A good comment area will democratically separate excellent opinions and reasonable but mundane opinions from trollish comments.  A good comment area will also allow a number of us to maintain our anonymity, our avatars, if we so choose.

Currently we have a voting system.  Good comments get voted up and bad comments get voted down.  That helps a bit.  Also we have dedicated volunteers in the background removing cruel comments, thoughtless comments, off-topic comments or extremely long comments simply meant to waste the time of our community of readers.

My initial software improvement would tie the up-votes and the down-votes to the comment’s current position relative to all of the other comments.  The top vote-getter should be at the top of the comments, where everyone would read it.  The bottom vote-getters the real trash, should be at the bottom.  This would be similar to a Disqus policy of sorting comments by “Best”, as opposed to sorting by “Newest”.

Disqus has one problem with its “Best” system.  If you comment on the “Best” comment, your own (possibly dumb) comment is automatically promoted to the second-best comment on the entire comment list.  In a pile of 500 comments this can be significant.  The best comment in a Disqus discussion can loaded with up to 100 such remora comments before readers finally scroll down to the second-best comment.  So, I recommend that the idea of “remora” comments (my term for these hangers-on) not be used.

Next, I’ve noticed that certain commenters have fairly uniformly excellent comments.  I want to read what these people have to say.  Certain trolls have a preponderance of trash comments, while others have a preponderance of so-so comments punctuated by a few disgusting personal rants against somebody else.  So, I propose that the commenting system keep track of the votes received on each of a commenter’s past 1000 comments.  The worst 5% of the comments are what counts, so, when a known commenter writes a new comment, look up that commenter’s fifth percentile score, the 1/20th worst comment on the list, and start her/him off with five artificial votes at that level.  This pushes known good commenters up and bad commenters down whenever they start.  In doing so, average people will be constrained to write good messages.

Preventing abuse:

Trolls can be subtle.  They can dump a spam that looks just barely enough like a real comment onto the system.  They can embed hidden messages into comments, things that would slip past a lone censor, but which a certain percentage of people in a crowd of readers would notice.

So, why can’t trolls simply abandon their old screen names and reinvent themselves with new, anonymous avatars?  They can, and they might.  So, as a general policy I recommend that all brand-new users with new screen names get assigned, on a scale of, say, 5 stars, five artificial comment ratings of 2 stars.  Their comments will be rather low on the list but will still be visible on the list for experienced hands to vote the comments up or down.

Next, why can’t trolls invent 100 zombie names and vote their own comments up into paradise?  They’ll probably try.  One solution is to heavily weight the votes of experienced and trusted hands on the comments board versus the votes of newbies and known trolls.  The votes of as many as one million recently created zombies, each with only a 2 star rating because they never have commented ever, shouldn’t count for more than an aggregate of 10% of the total vote on any comment.  Once a newbie’s rating starts to drift up to perhaps 3 stars out of 5, maybe their vote should fully count.  Between 2 and 3 stars, our trust in them is limited.  Let them prove to the village that they’re neither trolls or zombie voters before their votes fully count.  However, all newbies should participate fully in the voting process from day one as junior voters, slowly gaining seniority with well-graded participation.

An extension of this grading system allows us to shield any newbies to Reader Supported News from any cruel or foul-mouthed comments that have been dropped to the bottom of the page.  Why should a known troll have any right to publish an expletive-filled rant against people, even if it shows up at the bottom of a page?  So, if a brand new reader is assigned a lifetime rating of 2.0, she/he/ze should only be able to see comments rated 2.5 or higher.  All comments with a worse rating will have completely dropped off the bottom of the page.  If a newbie has risen to 2.5, they can see comments rated to 2.0.  With a rating of 3 they can see comments down to a rating of 1.5.  The general formula will be (reader rating) + (comment rating) must exceed 4.5 or else the comment is never shown on the page to that reader, except a reader will always be shown her/his/their own comments.  Experienced hands with high ratings will still see all comments.  In the event that a troll finds a way to deliberately sink some particular good comment, the experienced hands will see that good comment and will help to bring it back up.

It’s important that we slightly reward the good readers who vote our community’s messages up and down.  Perhaps a dedication to votes cast divided by the number of pages that a reader reads can result in a slightly higher lifetime score for their avatar, say, as much as .2 stars added to their score.

The intended results are as follows:

1.  Paid corporate trolls will give up, go elsewhere, find some easier prey elswhere on the Internet.

2.  Some of the people with an unfortunate tendency to start online personal vendettas will move on from Reader Supported News.  Other such people will feel generally constrained to be on their better behavior.  The official role of a good government is to constrain all ordinary people (whether they tend to be better or worse by their personal nature) to tend to do good.  That’s our hope, that ordinary people will participate and act well online.

3.  Voting will be hack-resistant.  We’ll see the community’s true desires.

4.  By moving to a five star system and away from a simple up-down system, we’ll be able to discern a five-star comment from a mundane three-star comment.  In doing so our community will have a tool for somewhat approximating consensus process.  Reader Supported News will have a method of making rather wise community decisions online, with the ability of some community participants to remain anonymous if they so choose.  I don’t think that too many people yet realize what kind of positive power this tool can unleash.