Post: #1

I wonder what the analytics say about banned members. There must be statistical significance that links them. Are there certain metrics that alert forum mods?

What metric do you think is the leading indicator for a troll?

What metric gives credence to authenticity?

As referenced by the past weeks bannings, it is only at the bitter end that neautral reps pile up. Surly there must be a leading positive or negative indicator to preclude direct knowledge of forum falsehood.

My simple thinking tells me that the total number of posts per (divided by) total number of positive reputation to indicates a members authenticity. A significantly low ratio would indicate troll.

Any thoughts? Can we put predictive analytics to forum use?

Post: #2

Like the requirement that presidential candidates approve each campaign ad, there should be a .wav file at the end of every post that says "I'm Roosh V...and I approve this message"

Post: #3

I use a sophisticated system called "reading each post and making a judgement on its content regardless of who said it". The bad news is the OP tipped off my word-salad-troll leading indicator.

Post: #4

It's simple, be authentic. If you're not, your bullshit will come to light eventually.

Post: #5

Are you actually suggesting we preemptively ban people because they may commit a bannable offense in the future according to your 'predictive analytics'?

Post: #6

(06-30-2018 12:16 AM)Delta Wrote:  Are you actually suggesting we preemptively ban people because they may commit a bannable offense in the future according to your 'predictive analytics'?

Post: #7

(06-29-2018 11:16 PM)WashedUPVet Wrote:  My simple thinking tells me that the total number of posts per (divided by) total number of positive reputation to indicates a members authenticity. A significantly low ratio would indicate troll.

Any thoughts? Can we put predictive analytics to forum use?

The rep system is flawed, but if you consider things carefully you can usually pick up on who is a troll or not. I believe I coined/introduced the Posts Per Rep ratio (PPR) roughly 2 years ago here but I'll quote it for this thread's sake:

(06-06-2016 07:40 PM)AneroidOcean Wrote:  This is not infallible by any means, but I go by the following guidelines:

1. What is their post per rep? Calculated by taking the person's # of posts divided by the number of rep points. A posts per rep approaching 50 is generally a solid sign. The LOWER the number generally the better. Keep in mind that generally this metric is more useful for higher post counts. It's easily swayed by a low post count member getting a couple reps and only a few posts. This can still be a good sign, but you have to consider other factors.

Examples as of this post from the top forum posters:
RooshV = 48 PPR (over 16k posts, nearly 8 years creating/posting on this forum)
Samseau = 51 PPR (over 11k posts, member over 6 years)
El Mechanico = 73 PPR (nearly 11k posts, member over 5 years)

Examples from this thread:
samsamsam = 103 PPR
Kona = 37 PPR
justforfun19713 = 2.6 PPR (VERY STRONG, but a low post count)
LINUX = 7.2 PPR (VERY STRONG, post count over 1k)
Merenguero = 57 PPR (over 3k posts, 4 year member)
Anabasis to Desta = 11.7 PPR (over 1k posts, member only 8 months)
cr33pin = 40 PPR (over 4k posts, member 4.5 years)
CleanSlate = 13 PPR (almost 2k posts, member 7.5 years)
hydrogonian = 35 PPR (almost 2k posts, member 6.5 years)
Suits = 33 PPR (over 5k posts, member 3.5 years)
2Wycked = 19 PPR (2.5k posts, member over 3 years)
Satoshi = 15 PPR (under 500 posts, member over 2.5 years)
kaotic = 61 PPR (over 6k posts, member 2.5 years)
262 = 35 PPR (over 1k posts, member 1.5 years)
Cobra = 13 PPR (1.5k posts, member over 5 years)

2. Length of time as a member is something to consider. Most posters that have been here in the 2-3 year range and higher tend to have more solid contributions.

3. Number of posts is also something to consider. If the member has been here for a long time, has a relatively low number of posts, but a high PPR that's something interesting and should be looked into.

Final word: Let's look at some examples. You should always look at the posts that people link to in their rep points or WHY they repped them. There will always be bullshit rep points but someone with some genuine contribution value to the forum will always have thoughtful rep point descriptions and links to valueable threads/posts. Also, members that have been around longer have gone through the times when there was no rep system or the rep system wasn't really used. Some members post in a lot more joke threads or just post a lot. They add solid value but their PPR is low compared to others because of post volume or being a bit post-whore-y.

Example: el mechanico - 73 PPR - He posts a lot of jokes and sometimes inflammatory statements. However, even in many of his jokes there is wisdom. Not only does he entertain people (which is some of his rep), but he drops knowledge. He has posted data sheets, he's verified a lot of automotive knowledge, etc...

Example: Suits - 33 PPR - If you consider how much of Suits' knowledge dropped and advice given circles around living in China, this is a very solid PPR in context.

Example: Cobra - 13 PPR - With only 1.5k posts and as a member for over 5 years, this is a very impressive PPR.

One last factor that I think is interesting is WHO reps someone. Do they get reps from high post count members who are in very high regard?

For example, if WestIndianArchie reps someone, that has a lot of weight versus a random lower rep member who hasn't been around as long (WIA has a PPR of 20 by the way and a current rep count of 284 which I believe is only bested by Roosh).

I encourage members to look into the posts / reasons that someone gets repped for and consider that strongly.

Finally, I would be remiss to not share my own PPR of ~54 which is actually closer to 50 than I thought. I consider myself a little post-whore-y as I like to post in the Donald Trump Thread and the Everything Else section a bit more than the game section (although I do try to post more in the Newbie Forum when I can). I could say that this is probably more because I've got a LTR for a couple years now, but I don't post that much in the Lifestyle or Deep Forum which would probably help me improve my life at a higher rate.

Post: #8

Post: #9

*Openly* discussing systems to root out trolls and saboteurs is not in the interests of the forum.

Post: #10

Solution above is needlessly complex. Reps should be for meet ups only.

Post: #11

(06-29-2018 11:44 PM)The Father Wrote:  Like the requirement that presidential candidates approve each campaign ad, there should be a .wav file at the end of every post that says "I'm Roosh V...and I approve this message"

Post: #12

Hilariously (given the op's silver membership), I remember "having silver membership" being pointed out about certain banned members who had flame-outs -- apparently it indicated a wishy-washiness (gold, or nothing! why do halfway?) which bled into the guy's posting persona.

We've seen a funny correlation over time with the rare, memorable silver medal, and strange posters.

No disrespect to OP who seems repped and vetted, but I remember cracking up when [member I can't remember] was about to be banned and someone pointed out "figures he has silver!" and broke it down -- what a hilarious small detail.

Post: #13

The Roosh V Forum Authenticity System Score (RVFASS) is the sum of these variables:

100 divided by Post Per Rep ratio

3.14159 multiplied by the number of confirmed RVF eskimo brothers you have

plus 30 if you defeated 30 white knights at the club

plus 2 if you and other RVF members were involved in a brawl over a hat

plus 20 if you and other RVF members were involved in a brawl over a hat but the hat didn't belong to any of you

minus 500 if you are a pathological liar or known as the Virgin of Belgrade

Post: #14

(06-29-2018 11:16 PM)WashedUPVet Wrote:  I wonder what the analytics say about banned members. There must be statistical significance that links them. Are there certain metrics that alert forum mods?

What metric do you think is the leading indicator for a troll?

What metric gives credence to authenticity?

As referenced by the past weeks bannings, it is only at the bitter end that neautral reps pile up. Surly there must be a leading positive or negative indicator to preclude direct knowledge of forum falsehood.

My simple thinking tells me that the total number of posts per (divided by) total number of positive reputation to indicates a members authenticity. A significantly low ratio would indicate troll.

Any thoughts? Can we put predictive analytics to forum use?

Popularity is rarely an indication of worth.