You are viewing a single comment's thread from:
RE: SteemReports - New Incoming Votes Analysis Tool
Thanks!
No, it doesn't use SteemData, my server downloads the blockchain data from the RPC server, and parses the information directly. The code is pretty simple, so little to go wrong, and for some services I have actually found admitted errors in other people's stats though trying to validate mine. I'm fairly confident in it, but of course, I'm not perfect, and can't afford to spend too long testing it.
This particular pie-chart uses 14 days worth of data, and updates every 5 minutes whilst the RPC server is working. This 'Info' button in the top-right should tell you more about each graph/tool.
Thanks for the detailed reply, that answers it. 🙂 Do you have your code open source?
I might use it as a comparison in the next report I do, thanks 😊
I'm afraid it isn't currently open source, as I need to try and recover my server costs and development time, so don't really want others capitalising on the work before @steemreports is better known. I will open source it if/when the reputation reaches around 60.
Two more things (that would really be suited to a project manager like GitHub).
1. weighted by percentage
These votes are all just counted equally right? Wouldn't it be interesting to see them weighted by percentage? Some votes might be 0.01% if user's have enough SP, which is 1,000x less than a full vote. When getting a picture of self voting this could be very relevant.
2. separate charts for posts / comments
This is of particular interest to me
These charts show the percentages calculated as the sum of all positive vote weights multiplied by the most recent total number of vesting shares owned by (and delegated too) the account.
There will be some small inaccuracy because the SP may change over the course of the 14 day period, and only the most recent SP value is used in the calculation.
Also, this doesn't take account of changes in voting power, but to handle all these things would be hard in software that updates on a five minute basis I think. I may reassess this in future, but think it's fairly representative for now.
http://www.steemreports.com/top-voters/ shows only votes for articles, not comments, but I agree it would be nice to show them together nicely.
Yes definitely very representative I would think, but it's good to know the limits of the accuracy. No representation can tell the "full story". But it's already far and away the best tool for this.
If you find the time it would be interesting to see posts and comments separated.
🤓
That's excellent and very commendable.
Since you don't have a GitHub or similar for it, where can I report bugs? I just found one which I consider a bug (but you might consider it in need of clarification only). Your calculation includes down votes as well as up votes, which while they are all of course votes, does not give the correct impression of what most understand a self vote ratio to be.
Example: http://www.steemreports.com/incoming-votes-info/?account=adm
adm self votes a lot, and also only votes for bayrene. But if you look at what adm is voting on you see that they are actually flagging bayrene, who is being targeted by steemcleaners.
Can I recommend you only include up votes? Or perhaps show two donut charts for up and down? I actually can't use this information as it stands for my project 😕
From the link you mention we see that bayrene is voting for adm.
If you're after who adm is voting for, you need to look at the 'outgoing' votes, so you should refer to:
http://www.steemreports.com/outgoing-votes-info/?account=adm
Flag votes are excluded for both incoming and outgoing charts.
Derp. Thank you! 😅
And btw that is quite interesting to see for different reasons, the amount of self votes which are incoming 🤔 Thanks again!