There are a few things that need to be considered here.
The first is that questions are easier to ask than answers, and answers have more innate value than questions. Everybody has questions, and anybody can ask, but answering a question correctly requires knowledge. As such losing a questioner is not going to be as much of a problem as losing an answerer. The aim of the voting system as it pertains to questions is to help guide our users to the better questions so that they do not get bored of the website. Jeff Atwood mentions that the website has a preference to answerers in Are Some Questions Too Simple, and goes into greater detail regarding the principle in Optimizing for Sand, Not Pearls. Things were equal, but he rebalanced them to be reflective of the value.
The second thing is that questions and answers are motivated differently. Questions are inspired by a desire for knowledge, and receiving a satisfactory answer is reward enough irrespective of what other people think. Answers on the other hand are motivated by the desire to be helpful, and if people indicate that you have done more harm than good, then they might stop to think why should they even bother trying to help? Voting against answers has a greater prospective harm against user participation than voting against questions, so there are a number of protections against it that Jeff mentioned in The Value of Downvoting. Jeff also notes how indifference is also a valuable data point there, so the penalty against voting down also discourages voting against answers simply because they are not great.
The third thing is I doubt that closure is really supposed to be the mechanism that separates the good questions from the bad ones. Closure reasons are usually implemented because certain categories of questions are so problematic that they are likely to receive problematic answers. Also there is a penalty for everybody involved, in that answers can not be submitted to a closed question, and we can not simply ask a new question free of defect because the closed one remains a potential duplicate target. The end result does not only punish the questioner in a much more real way than reputation deductions ever could, by possibly denying them the best answer, but our entire community. Any insightful person who might wish to answer the question can not do so, and our general readership, including thousands of non-contributors, are unable to benefit from it. On Stack Exchange, questioners are not considered the sole beneficiaries of a question, or even the most important ones, so closure needs to be used in a way.
Having questions in a permanently closed state is perhaps the very worst outcome for our website, since it is letting one rotten apple spoil the whole bunch. The system was designed so that closed questions would either be eventually deleted or reopened, as noted by Gracenote in Off-topic questions should be deleted rather than closed
(top voted answer at general S.E. meta; supported by Shog9) and Why is this question not being deleted? (Security Information meta), and while she does mention the term "off-topic", that is term unlikely to carry its usual meaning. Every closable question is considered off-topic in the context of Stack-Exchange, even if it would be within the scope of the subject matter. Evidence to this is found in the closing and flagging interfaces, which have "off-topic" submenus, and to a lesser extent in the subjectivity guidelines which also show that "opinion based" questions are subject to possible deletion on that basis alone. As a matter of fact, the help center page that mentions this presently has a U.R.L. that titles the page as "dont-ask" [sic].
Considering this, if we were to introduce a new deduction, I would prefer for it to be imposed on close voting until the question is either reopened, in order to encourage commentary, or deleted, in order to encourage clean-up that permits new questioners to try and ask again until somebody gets it right, if possible. Yes, closing questions is more effective at eliminating bad questions, but in the same way that a bomb is more effective at causing destruction than a handgun: The greater potential for unintended collateral damage is very, very large and we should probably have a preference for the more selective weapon, that only targets the questioner with negative feedback, rather than denying others the opportunity to teach and learn. I would greatly prefer if we reserve closure for questions which must be fixed in order to be productively answered in accordance with our goal of sorting the best answers to the top.
The final thing is that there is a distinct difference between our corner of Stack Exchange, and the network as a whole. We have a general reference type closure reason, but that closure reason is not a network wide closure reason, and it never was. Other websites which opt against having such a closure reason may still need a way to deal with under-researched questions, and may have a greater need to vote against them with impunity. Jeff thought we set a bad example for what a Stack Exchange website should be, we kept ours as a custom closure reason due to special need (dictionary questions, mostly), and that merely voting against answers is the way other websites are meant to deal with poorly researched questions as evidenced by the tooltip. The first point is important because we probably won't persuade the rest of the network to implement a general reference closure reason, and the latter point is important because it demonstrates a difference in attitude towards closure (we're one of the most agressive sites to my understanding) and how a difference in closure reasons changes the dynamics of voting:
Without a relevant closure reason, other websites are expected to vote against under-researched questions as the primary method of quality control in that regard, as the tooltips for voting for and against questions explicitly mention research efforts on every website and for those rare occasions when a question is really bad.
The way we do it has merits for very well documented sciences, where persuasive answers from authoritative sources like dictionaries are immediately on-hand, since the risk of a useless or outright harmful answer can outweigh the prospective benefit of a rare gem of an answer. We want to sort the best answers to the top after-all, but that might not happen if we merely repeat a trusted, yet incorrect or under-detailed source. However, at the same time I suspect that the current state of affairs, where closed questions are left to simply languish lowers the average site quality since it prematurely forecloses upon good answers, and for other subject matter that is less well documented that closure reason is a pretty bad fit. When misapplied, our system can be counterproductive to Stack Exchange's general goals of making the internet a better place by creating a comprehensive reference work indexed by google. Having a general reference closure reason makes sense in our case because language, and especially the English Language is already a very well documented science, and we merely aim to be more of a supplement to existing works because there is no need to tread the exact same ground twice. However, I would not recommend our approach for less well documented subject matter: Anime and manga does not currently have such a closure reason, and I would recommend against them implementing one if I were to see the matter brought up there.
Anyway, my point in bringing the difference in closure reasons up now is that the voting system as it currently stands is balanced for the network as a whole, and not just in consideration of our specific circumstances, and while you could perhaps make an argument for per-site voting scores, you would also have to make an argument against having a consistent scoring system giving users a consistent experience across the entire network. When votes are the only practical control against some categories of questions, more aggressive use of the voting system is needed to sort out the good from the bad, and I believe that the 0 point score is intended to encourage that.
Addressing Mari-LouA's comments:
There is nothing easy about writing good questions, I sometimes spend hours, for example, in doing research before writing the final draft A good question will nearly always receive at least one answer, and if it doesn't it will nearly always be upvoted, it will generate excitement.
I agree Mari-Lou A, but to clarify, "easier" as I put it earlier, is not necessarily the same thing as "easy". The assessment is comparative in nature, and generally holds true irrespective of how hard the question is to ask. Why do I say that? Part of the reason is because many of the difficult aspects of both are shared in common, however in addition to those it is because a research effort is an effort to find an answer. Answering a genuine question requires an extra bit of knowledge that asking the question does not, which is the answer itself. All of the effort a person may put into researching those questions which remain genuine before asking them here are an effort to produce the answer that have failed, so there is no need for me to "test drive it" because it has already been tested. When people show us everything they have tried, it shows us how hard it is to satisfactorily answer a question in the first place.
Even in those cases where you ultimately do find an answer, then the only reason you would ask a question here is to share it in recognition of how hard it was to find in the first place, in the hopes that your question would make the answer easier to find.
Granted, there are other factors which make writing good answers hard, such as proper presentation, so it is not my intention to over-marginalize the value of good questions, however factors such as presentation apply equally to answers, and, we are discussing the penalty systems, so good questions are a secondary consideration at best. We are primarily concerned with the bad ones, since somebody voting against a question is of the opinion that it does more harm than good, and is hence bad. Good questions are only a concern for cases when that assessment is incorrect, and our solution for that is to hope that the aggregate sum total of votes corrects for the marginal defect.
Yeah, Stack Exchange does not value questions b/c on SO answers are far easier and quicker to verify or not.
I don't even really see why that matters.
The system is flawed, downvoting doesn't work. The flood of L.Q. banal-beyond-belief questions continues uninterrupted. The site still looks horrible, there is no incentive to answer because these drive-by users drop their questions and leave the site.
Stopping the questions isn't the sole goal of voting. Votes serve as a system of feedback which help users to improve, and also affect sorting. If I see a question that is rated 11, and a question that is rated 10, I should be able to presume that the question rated 11 is of greater interest than the question rated 10, which is to say nothing of questions that end up going into the negatives. People don't really have to bother clicking into a question if they suspect it's going to be bad from the start, and the system even sorts negatively rated posts away from the front page so you don't even have to see it.
Also, I wouldn't be so quick to assume that it isn't stopping some bad questions. We'd only know if that is a fact if the change was implemented and the rate of bad questions didn't exceed the projected expectations of the current system.
Some questions appear deceptively easy when in reality they are reveal more complex issues.
Yes, I know that. Sometimes I spend days trying to write detailed answers to seemingly simple questions, and sometimes I am greatly disappointed to see those questions closed before I can even post them.
However, just because a question has some deeper interest does not necessarily mean that it will be conducive to our goals. The problem with deceptively easy question is that they are deceptively easy, which has a detrimental effect on the ability for votes to sort out the best answer. The obvious answer can be given immediately, even if it is problematic, and place a burden on more insightful answers that get less exposure later down the line. Closure gives us a chance to fix the deceptiveness by establishing that further insight is needful.
It's one of the main reason I often insist that some questions should be closed. Voting alone can't be trusted to sort out insightful answers from uninsightful ones. Ultimately, closing a question puts a stop to answers, not questions.
There are no upvotes, only when a question hits the HNQ. Note, it's called Hot Network Questions. Then answers consisting little more than a line can be wildly upvoted. Some answers can be brilliant and enlightening but if they don't see the HNQ they attract maybe 5 upvotes from the same users.
H.Q.Ns. are a separate issue in my opinion, but my opinion of them is that they break the balance of the system. They generate a whole bunch of artificial interest in a question, only to have them rated by people who are less likely to possess insight regarding the subject matter, since their fields of interest and expertise regards other subjects on the network. I don't think they're very conducive to our goals.
Paying 1+ for downvoting a question is peanuts.
Every penny adds up. If you vote against as many questions as you can in a day, it costs 40 rep, which is 80% of the lowest bounty. Since we have so many poor questions in need of remedy, there is no need to risk making the system even less effective than it already is by adding an associated cost to one of the controls against poor questions.
Besides, the same thing can be said about voting against answers. The only difference is that we have more reason to penalize that.