Some Interesting Facts About Reputation on Steemit

Since joining Steemit in what seems like June of 1995 (actually last year), I have always wondered how these things work. Like many users, I have many questions such as; How does a post end up on trending? What is the algorithm for the hot tab? How does reputation work?

Recently, thanks to a great post I found by Hive account@arcange, I was finally able to look at the cold unadulterated facts regarding how reputation is calculated according to the actual code. I found the information enlightening. It actually contradicted some of the information I had received from other Steemians. So I conducted some more research and testing in order to gain full understanding reputation. Some of what I found were just confirmations of stuff I already knew (or at least thought I knew). Others were new revelations to me. I have listed my findings in a list below.

The 2 digit number you see near your name isn’t your actual reputation. To find your actual reputation, you can go to your steemd page. It is near the bottom of the table on the left.
Accounts with a negative rep cannot affect rep in any way (neither by upvote nor downvote). So that high payout upvote from Bernie won’t help your rep, but at least it helps your wallet.
Self-votes affect reputation the exact same way that upvotes from someone else do. So you can vote yourself to 78 rep if you’re diligent enough about it and if you have enough Steempower.
Upvotes on old posts have no effect on reputation. That’s too bad. Old but popular posts should earn something.
Upvotes on posts with declined payout will still affect your rep.
As long as an upvoter has a positive reputation, their effect on your reputation is proportional to their SteemPower, not their rep. For example, an upvote from someone with a rep of 77 and 15kSP will be worth less than an upvote from someone with a rep of 25 and SP of 30K.
Reputation is a function of the condenser, not the blockchain itself. Meaning we don’t have to wait for a hardfork to change it. It can be changed by Steemit.inc or another condenser such as busy.org.

With all that in mind, I find the current state of the reputation system to not only be ineffective, but misleading. It only addresses one aspect, rewards received. It doesn’t offer any insight into behavior, trustworthiness, community involvement, transparency or even longevity (mainly in the case of the heavily invested). While I mostly agree that investors should be able to retain greater influence as a perk for their investment, I do not feel they should have an easier path to earning reputation than anybody else. The current system allows them to basically purchase reputation and actually encourages or rewards those who partake in voting abuse.

So how could we fix the system? Well, I think there are many ways in which we can address a problem. One way is by algorithm. In order to use this method, the first thing we would have to do is define behaviors that we feel should result in a positive reputation. So we would look at things like the number comments left, number of unique people upvoted, communities supported, number of downvotes/flags received, average length of comments, etc. Then we would have to figure out the weights of those factors on the algorithm. Hive account@steemchiller’s CSI might be a good starting point for such an algorithm.

The good thing about an algorithm is that it is usually based on good old quantitative analysis. So it takes out some of the subjectivity. However, isn’t reputation more of a subjective aspect by its nature? What are some good qualitative approaches we could implement?

Reputation whether good or bad is something that is earned from those with whom you interact regardless of their success on Steemit. With that in mind, I think a l simple rating system on the account itself will suffice. This way, the outcome of all interactions could potentially be captured.

The best thing about these changes is that they do not have to rely on a hardfork in order to change it.
They could be implemented at the condenser level. It would be nice if Steemit would take the lead though. What are your thoughts? Hive account@busy, do you have an opinion on this?