After a while of thinking about it and in reflection to the comments, I'm updating this with the following proposal:
Remove the passage from the help centre, as it is a policy that is not enforced.
Keep answers that try to promote unconventional theories around and let the score indicate that they are "bad". (Delete if they do not try to actually answer the question, NAA - not an actual answer, or link-only, spam, etc..)
My 2 yen:
Should the existing policy for questions be extended to answers?
No, not really, that's what down votes are for. I actually needed to think about this quite a bit and I'm somewhat undecided still, for reasons I try to include at the end.
Are the definitions of mainstream chemistry and reputable journal clear enough?
I don't think they are. But I also don't think they matter much. (Again, see the end of the post.) At what point is a journal reputable? When does a new idea become mainstream? It's usually when you get other scientists to understand it and support it. On SE we usually signal that with up and down votes.
I understand that we are not here to invent new things, and rather building a library of existing knowledge, easily accessible and in a more generalised language. We have the occasional question asking about a proposed synthesis
Should the guideline be limited to unconventional theories or be generalized for any answers that are considered extremely wrong?
Where is the line? Who decides that an answer is extremely wrong?
One of the worst answers that I would happy nuke from orbit happens to be one of the highest scoring answers on the site. In addition it actually does promote some strange new form of notation. (If you don't know what I am talking about, look here $*% http://chemistry.stackexchange.com/a/461$ in the source code to find a link; I just want to prevent too much meta-effect.)
How would we justify deleting such answers, who decides that? The site is about content and sometimes wrong content is equally important as right (or accepted) content. It just needs the proper indication as such and that is where the voting system comes in.
Should such answers be deleted, or wouldn’t it be more to the point if they stay – with downvotes that clearly indicate the assessment of the community?
They need to stay and be voted to oblivion for reasons stated above.
Would users prefer to delete answers because downvotes on answers remove 1 reputation from the voter?
I am very much against deleting wrong answers/ answers with non-mainstream chemistry, like I outlined above. Additionally I think that one reputation penalty is not much. If you are invested in this page, within a month or two you'll be in the realm of "It hardly matters".
Lastly: Why do we need such a policy in the first place?
I can't see how voting wouldn't take care of that by itself. If a question is non-mainstream, there will be enough users giving it a negative score and probably a close vote, basically ensuring that the question will get deleted in due time.
If an answer is unconventional it will be indicated by a desperate negative score, the link in this question is case in point. At some point it becomes barely readable, it is clearly recognisable as not helpful.
As I argued before, I want these answers to stay, so I don't see a policy enabling us to delete them doing any good - and as a moderator I do not want to make this kind of decision.
Pitches for your own personal theories or work: We deal with mainstream chemistry here. Anything that couldn’t be published in a reputable journal is probably not appropriate at this site.
I don't read that as a policy in the first place. It is a gentle reminder that these questions won't get you very far and that you better not waste your time putting them up in the first place. This is especially indicated by the fuzzy choice of words: "probably not appropriate". It does not say it is off-topic in general.
I personally find such a policy superfluous, if not counterproductive. I rather debunk or myth-bust a question dealing with non-mainstream chemistry, than deleting it, leaving everybody in the unclear.
My parents said to me "there are no dumb questions", and while I not completely agree with that, I understand that a question means a stating an interest or misconception, a quest for knowledge and clarification. If a question is based on a wrong premise it needs to be clarified, probably even more carefully answered so that hoaxes and pseudo-science doesn't spread.
After Lastly: I see where such a policy could be coming from. It's from the early days of chemistry where the general scope of the site needed to still be formed. Removing such questions to not have them around for examples of how to ask questions here was important. It isn't anymore. We have a fairly well defined scope, a broad user base, and such questions will be dealt with the built-in mechanisms of the site.
In rare cases we would probably need a handle to delete a non-mainstream question. In most of these times I guess we already have another mechanism in place to deal with that. In the last instance we have specific-question to discuss this here.
I don't argue for removing the passage from the help-centre, as a discouragement it is probably effective enough and it's wording actually does not prevent any question per se. I just wouldn't interpret it as policy and certainly do not want to extend it towards answers.