 0. Therefore the self-vote rate is being taken into account too.

Do you have an example for an account for which the CSI is higher than it should be? Your value is at 30.125 and you are supporting many people with 1% but also many with different higher percentages (100, 60, 40, 39, 4.25, 5 I have seen) and your self-vote rate is in a normal range. I think your CSI is really good and works great for your account.

As I've stated in my post, the weight of the vote is already taken into account ;) Ausbitbank's value for example is at ~ 100 and he supports many people with his votes as you do :)

PS:
As I've mentioned, the algorithm may still not be perfect at this point and I will optimize it, if it is required. Currently I still don't see a reason for it. Thanks for your feedback!

My account can be the example because while voting a bunch of accounts with 1% I raised the CSI all the way to 50. Its just overly biased to number of users supported and can be misleading if a bunch of 1% votes are used.

Can it be that it is just kind of a display error? Please refresh the page (Ctrl + F5 for forced refresh) and tell me the value. Thanks for your help!

This is what I see when I open your account:


Edit:
If the value should remain the same after refreshing the page, it would be helpful for me to know which browser you are using and in which time zone you are living (How late is it in your country? For me it is 23:07 now). The UTC time is being used but there is one thing I have to change regarding the day change and that point is already on my list ;)

I was doing this yesterday, I've since brought the number back down with my antics today. I was testing things throughout the day yesterday and was refreshing the page throughout the day, I pushed it up from ~22 to about 50 by giving out 1% votes to new users in the introduce yourself tag. I did not have the website open (all of the time) throughout this process. I periodically checked, as I wanted to see if that had a significant impact.

It is UTC 23:00 here, I am in Boston MA (EST)

Thanks for the info! I will check it now with your account.

Some details on the calculation:

  • 1 x 100% on not voted account = 100
  • 5 x 1% on not voted account = 5
  • = 105

(+)

  • 1 x 100% self-vote

(=) 5


Some more calculations to make the value smaller follow afterwards and for already voted accounts there is also some extra logic... (may change a bit in the coming days) The number of votes itself is only taken into account for detecting inactivity and I'm only working with sums based on the votes weight but I will see if I can find something that I may have overseen.

Hmm, this makes a lot of sense to me and I should not have observed what I had. Perhaps the rise was due to the relatively low proportion of self votes I had? I rarely vote my self unless I post something, or I am purposefully trying to piss someone off :D (which I did today, bringing that percentage up quite a bit).

Not sure, I just found this metric to be pretty interesting and potentially very useful, and wanted to ensure it gets thoroughly vetted.

Yes, it's really a simple but still awesome algorithm :) Maybe the self-vote had really a big impact and brought your value back to 24. That would mean by not self-voting your CSI would be one of the highest I have seen so far.

Even if I won't find anything... I'm always thankful for such responses as you did because it can help a lot when more than just one person have looked into it. If I should find anything I will tell you ;)

Sounds good, I am hopeful that this metric will be useful for me in the future when researching users. Could make a difference when curating and deciding whether to really push for a user or not.