To go off of the initial goal of this thread and not directly respond to any one post besides the OP, I have come up with what I feel is the best approach for which sets go on-site and which don't.
I propose the following set of rules for C&C:
1. Viable and competitive sets are presented in analyses, regardless of if they are common or not.
This is the good stuff. I mention that they don't have to be common, either, and for good reason. If the set is viable and competitive, it'll become common after we put it on-site! Good examples of this are the Attacking Lead Machamp set and the ChestoRest Kingdra set. They're great, we all know that, but when they were submitted they weren't popular. They still got approved. This rule is the one that lets QC test new things and validate them on their own merits.
2. If a Pokemon is ineffective in the tier that it exists in, but banned from lower tiers, then the best set that it can perform must be presented in the analysis.
This rule applies to the Umbreon and Honchkrow of the world. They are not good Pokemon in the tier, but the best that they can do must be given and no more. This means that whatever it is deemed this thing can do better than all other things it can do gets the analysis. We need to give them competitive analyses, and we need to list
something, even if it's truly and thoroughly bad. We cannot have an empty analysis; that helps no one.
3. If a Pokemon is ineffective in the tier that it exists in, but competitively viable in the hands of a good player, then the best set that it can perform must be presented in the analysis.
This is for the Parasect and the Solrock of UU, pretty much. If these Pokemon are not so bad that they are non-competitive, like Spinda or Unown, then they must get the 1 set that is most effective in their analysis.
Consider that if they 'can' run some other sets, but they fall into rule 5 below, they can and should be mentioned in OC for that analysis but not written up! This way people can still reference the analysis for valuable information on bizarre things the Pokemon might do, but we are not endorsing that set as competitively viable in the metagame within which it exists. That difference is crucial to grasp.
Note that QC and only QC determines whether or not these Pokemon are 'competitively viable' or not.
4. Sets that are outclassed by other sets on the same Pokemon are not presented in analyses, even if they are common.
A good example of this would be how a mixed attacker Tyranitar pretty much outclasses a Choice Specs Tyranitar in every manner of speaking. We would not put a Choice Specs Tyranitar analysis on-site because it is awful compared to mixed Tyranitar that can lure in and eliminate certain threats. Another good example of this is Choice Band Heatran. Sure, it can Fire Fang something, but a Choice Specs Fire Blast is far more effective.
I qualify this rule with "even if it is common" to prevent us from listing poor sets that people are popularizing by spamming the ladder with their buddies all with the set. The quality of our analyses and the veracity that "These are the most competitive things this Pokemon can do" should not be put in jeopardy by trends on the ladders.
5. Sets that are totally outclassed by other sets on different Pokemon should not be presented in analyses, even if they are common.
So let's say that a set is totally outclassed. Choice Band Honchkrow versus Choice Band Tyranitar. Sure, you can probably run a Choice Band on Honchkrow, but it is outclassed in absolutely every single necessary way (defenses, offenses, typing, movepool, coverage) by Tyranitar. Sure Honchkrow is slightly faster, but it is not sufficient for the set to be regarded as even remotely on-par with Tyranitar. These sets should not be approved to be presented in analyses unless they are the sets forced into the analyses by rule 2 above.
6. Sets that are partially outclassed by other sets on different Pokemon may only be presented in analyses if Quality Control agrees that the differences between the two Pokemon are substantial enough that both are legitimately viable in different ways.
This is one of the stickiest and likely most debatable of all of my proposed rules. Let's consider the pioneering set that defines this rule: Curse Umbreon. We all know that it's outclassed by Curse Tyranitar, which has higher Special Defense, higher base Attack, better all around stats, a better ability, and better team support options. However, Tyranitar is weak to Water-, Grass-, and Steel-type attacks, with a 4x Fighting-type weakness in there; Umbreon doesn't have that. Umbreon also has access to Wish, which doesn't force it to use Rest and Sleep Talk to heal itself. Of course, this then makes Umbreon vulnerable to poison and burn! As you can see, Curse Umbreon is partially outclassed by Tyranitar. As a competitive player, you would probably use Curse Tyranitar—not to say that Curse Umbreon is necessarily bad or non-viable, however!
In this case, I place that it is up to QC to determine whether this set is useful or not. They weigh the options, both sides, and they decide whether this set is 'good enough' to get an analysis. Then, once QC decides, we stick with it through thick and thin. QC decided that Curse Umbreon was terrible and poor enough to not merit a set. That means that if someone submitted Curse Umbreon to QC now, it'd get rejected because QC made their decision and nothing about the game has changed to improve Curse Umbreon. In this case, you would say that QC has determined that Curse Umbreon is not viable enough for a set, despite that people use it. These almost-sets get described in OC, as usual.
So you see, some things that are outclassed
can and
will get analyses. A good example of one that succeeded was
Espeon (Substitute + 3 Attacks), which is partially outclassed by Alakazam. It was decided by QC that it was viable enough for a set, so it got a set. It's that simple; it's all up to QC.
Note that QC generally decides upon the degree of outclassing as well. They decide if "X is totally outclassed" and thus meets rule 5 above or if it's "partially outclassed" and they must test it to decide if it should get an analysis or not. This and the last rule are just formal representations of how QC functions.
7. Poor performing sets are not presented in analyses, even if they are common.
This is pretty straight-forward. We do not list bad sets in analyses even if they are common. I can think of one good example of a very common set for a competitive Pokemon that is very popular, but that the set is so bad that it really does not deserve an analysis: Encore Lead Machamp. Some would say "Hey, that's outclassed by the Attacking Lead Machamp!", but I would say that they do slightly different things. This set was actually remotely common, believe it or not; it was used on 8.7% of OU Machamp back in June! This set, however, was considered not viable enough by QC, and was rejected from Machamp's update at the time. It is important that we don't just put anything popular on-site and that we qualify what is good and what is not. Just because you might run into Encore Machamp in-game does not mean that we need to list it as a set! We have OC for those bizarre and weird things people might run that aren't really that good.
8. If a Pokemon is so bad in all tiers that using it in battle is attributable to not playing the game competitively, then that Pokemon is a candidate for a joke analysis.
These are the Spinda, Luvdisc, and Unown of the world. No matter what anyone says, using these Pokemon is a bad idea. The Pokemon are bad, the things they do are laughable at best, and you are not playing competitively if you are using these Pokemon. When you can say that, you can write a joke analysis for that Pokemon.
Note that I mention
all tiers. This means that BL Pokemon cannot get joke analyses just because they are bad in OU. They are BL, meaning they were too good for UU, which prevents them from getting joke analyses. I think this rule covers pretty much any time we would ever want to make a joke analysis.
A joke analysis may be written for any tier. This means that Spinda might get written for UU, while Farfetch'd might get written for OU and Magikarp might get written for Uber. That's fine. Because these are joke analyses and the Pokemon is non-competitive anywhere, the joke analysis can be written for any tier.
--------
And those rules cover pretty much everything. They delegate exactly which sets go on-site, exactly which sets
don't go on-site, and they allow us to accommodate for bad Pokemon and Pokemon that deserve joke analyses. If you guys have comments or criticism of my proposal, please feel free to lay it on thick; I'm open for discussion on the matter.