Meritocracy: A Fierce and Fair Meritocracy
LogiLogi combines openness with quality control. It does this by allowing logis to be rated, and then showing the best rated logis first. In addition, voting-power varies between authors depending on how well their own writings were rated previously. Authors can thus gain `standing' and `influence' through their work. This makes LogiLogi not just a democracy, but a peer-reviewed meritocracy, quite comparable to what we, according to Bruno Latours philosophy of science, encounter in the various structures surrounding journals.
The ratings in LogiLogi are essentially grades, given by visitors and other authors. With each vote a score can be given on a scale of -2 to 5. The average of these scores forms the rating of the logi. These averages are weighted averages, because voting-powers can vary. Anonymous users and people with accounts begin with 0.1 resp. 1.0 voting power. This is their base power. In addition, people with an account can receive extra voting powers (so called honours powers) for each of their logis which are positively rated.
The formula for awarding the extra voting-powers based on the rating, is currently quite simple. It is calculated as follows:
$rating ^ 2 * 0.05$
So it is 0.05 percent of the square of the rating (0.05, 0.2, 0.45, 0.8, 1.25 for scores 1 to 5). The rating is - as noted - the weighted average of all votes given to the logi. So honours powers are not given for every vote, only for their standing weighted average. And they are given in realtime, so when the scores given in new votes are lower than this average, the extra voting-power received from the rating can be reduced again.
Now for the calculation of ratings: the rating of a logi is the weighted average of all the scores it received through votes. It, besides having a score between -2 and 5 (let's call it its height), also has a weight. Initially this weight would be equal to the powers of all votes it received. Thus for example a new vote of 5, with power 1 added to a current rating of 1 with weight 3, results in a new rating of 2, with weight 4. Now of course this would lead to the entrenchment of ratings over time: it would make ratings ever harder to change by subsequent votes.
To fix this problem - and to give new votes a chance - the weight of the rating is decreased each night with a fraction in such a way as to result in a half-life of one week. So at the end of the week the weight of the rating is half as big as it was at the beginning of the week. If no new votes come in the height of the rating remains as it is (its weight just drops), but if they do come in, they can influence the rating more easily because of its lesser weight. It should be noted that no half-life applies to the voting-power of authors. Their voting-power does not change as long as their logis ratings are not voted up or down (because honours powers are based on the height of ratings, not their weight).