This is request for comment from the community regarding the following question:
in SAT resolution proofs, are all DAGs possible?
Is the question in its current form suitable for the site?
If it is not, was closing it by a community moderator justified? Or was it excessive use of moderator powers?
If the question is not suitable in its current form and moderator action was excessive, how should we deal with it?
My view is simply this: moderator powers should be used lightly, and only when the community mechanisms are insufficient to correct a situation. In the case of the above question, here's what my personal (i.e non-moderator) take is:
So the moderator role here is to help suggest modifications (especially since there appears to be a germ of substance in the question) and encourage the OP to make the changes (which I will note the OP has not done). As an individual, the moderator is free to downvote the post in its current form.
But as a community moderator, I prefer to do nothing rather than do something, and in this case I have to ask what the consequences are of doing nothing. Either the OP will fix the post or not. If not, it's quite likely that other members of the community will downvote the post and/or vote to close it. Or they might attempt to answer the question.
The "worst case scenario" from my point of view might be that an ill-formed question sticks around for a while. But what is the downside of this ? The OP regularly posts ill-formed questions, and they get hammered (about 2/3 of the questions posted have a net negative score). Clearly the OP is not learning anything from the community view on their question formulation, but I don't see any evidence that these questions are affecting others, or reducing the quality of the site.
Either way, there are sufficiently many options available to the community at large that I (as moderator) don't need to step in. And that's my underlying philosophy.
Regarding the question of suitability for this site:
I do not believe the question in its present form is well worded. I am ambivalent about its appropriateness. Ill formed questions appear occasionally on this site. I myself have posted a question that was sub-optimal for the site, and was glad that the community engaged me and pointed out why my question was inappropriate rather than closing it. I believe the community has tolerated and helped to improve vague, unclear, but well intentioned questions. Even if intent is unclear, we have helped improve unclear questions. I feel that this question can easily be improved to a clear, technical question and ideally, I would prefer this route.
I am aware that the OP is very unlikely to incorporate feedback from the community and improve the question. I agree that this is a problem and sympathise with the frustration of both the site users, and particularly moderators who must confront such issues.
Nevertheless, I would prefer to be part of a relaxed community. This is a personal preference. I am averse to forums that feature flame wars, heavy moderation and contentious discussions that essentially amount to shouting contests for intellectual bullies. I would like this site to have an image that is welcoming and not intimidating to give computer scientists the confidence that they can ask questions and be listened to rather than fear admonishment if their query does not satisfy some rigid set of rules. Let me clarify that I am currently fairly happy with this site and I do not believe any of the concerns I am stating currently apply.
Closing a question so quickly, especially when it is not obvious why (to me as a user, viewing the question) feels strong to me. I would have preferred for the community to go through the potentially futile process of suggesting improvements and then either expressing our displeasure with our votes, or ignoring the question beyond a point. To Kaveh, I would like to further add that this is a preference of mine, not a strong opinion. My desire for a relaxed environment and concern that closing a few questions quickly will cause us to descend into a dictatorship is less realistic than your concern that the site will attract cranks, but neither concern seems empirically justified. In the latter case, you may have more information as a moderator and if people are indeed being driven away by cranks I would agree that we should behave proactively. This concern does not seem to apply to this question because the user in question has been on the site for a while and the community has survived.
Regarding improving the question itself.
Here is one potential wording.
This question is about the combinatorial structure of resolution proofs.
Let $Prop$ be a set of atomic propositions. A literal is an atomic proposition or its negation. A clause is a disjunction of literals. Let $Clause$ be the set of clauses over $Prop$.
The resolvent of two clauses $C \lor p$ and $\neg p \lor D$ with pivot $p$ is the clause $C \lor D$.
A resolution proof is a finite, labelled DAG $P = (V, E, piv, clause)$, where
A clause $C$ is derived by resolution from a set of clauses $F$ if there exists a resolution proof $P$ in which $clause(v)$ is $C$ for some vertex $v$ and every leaf vertex on a path to $v$ is labelled with a clause from $F$. A resolution refutation is a derivation of the empty clause.
The original questions can be reformulated as below.
- For every DAG $G$, does there exist a resolution proof $P$ whose underlying graph is isomorphic to $G$?
- For every DAG $G$, does there exist a resolution refutation $P$ whose underlying graph is isomorphic to $G$?
- Does there exist an ordering on clauses such that the set of clauses labelling leaves is minimal with respect to that ordering?
Let me point out that I believe (1) and (2) are directly in the question and (3) includes a minor interpretation on my part. I would further extend (3) as below.
3'. If $E$ and $F$ are sets of clauses, define the order $E \le F$ if for every clause $C$ in $E$ there exists a clause $D$ in $F$ such that every literal in $C$ occurs in $D$. For every DAG $G$, does there exist a proof $P$ such that the clauses labelling leaves are minimal with respect to this order. Is there a unique minimum labelling?
My reformulation does leave open whether the question is research level. I believe the reformulated question is not difficult to answer, but I find it appropriate for this site. The proof required would be a background exercise for approaching the type of work in The complexity of resolution refinements.
Here is my explanation for closing the question:
The OP states clearly in the first paragraph that his intention is to share his ideas and states that he does not expect an answer for the question he is asking. That I think clearly means that the question in the post is not a real one and the post is not suitable in its current form and should be closed as such.
If those extra claims and statements are removed and the post becomes a real question it can be reopened. I think the post is clearly not suitable in its current form and should get closed until issues with it are resolved at which point it can be reopened.
Unilaterally closing it might be excessive however as we know closing is not permanent and can be reserved and if this podt is a clearly unsuitable in its current form then moderator intervention is not too excessive.
Now I could have simply commented and asked OP to modify the question to become suitable, however in my experience in dealing with this user he is not receptive to such comments so I felt it is unlikely that without a closure the user will fix the issues, particularly since from the first paragraph it seems that expressing those ideas and claims are the main intention behind the post. It is often possible to notice the difference in attitude and language of a post really asking a question from a post which is intended for expressing/soliciting/sharing opinions/claims/ideas even when it is not as explicit as in this question.
More generally, I think it is an abuse of cstheory to promote one's opinions, particularly on topics where one is not an expert on, in fact it looks to boarder a post by a crank which are off-topic for cstheory.
I don't think this kind of posts and such expression of claims and ideas are suitable for cstheory, a user can express them somewhere else and people interested in the opinions of the user can read them there. One should abuse the fact that experts and researchers read cstheory to promote and advertise personal opinions and claims.
I think cstheory is not and should not become a medium for non-experts to express such claims and ideas, that is not a goal of the site and frequency of such posts would hinder the usefulness of the site for its target community. Right now we have a single user who insists on abusing the site (comments, answers, questions, and even cstheory associated chat-room) in this way so the effect might not be very noticeable, however it will be if such posts become acceptable and frequent and will cause the site to suffer from the problems that tcs Usenet groups suffered because of such posts.
This is my perspective. If you have different opinions and perspectives about this question or the general issue please share them with me and others. As a moderator my intention is to follow as much as I can any view or policy we arrive at as the community of cstheory.
