> New idea for CI Trust Scores

Kanban forum

Meen shood bee band

Also when high trust people like a post, increase that posters score slightly.  The same for when they dislike.  Also make it so people with lower trust cannot lower trust scores for posters with higher trust.  In fact, maybe make it so if a low trust person keeps claiming "fake news" or dislikes a high trust person's posts excessively, it just lowers their trust score more.

wrote:

I know, how about a trust score isn’t relative to another poster’s score until one or both have rated a post of the other’s.

Then you can reward finding good and pointing out evil. Who rated first could be a thing.

It could be a game.

For any web of trust, initial seeding is very important.

anon user #3 wrote:

Also when high trust people like a post, increase that posters score slightly.  The same for when they dislike.  Also make it so people with lower trust cannot lower trust scores for posters with higher trust.  In fact, maybe make it so if a low trust person keeps claiming "fake news" or dislikes a high trust person's posts excessively, it just lowers their trust score more.

Just abolish trust scores and let people post.

Seems to work for every other forum on the internet.

wrote:

Just abolish trust scores and let people post.

Seems to work for every other forum on the internet.

That’s simply not true.

how would you determine which IP's are trustworthy??

A lot of people connect through hotels, airports, coffee shops, different countries, etc... and may then use a VPN.

Also you could eliminate the "SuckItLiberals" spam without even touching the "trust score"

You need to be more honest about what the point of these scores is... because from our perspective, it is there to make the new users suffer needlessly.

., wrote:

how would you determine which IP's are trustworthy??

A lot of people connect through hotels, airports, coffee shops, different countries, etc... and may then use a VPN.

I pay for a service that provides metadata about IPs. That would be a big key to the score. On top of that we have previous history for each IP. Those two things combined would tell us a lot.

A Comcast address with two years of history on CI and it comes in without a cookie? Give the cookie the benefit of the doubt and boost the trust score.

A VPN from Malaysia that has repeatedly been banned on CI? Penalize.

Etc.

You don’t understand the problem space.

This is exactly like a credit score.

“Is this poster going to wreck the board” is the important question but you do not recognize the efforts people go through to try and wreck the board. Many aspects of time come into play in determining that. So time must be a critical aspect of the system.

That said, after the optimizations earlier this year, I don’t think anybody behaving normally would be stuck in medium-low for a week.

HOW ABOUT IF THEY DONT SPAM FOR 30 MINUTES YOU TREAT THEM AS VERY HIGH TRUST?
\

wrote:

The distribution should be about 90-99% very high and then a few board wreckers.


This isn't a credit score or points on your driver's license... this is a binary variable not a continuous one... and that variable is:  "Is this poster going to wreck the board with spam?"

So we don't need people dwelling in "Medium-low" for a week. 


But when I look at math books there's a distribution.
\

Name and shame every regular poster here.

Give them an estimated score.