Board Game Arena

Ranif88 wrote: A green/red ratio metric PLUS a karmic element of player presence remains a better approach to addressing the concerns of most of the players here.

I would finally mention that the system approach continues to not address a primary negative which is any redress towards any red thumbs awarded to a player without courtesy of a text note (the "drive by"). As reds still contribute negatively in the new karma system, an address system to allow counter or appeal should exist with the easiest being who gave what to whom. This continues to be a sore point and often muddled and lost in these discussions about "thumbs" or "karma".

For another way to make the above points, what the "karma" system is attempting to do (in part) is to create a "sentiment analysis system" or SAS which is blending numerous metrics on player presence and effect to create a "brand quality" assessment for other users/consumers.

This is a very common effort within most corporate brand management efforts but must include the two elements noted in my quote above, again:
1. Inclusion of positive considerations (the green thumb count and/or ratio to reds) in weightings
2. Addressing of false positives (the ability to counter/remote/request reason for received red thumbs)

A simple sentiment system (good/bad or green/red evaluations) fails over time primarily due to #2 and has been the case with past use here.
A presence-focused system (time/exits/play statistics) dilutes opinion impressions gathered from the masses and becomes limited in use.
An opinion input that is only negative (including only red thumbs in karma vs a green/red ratio use) skews to false positives and is seen now here.

There are numerous examples of how to develop a more balanced evaluation metric AND to include a method for redressing negative metrics.
Until these two elements are included the solution will remain only partially solved.

(Note: While this is mentioned in earlier red thumb/ELO/algorithm discussions, several recent questions on what is an ELO vs karma spurred this:)

Several ongoing comments and questions appear to confuse the idea of what BGA is developing with a player ranking and evaluation metric.
The "karma" is actually a bit of a blend of player presence + implied ability + subjective basis of how one plays and good sportsmanship.
This is a challenging metric and previous comments suggest modifications such as reducing the presence elements and increasing green thumb weight.

Several systems including simple win/loss summaries exist for establishing a player's relative skill. Additionally, presence metrics (time on site, finish game, exits of game, etc) exist to incorporate. Most of these are fairly quantitative and assembled using methods such as ELO and/or site selected player metrics.

The player quality metric however (see numerous previous entries) remains the challenge, as well does a fully combined metric of the three elements.

Any quality metric requires a positive and negative evaluation to be included. The use of a normalized set of values that are negated by implied actions and reset only by time is a relatively incomplete solution. The inclusion of accrued positive metrics (greens) should be used to either de-weight accrued negatives or (via a green/red ratio) scale any decline or reset/become whole algorithm to more correctly assess a player's effect with others.

So, just a reminder that there are three blended elements required for a true full player evaluation metric (PEM), and it's a challenge.