Scott's definition of necessity is that something be "on topic, and not only contributes something to the discussion but contributes more to the discussion than it’s likely to take away through starting a fight". I think that dovetails pretty well with "adds value to the discussion" from my post.
Here are the four definitions of “necessary” that we have so far. Scott’s two:
Necessary in that it’s on topic, and not only contributes something to the discussion but contributes more to the discussion than it’s likely to take away through starting a fight.
And it had better be necessary, in that you are quashing a false opinion which is doing real damage and which is so persistent that you don’t think any more measured refutation would be effective.
And your two:
Does it add value to a discussion?
Did the poster think they were ethically or practically obligated to make this post, or is it just empty cruelty or virtue signaling?
These are, again, four different definitions!
Now, you may not think they’re different. But I do. And that’s the point, right? Let’s say I’m a DSL member and I’m writing a post. I think: “Is my post necessary? Well, am I ethically obligated to make this post? I think so, yeah. It’s definitely not just empty cruelty or virtue signaling! And I know it’s true, so… <hits Post button>”
Then someone complains about my post, for whatever reason. You come along and give me a strike. “Not necessary, not kind,” you say. I demand an explanation; you oblige: “Your post didn’t add value to the discussion.”
You see? I used one of your definitions of “necessary”. You used a different one of your definitions of “necessary”. Your strike came as a total surprise to me, because I thought I was following the rules, but you disagreed.
And this is a scenario that can play out with just one moderator—but we have five mods! (And, of course, I used two of your definitions in my example; if we bring in Scott’s definitions as well, the problem gets much, much worse.) This sort of massive ambiguity is guaranteed to lead to total confusion about what the heck the rules actually are.
That is terrible for a forum.
The “kind” criterion isn’t much better; here, too, your definition is not the same as Scott’s. So which do we use? Do you suppose someone might disagree with you on what constitutes “kindness”? I think it’s guaranteed that someone will. (For example, in a currently ongoing thread, we have people saying that “love” demands that you kill your neighbor for being gay. Think about that! How, exactly, do you propose to use “kindness” as anything even approaching a workable, mostly-objective standard of evaluation, given the non-trivial prevalence of such views among the commentariat?)
IMO, there's no non-sophistic way in which "kindness" can be tortured to justify murder (including state-sanctioned murder, as [REDACTED] is proposing).
I agree with you on the object level, but what does that actually get us? For instance, [REDACTED] clearly does not consider what he advocates to constitute murder, and as far as I can tell, that is a true and genuine belief on his part, not just sophistry. But my point was just that if you ask people to be “kind”, you’re not going to get anything even remotely resembling a consensus on what that means in actual practice.
Basically, the rules we want are those that can most effectively be used as binary predicates: does any given post violate any given rule, yes or no? In order for the moderators to apply the rule consistently, the predicate must:
1. Return the same answer for a given post if a moderator applies it today or next week (the most trivial criterion, but we’ve already seen failures to meet it)
2. Return the same answer for a given post no matter which moderator applies it
3. Return the same answer for two different posts with a probability proportional to the similarity between the posts (along relevant dimensions), when applied either by the same moderator or two different moderators
And in order for the members to follow the rule, the predicate must:
4. Return the same answer for a given post when a member applies it as when a moderator applies it (this assumes criteria #1 and #2 are met)
4a. Return the same answer for a given post when one member applies it as when another member does (implied by #4 but worth stating explicitly)
5. Return the same answer for two different posts with a probability proportional to the similarity between the posts (along relevant dimensions), when applied either by the same member or two different member (analogue of #3)
In other words, the predicate that a rule is based on needs to show inter-rater, intra-rater, and test-retest reliability, when applied by both moderators and members; and its answers have to map onto some sort of sensible space of post variation. These are extremely basic requirements; without them, a rule just doesn’t work. It doesn’t do anything.
Note what I’m not demanding, here:
Equality of treatment—that the predicate return the same answer for the same post made by two different members.
Construct validity—that the answers the predicate gives us properly represent what the predicate is claimed to describe (i.e., is it actually “kindness” that we’re measuring, as commonly understood, or something else?).
These may be desirable criteria, or they may not. But we can punt on that question. What we can’t punt on is the very basics, those criteria I listed above. Again: if we don’t have those things, the rule just does not do anything except cause confusion and chaos.
