Follow Slashdot stories on Twitter


Forgot your password?

Journal fermion's Journal: on moderation

"A right of free correspondence between citizen and citizen on their joint interests, whether public or private and under whatsoever laws these interests arise (to wit: of the State, of Congress, of France, Spain, or Turkey), is a natural right; it is not the gift of any municipal law, either of England, or Virginia, or of Congress, but in common with all other natural rights, it is one of the objects for the protection of which society is formed and municipal laws established." --Thomas Jefferson to James Monroe, 1797.

The following is some thoughts on altering the /. moderation system. I do not believe any of these ideas are novel or unique, but the application of these ideas may help with issues such as scalability and the promotion of long term discussions within an article posting. The main concepts in this modified moderation proposal are moving to labels instead of numbers for moderation, continuously variable starting moderation points based on karma, and non linear moderation scoring.

First, I believe /. should move to labels for moderation in leu of the current numbers. These labels could be 'horrible', 'unrated', 'default', 'fine', 'great', 'terrific', and 'super' , respectively corresponding to moderation of -1 to 5. The labels will need to be tweaked with the idea that labels will reflect the status of a comment. For instance, a current '1' comment would be unrated because this is the default posting for a logged in user. I do not use the karma labels to avoid confusion.

Underlying these labels would be an expanded, and scalable counter. To determine the range of the counter, one would set the increment for a category. If we select the increment to be 20, for instance, the counter would go from 0 to 101. A comment at 0 is 'bad', a comment at 101 is 'terrific', and everything else is evenly divided into groups of 20. For instance, the structure might look like
horrible 0
unrated 1-20
default 21-40
fine 41-60
great 61-80
terrif 81-100
super 101

In the discussion below I will assume that moderation counter starts at zero. I will also refer to the increment as defined above. As an aside, is feasible to make the increments nonlinear, and there are reasons to do this, but I believe the such functionally can be incorporated into the moderation procedure.

The next issue is the starting moderation based on Karma. Currently logged in users start at 1 and eventually get a point. If the user is subsequently very bad, they will lose the point. This system is effective, but imprecise. I feel it gives a new user excessive benefits, delays additional benefits until the user reaches top status, and does not quickly punish bad behavior. My suggestion is that anonymous users start at the low end of unrated, which in general is 1. All registered users would have their starting moderation calculated based on karma, as described below.

If we assume that the underlying karma count sets a neutral karma to zero and goes positive and negative with increasing and decreasing karma, we might calculate default moderation as

default_moderation=offset + karma*scale.

This equation has two variable. The first is offset, which is where we give logged in users a boost over anonymous users. For instance, if the offset is equal to the increment, a new logged in user will still start with 'unrated' comments, but a single moderation would guarantee the comment would become 'default'. The other variable is scale. To be consistent with the current system, this should cause a user with maximum karma to in the 'great' rating. This could be done by having each label change in karma add half the increment. Therefore, a karma of 'positive' and 'good' would make the moderation increasingly 'fine', while a karma of 'excellent' would have moderation in the middle of great. The application of this is that the default moderation is a continuous function based on karma, which we assume also changes continuously with user behavior.

Moderation itself should not change from the point of view of the moderator. The moderator will still choose a label and moderate. However, /. will now have fine control over the points awarded. For instance, each moderation might only award 3/4 of an increment. Or perhaps we want to encourage users to look for new good comments rather than just continue to moderated existing highly moderating comments, so we might only award 1/3 of an increment to any comment that is already 'terrific' Maybe we see that most comments posted in the first 5 minutes of an article are useless, so the moderators on those articles on get to award 1/3 of an increment until the comment reaches 'fine' status. Maybe we want to encourage moderators to look at new comments, so a moderator will only be able to award 1/4 increments to any comments that is past a certain time threshold. It may be decided that a funny comment is less valuable than an insightful comment, or an overrated tag is less valuable than a insightful tag. In summary, because moderators will still only have a certain number of moderations, regardless of the actual points awarded, the moderator can be discouraged from certain actions by making those actions less forceful.

There are several possible pitfalls in this proposal. First, the moderation of a comment will not necessarily lead to the change in moderation status for an article. This may confuse moderators. Second, communicating the variable moderation points may be prohibitively difficult. If such information is not communicated to user at the time of moderation, this proposal may not be an improvement over the current system. Third, an increased amount of computation may be necessary to display as moderation page. This increased load on the web-server may prove excessive.

A final comment on the '3 day limit' for moderation. I believe this limit is excessive and does not encourage the important activity of moderation. My suggestion is based on when the user logs into /. First, if a user does not log into /. within 24 hours of being awarded moderation points, the moderation points go back into the pool. If the user logs in within the 24 period, the user will then have 24 hours to use the points. If the points are not used in that period, they go back into the pool.

This discussion has been archived. No new comments can be posted.

on moderation

Comments Filter:

"The number of Unix installations has grown to 10, with more expected." -- The Unix Programmer's Manual, 2nd Edition, June, 1972