Moderation

Is it really about "beta"?

It is claimed the reason of migrating out of slashdot is the "beta". But is it really so? Let's try to do some speculations...

"Beta" is said to be made to fit tastes of the younger audience, flowing recently en masse to Slashdot - in fact, stats show, that Slashdot is currently accesses mosltly from... schools. The same audience is, though, often scorned upon by the "low uids" as the reason of progressively "destroying the old Slashdot". Now, we could conclude, that the deeper reason for both beta and discussion quality are really kids. But it seems, it is still not so simple.

Let us consider an example. So, (i) a bunch of teenagers mods ups some old joke, as things like that still amuse and educate them. Now, a low uid hates that. But, despite that, (ii) the younger and the older Slashdot might still share some common interests - be it some comment insightfull for all, so a single site might make sense. How to agree that? Would some special sort of moderation be able to please both of the discussed groups?

Attach your proposals below.

A local moderation

The moderation and metamoderation work globally now - it is one of the factors which decide, how many mod points a user gets. What about making it also local? A moderating or metamoderating user would express his/her preferences in this way. An example: it turns out, the user X likes comments of "low uids", but dislikes comments frequently moderated by teenagers? So the system increases that user's "experience", what translates to mods of experienced users being somewhat more visible by this particular user, as opposed to mods of the school crowd. In other words, there would not be a single score, seen by all -- a user by moderating would tune, or bias the scoring according to his/her needs. Meanwhile, a kid might still admire, and mod up with "novice points", old jokes, but also high-quality school-grade educational comments.

This might be made even more complex -- a dynamic cluster analysis might find out by itself groups of users with common tastes, and somewhat tune the scoring to their likes. And, if the said kid writes a quality comment, modded up by experienced users, it would in turn move the kid a bit in the direction of a cluster of "low uids".

Surely, a careful tuning of the system might be needed. To alleviate that a bit and to make the local moderation opt-in, every user might be given a choice of his/her sweet spot between a "flat" and an "adaptable" scoring.

Example implementation

Let us discuss a simple, yet intuitive variant of the above. This variant does not require any clustering methods, and most of this variant's complexities would be handled in a way invisible to the user, but still, many people might prefer a simpler moderation scheme. This is why a plain moderation might still be the default, and the system below would then be an opt-in, for tweakers that do not like the "flat" scoring. The variant discussed may easily be used together with a plain moderation -- the latter would just make the resulting scoring more flat.

The quality of user contributions would consist not of a single value, but of three values: novice, neutral, veteran. A user might thus be e.g. valued a lot by teens, thanks to his school--level yet very educational contributions, or conversely, a user might be a long--bearded Unix man that is not into teaching kids at all, but loves deep, clever references to the geek culture. More examples:

Novice	Neutral	Veteran	User type
0%	0%	0%	Tabula rasa, or AC, rating 0%
30%	10%	10%	Obvious or shallow, rating of 50% - might be valuable for novices, though
70%	10%	20%	Possibly a skilled teacher, high educational value for kids, but rare contributions directed at educated users
20%	60%	20%	A "middle" user, but a respectable total rating of 100%
-10%	0%	90%	A long-bearded Unix woman, likes some trolling, too; rarely useful for novices but low-uids love him
40%	0%	90%	A long-bearded Unix man, but occasionally likes to teach the younger audience, too. This universality gives him a total rating of 130%

Depending on the "NNV" ratings, a user with a total rating being high enough would be given various types of mod points. For example, a "middle" user might see:

You have 5 modpoints! 1 novice, 2 neutral, 2 veteran

How would the system boot up? The types of modpoint would initially be distributed equally for everyone with a high total rating, like a cautious yet type-complete "2 novice, 2 neutral, 1 veteran". But eventually moderation would change ratings: a user whose comments are modded mostly novice would have its novice rating increased etc. This would iteratively lead to novices having mostly novice modpoints at disposal, veterans having mostly veteran modpoints to use etc.

But, what about a veteran that spots a school--grade yet quality educational comment? It would be a nonsense to mod it up with his veteran point. In fact, it would hijack the system a bit. The solution to this would be, that a "more experienced" mod point might be transformed into a "less experienced" one, depending on a user's decision. Thus, a user might transform his "veteran" points into "neutral" or "novice" ones, and his "neutral" points into "novice", if he wishes so.

Occasionally, even a veteran would get a "novice" mod point, what would be a hint to him to award good school teachers, too. He would need to browse comments at a low threshold for this, though, as he would generally see school--level educational comments scored low, unless he would choose to see a "flat" scoring, as discussed in the previous section.

A special caution should be given to certain artifacts of the discussed system. For example, let us consider users, that rarely have "experienced" modpoints at disposal. What e.g. about a user with ratings 80%:10%:5% that wants to mod up a nerdy comment, but has only "novice" modpoints? First of all, he should not do it, because he does not have a proper type of a modpoint and would thus skew the system. Yet if he does it anyway, he should eventually be catched by metamoderation.