Whenever I actually want to know about the product I'm buying: I only read the 3-star ratings. You automatically can't trust a five-star reviews because that's exactly how the system is gamed to the fullest. You also can't trust a 1 star reviews because it's usually just some pissed off entitled jackass that got unlucky or it shipped late or it was broken yada yada yada. Those reviews aren't helpful.
However, the three star reviews are almost always way more unbiased and give you a pretty fair shake of the product. Sure, it's not perfect... or even great... but it does give some nice information on a product we know not much else about.
Why do I bring this up?
Reputation matters. I've been thinking a lot about potential reputation systems on Hive including my idea for odd-point rankings. Now that I actually have a database up and running this becomes a much more doable project.
Reviewers must have something to lose!
This is the missing link I've noticed about reviews: the reviewer doesn't have a reputation. The reviewer needs to be reviewed and audited by the network itself. We shouldn't be listening to what poor reviewers have to say.
How do we know if someone reviews things well (products/movies/whatever)?
Well it's hard to figure these things out but there are some metrics out there that can help. For example, if a user ranks things on a 5 point scale (5-stars, A/B/C/D/F, etc.), but then issues mostly 5 stars and/or 1 star reviews, we automatically know they are grading on a curve they can't handle. On a 5-star scale, 1-star and 5-star rankings should only occur like 10% of the time on a proper bell curve.
Recursive Reputation
This is also another important thing I've been considering for reputation on Hive. The more people that use the system, the more synergizing network effect the rep system can cultivate. In particular, the outputs of who we trust/support can be used as inputs for those who trust us.
For example, say I've ranked various accounts highly on Hive like @tarazkp or @taskmaster4450. Now, anyone who trusts me within the system will trust my high ranking accounts by proxy until the user says otherwise. There are various algorithms that could be employed to give everyone on the network a personalized experience much like the feeds/walls on Facebook/Twitter and other Web 2.0 social media sites.
There's also an option to give high ranking users certain permissions should they qualify. For example, users could start giving posting permissions to others with the express command that this power is only to be used to downvote malicious content. That way whenever anyone in the group see something that needs to be dealt with they all have massive downvoting powers.
Conclusion
Brain is all over the place today.
No focus.
Perhaps too much FOMO as the bulls awaken.
Posted Using LeoFinance Beta
Return from Always read the 3-star ratings. to edicted's Web3 Blog
