Wikipedia talk:Editing scientific articles

The Essay in question

Editors are editing away the actual content of this essay. That means that instead of having people discuss and think about this proposal, it is being reworded to mean the exact opposite of what it is saying. In order to prevent this, I will put a copy down here in the talk page. This is what was being discussed above (more or less, there is also the original Count Iblis version, which he might want to put here too)

The nutshell bit that says "If you are an expert with a working knowledge of the subject, make sure you are as rigorous when writing for Wikipedia as you are when you write for scientific journals" was added later, and was part of the original proposal, but I find no support for it in the essay (nothing about experts editing, nor rigor, nor writing for journals); isn't a nutshell supposed to be a summary? Dicklyon (talk) 06:00, 11 November 2009 (UTC)

When writing for a scientific journal, you check what you say for logical consistency, for first-principles accuracy, and for accuracy about sources. That's true regardless of whether you are writing original material, or just reviewing a well understood topic. It's just how scientists write. The article is just explaining how this type of writing works in a more verbose way, so that editors who haven't done this will do it too.Likebox (talk) 07:12, 11 November 2009 (UTC)

I agree that's a characterization of how we write for journals. But why should it be addressed only to "experts with a working knowledge of the subject"? And if it's supposed to summarize the proposal, why doesn't the proposal say these things? The connection is too implicit to make sense here. And the proposal is really more about how to deal with disagreements with other editors than about how to write. Dicklyon (talk) 07:32, 11 November 2009 (UTC)

Anyone persuing a Ph.D has to write articles, so it is something that many experts will be familiar with. Now, especially if you are an expert you may have a mentality that since this is "just wikipedia", you don't need to be careful.

Only point 3 deals with disputes, point 5 is a good way to prevent a dispute. Disgreements between statements are not necessarily disgreements between editors. Count Iblis (talk) 03:20, 12 November 2009 (UTC)

The actual essay (not inverted)

Wikipedia's content policies are often unambiguous when working with sources which use ordinary language and everyday concepts. But scientific terms are much more precise, and require much more care. When writing about science and mathematics, accuracy is paramount, and the policies must be followed carefully to avoid introducing technical errors.

For scientific articles, following the letter of Wikipedia policies often is not enough to guarantee that an article won't contain serious factual mistakes or misleading statements. When editing scientific articles, try to follow these suggestions:

1. Check any non-trivial statements you intend to insert into an article. Determine whether your statement could be invalid under some circumstances. To find out, you may need to study the entire source in which the statement is made, or look in other sources. The validity of a statement made on some particular page of a technical book may well rely upon necessary conditions mentioned many pages earlier, or even in another source. If you find that the statement is valid only within a specific context, you should take the effort find out what this context is, and to include that context in the article.
2. After checking carefully, you may find that a statement you want to insert still disagrees with other statements made in the same article. It may be that the conflicting statements are true under some conditions not explicitly mentioned in the article. Any apparent conflicts should be worked out by the editors through discussions.
3. In resolving technical disputes, the exact wording of quotes from sources is often unhelpful. It is essential that the editors sort out the scientific issues from first principles as much as possible, referring to all the ideas in the sources, not to out-of-context fragments. This means that editors should read the technical literature with the goal of acquiring a full understanding of all the relevant points while editing the article. It is important that every disputed point be explained clearly, so that any remaining disagreements which appear in the encyclopedia reflect actual diverging points of view in the literature, not the misunderstandings between editors.
4. When editing a scientific article, be careful to be as complete as possible. Filling in intermediate steps which are omitted from more condensed literature sources is not OR. Rephrasing content is not synthesis, as long as the ideas faithfully represent the ideas in the sources; remember that “Carefully summarizing or rephrasing source material without changing its meaning is not synthesis—it is good editing.”
5. Assume from the outset that multiple meanings of technical terms are likely to occur, whether or not you are aware of them, so search for meanings proposed by other editors, rather than searching only to back up your own understanding. The goal is for the editors to gain a comprehensive familiarity with as much of the literature as possible.
6. Different approaches or explanatory models are often all correct, and different readers will find different explanations useful. Don't delete existing explanations just because they use a different model; add your explanation to the article, so long as it is carefully sourced.
7. Discussions from first principles or evidence are not violations of the ban on original research if they are conducted on the talk pages.
8. The perception of what is original research varies according to the level of expertise of the editor; it would not constitute WP:OR to provide the logical connection between sourced premises and sourced conclusions, when the arguments are well understood by experts in the field, but this might be disputed by someone less familar with the topic.

administrative action

I recieved a comment on my talk page that Jayjg thinks I am being tendentious. This editor would like to take administrative action.Likebox (talk) 04:51, 11 November 2009 (UTC)

Jayjg, it's just an essay for gawd sake. Essays can represent minority views. Calm down. --Michael C. Price ^talk 08:34, 11 November 2009 (UTC)

There is no requirement that essays have to agree with actual policies or practice. For example, Wikipedia:Delete the junk and Wikipedia:No original biographies. This is why they are called "essays". — Carl (CBM · talk) 12:26, 11 November 2009 (UTC)

FYI Jayjg has succeeded in getting Likebox blocked for a week. --Michael C. Price ^talk 12:47, 11 November 2009 (UTC)

One may well ask whether Admins are following WP:Block. In my opinion, they are not. Brews ohare (talk) 18:52, 11 November 2009 (UTC)

Well, I'm unblocked now. Thanks for the help. But it was annoying to deal with.Likebox (talk) 05:23, 13 November 2009 (UTC)

Good to have you back. If Jayjg causes trouble again it might be relevant that he has a long history of repeated edit warring, POV pushing, rushing to sanction others without caution and "unbecoming conduct", for which he he received an indefinite editing restriction. See here for details. --Michael C. Price ^talk 14:54, 13 November 2009 (UTC)

Hmmm, this is new to me. I remember that Jayjg was under some sort of restriction from a much earlier case, but this is very recent (May of this year). This is what I mean when I comment on Brews ban from editng physics articles. The troublemakers on politics pages are dealt with infinitely more lenient by the same Arbitrators. It is not that I deny that Brews editing style did not cause any problems. My opinion is that having Brews back to editing physics articles, perhaps under some sort of mentoring agreement, would benefit Wikipedia a lot. Count Iblis (talk) 15:28, 13 November 2009 (UTC)

Count Iblis: Thanks for your vote of confidence. I don't know anything about mentorship, but unless it involves a shield from ridiculous blocks and motions that are simply time sinks over complete trivia, I don't think my participation on WP is likely. Statements by TenOfAllTrades, MBisanz, Jehochman and others indicate no tolerance for any activity I might attempt. Brews ohare (talk) 17:25, 13 November 2009 (UTC)

Not to derail the thread, but that is not true. We suggested several venues for you to edit that wouldn't be controversial in the least and which would be much appreciated, like expanding stubs or some cleanup. Headbomb {^ταλκ_{κοντριβς} – WP Physics} 18:09, 13 November 2009 (UTC)

(edit conflict)Brews: That is a misreading of what they are saying. They would have no objection to your non-disruptively editing and discussing articles that are outside your topic ban, for example. There is wide range of "activity [you] might attempt" that would help build the encyclopedia, and to which they would not object. What they do object to is your constantly keeping the Speed of light drama alive by trying to undermine good policies, complaining about arbitrators, complaining about admins, complaining about Wikipedia's culture, complaining about how you've been treated, and other miscellaneous whining, which appears to be all that you are doing here. It's getting very old, tiresome, is not contributing to the project, and will continue to get you in trouble. —Finell 18:19, 13 November 2009 (UTC)

Sorry, but misreading is from all sides. If you find all this tiresome, try experiencing harassment over trivia for months. Nothing I have been doing is related to Case/Speed of light apart from the fact that that case underlined some major problems on WP. Those problems are endemic and transcend any particular case. They can be found all over WP and occupy WP:AN/I and administrators for weeks and months. It's hard for me to understand why my actions to mitigate this situation should be continuously distorted and misinterpreted. This strange campaign extends the sanctions to include a remark like Superman travels at the "vitesse de la lumière". (Pardon my French.) And even to objections to composing an essay on my user page and a block for answering a question about this essay!!! Brews ohare (talk) 18:33, 13 November 2009 (UTC)

My attempt to keep Jayjg to stick to politics has failed Count Iblis (talk) 21:59, 13 November 2009 (UTC)

"Opposing viewpoint" section in essay

The poor quality of this section speaks for itself. The editor who inserted allows no correction, with the comment " if you npov my section, I'll npov yours.". Fine, have your way. I'm sure the section won't survive in the long run but if, in the meantime, it prevents disruption to the evolution of the main text, then I'm okay with it. --Michael C. Price ^talk 13:04, 12 November 2009 (UTC)

I removed it, we already know that the essay gives the view of a minority. Now if that section is not going to discuss in detail what the problems are, then it is a useless section. It would be similar to Woit coming to the string theory article and writing in that article a one line sentence that Woit, Professor of mathematics disagrees with string theory, without even discussing why. Count Iblis (talk) 14:39, 12 November 2009 (UTC)

This essay is the statement of a minority? Please acknoledge that in the essay. Thanks. Hipocrite (talk) 14:40, 12 November 2009 (UTC)

This essay can only discuss the editing of scientific articles and perhaps any potential problems with the mentioned points. So, a discussion section in the essay where some pitfalls are discussed is, in principle, ok. So, if you can argue why some points in the essay could lead to problems and what one can do to avoid such problems, you are welcome to add that in such a discussion section. Count Iblis (talk) 14:47, 12 November 2009 (UTC)

Count, you're correct, but let them have their silly section for the present if it prevents their disruption to the rest of the essay while we develop it. --Michael C. Price ^talk 14:55, 12 November 2009 (UTC)

Are you going to make it clear this essay is the point of view of the minority in any way, or are you both just going to tag-team revert to your preferred version? Would you prefer this essay be userified? Hipocrite (talk) 15:05, 12 November 2009 (UTC)

Is it a minority view of editors, as opposed to editors who shout a lot? I think it describes actual practice pretty accurately, so it implicitly reflects the views of the majority of science-article-editors. Whether it is a minority view or nor is debatable. As the essay template says, it may be a minority view.

"Essays may represent widespread norms or minority viewpoints." seems to sum it up very nicely. This essay describes what actually goes on on science article talk pages already (where there is widespread discussion from principles) along with some anodyne suggestions (which have been incorporated into your essay).

--Michael C. Price ^talk 15:17, 12 November 2009 (UTC)

There is no assumption that an essay reflects community consensus. To the contrary, the essay banner template states clearly that an essay may represent the views of just one editor or a minority. Therefore, there is no need to state that an essay represents a minority view. —Finell 17:57, 13 November 2009 (UTC)

Original research

My take is that this essay has wandered from its original purpose and has become less effective. It seems to me that the main area of dispute concerns OR, and that has infected the entire article development. Perhaps it would be clearer for the article to have a separate section dealing directly with the OR issue. Then the rest of the article could be returned to more like its original form, and dispute would be more focused. That division also would be clearer for the reader. Brews ohare (talk) 18:10, 13 November 2009 (UTC)

The OR issue has arisen already on WP:NOR in discussion of syllogism. There is a great fear on the part of some that they cannot tell when a technical derivation is accurate, so it's best to require it to be verbatim from a source. This is the nub of the problem: how can a non-expert tell whether a section is correct or not. If that cannot be done, how can a non-expert determine that another editor is more qualified than they? Or, can non-experts be relied upon to admit to incompetence that they know themselves to be so? Brews ohare (talk)

A different solution, based upon sourced premises and conclusions, would seem to limit the possible damage of an incorrect deduction, but even that fails to reassure the non-expert. Brews ohare (talk) 18:19, 13 November 2009 (UTC)

Yet another solution, the argument from first principles, also is resisted by the non-expert because (i) they don't understand the principles, and (ii) they don't have any confidence in their ability to track an argument. Brews ohare (talk) 18:21, 13 November 2009 (UTC)

I think that we have to say that non-experts cannot be expected to be able to always verify everything in a straightforward way. Because understanding the sources may require the non-expert to study the topic from textbooks for quite some time.

About OR, we could say that what matters is that the text in an article as judged by scientists working in the field does not contain any novel statements. The editors here have to make a judgement whether or not this criterium is met. If they are in agreement about that, but the text happens to be in conflict with the exact wording of the OR policy, then the text can be admitted per WP:IAR. Count Iblis (talk) 18:29, 13 November 2009 (UTC)

I never thought the original essay [1] was about OR in the first place. Of course you are right that we simply do not require all deductions to appear verbatim in sources, and people who believe we do are overreading the actual NOR policy. But I don't think it is practical to try to clarify that sort of thing in a guideline, because there is little one can say in general; the actual policy is to make decisions on an article-by-article basis.

It is pretty common for editors who follow some policy page closely to read that policy as being more strict than it is, and this is visible at least among people who follow WP:V, people who follow WP:NOR, people who follow WP:NFCC, and people who follow WP:CSD. I am sure that if I proposed a guideline that clarified what the CSD policy actually is (versus what the page WP:CSD claims the policy is), that proposal would get opposition for the same underlying reasons that this essay did when it was proposed as a guideline. The same would be true for WP:NFCC, and it is clearly true here for WP:NOR.

So I think that we should just make sure any attempted explanation here is clearly marked as an essay. It might be better, actually, to just put it in userspace. — Carl (CBM · talk) 18:30, 13 November 2009 (UTC)

I also think that, if I were going to propose any clarifications of the actual NOR policy, the least helpful thing would be to have my proposal strongly supported by editors who have recently been sanctioned, by arbcom or otherwise, for tendentious editing and original research. In the outside world, groups seeking to have laws modified often wait for the perfect case before pursuing appeals to the Supreme Court, and avoid cases where there are distracting or negative factors that might bias the case against them. A similar sort of discretion would be helpful here. — Carl (CBM · talk) 18:39, 13 November 2009 (UTC)

I take that as an invitation to butt out. I would hope that even a felon might have an occasional good idea that might be considered on its merits without the paranoid delusion that it is a Trojan horse. Hannibal Lecter was used as a consultant. Brews ohare (talk) 18:45, 13 November 2009 (UTC)

(deindent) On expertise and noOR: There is a legitimate fear among editors that with nontrivial content, which includes derivations of mathematical statements, and detailed exposition, then verification will become a circus of diverging ignorant opinions. This is a reasonable worry, but I think it is overblown. The question is do we need a big boss to step in and say "this is correct" and "this is incorrect"? Or can we rely on non-expert editors to acquire this knowledge through editing, reading, and talk-page discussions.

The notion of expertise was important in the print era, when access to information was slow, and acquiring expertise took years and years of patient study in a university with a mentor. We are living through a transformation in the availability of technical knowledge, and we are now at a point where even the most detailed stuff is widely accessible to anyone with a search engine. This means that it is not absolutely necessary to defer to expert opinions, because anyone can know all there is to know, with patience and help.

This does not mean that every editor will have an easy time verifying technical information in articles--- far from it. They will have to study the subject, read the literature, read the sources, and learn about the topic in an in-depth way. But I think Wikipedia can ask this of its contributors, and in many cases, non-expert contributors have taken the time to do this, and have written wonderful expositions. For historical material, possibly non-expert contributors were able to present technical material in a lucid way that went far beyond standard textbook expositions (although not beyond the literature).

For example, the historical information about Moseley's law on Bohr model is not usually found in textbooks, but it is found in the literature. Same with the discussions of Laplace's argument about the speed of gravity. These contributions are known, but not included as part of the standard curriculum. If we treat out contributors as capable readers, we can expect them to understand the literature that they are reviewing at the highest level. If there is a problem of understanding, we can hope that experts will step in to correct the errors. If the discussions with the experts goes along the way suggested by these guidelines, everyone will get up to speed. Verifying material is a time consuming task, but there is no reason to suppose that it cannot be done.Likebox (talk) 23:04, 13 November 2009 (UTC)

And yet experts have corrected you repeatedly, Likebox, but you never get the message. To place the blame on the experts (they didn't follow "the guidelines", so it's the experts fault and I can keep being disruptive) is extraordinarily offensive, but Count Iblis has implied exactly that on more than one occasion. I'm afraid Carl missed the intent of this proposed guideline as it was clear that from the beginning it was always about weakening verifiability to allow original research in the areas favored by its proponents. Likebox and Brews support the essay because they want to use it to justify their behavior, but this behavior is outside of community norms and it should not be surprising that there is significant opposition to a proposal seen as enabling disruptive editors. The heavy-handed attempt to spam it as a hatnote on several article talk pages was unfortunate too as this suggests an attempt to try to circumvent policy through the back door. If the proponents here have so many problems with WP:NOR they should go to WT:NOR to try to get policy changed there rather than trying to build a walled garden in their favorite science articles in which pesky wikipedia policies and guidelines are not applied. Quale (talk) 23:35, 13 November 2009 (UTC)

Quale, you are incorrect that the original intent was to weaken OR. The original intent was to help discussions on the talk pages which are not OR. The essay has flirted with weakening OR, but if you read the current state of the essay you will see no violation or even weakening of OR. --Michael C. Price ^talk 00:20, 14 November 2009 (UTC)

I agree with Michael. Let me try to explain it differently. The purpose of this essay is to make sure no errors slip into a certain class of scientific articles. There exists articles here on Wikipedia for which simply sticking to the WP:V and WP:OR policies are not sufficient to guarantee that no errors slip in. You can think of articles in which there are a lot of mathematical derivations. Far from weaking verifiability and allowing in original research (which in practice means allowing in errors, as typically a novel statement equates to an erroneous statement in most wiki articles on well established subjects were nothing new is to be discovered), we want to strengthen verifiability and make sure that no statements are made that are not consistent with current scientific knowledge.

To achieve these goals, you need to do what this essay says. I have given plenty of examples. E.g. many thermodynamics articles were saying for two years that dU <= T dS - P dV. That error could have been prevented had the editors stuck to the recommendations in this essay. These editors were experienced Wiki editors who knew about WP:V and WP:OR. Saying that sticking to these guidelines would have been enough, because there are plenty of textbooks that say that dU = TdS - P dV, is not a good argument against this essay, because editing practice shows in this and many other cases that the (simple) errors were not promptly corrected.

The problem was that the statement dU <= T dS - P dV was sourced. A flawed derivation for this flawed statement was also given. It was the dumbed down Wiki-mentality that stopped people from even questioning this ridiculous statement. Had they done so, they could have found out that it was wrong and then they could have confirmed their suspicion by consulting another source.

Had someone without much knowledge of thermodynamics seen the correct equation, then a posting by such a person that something seems to be wrong on the talk page would also have been helpful. But that would require a willingness of the regular editors to engage on the basis of the other points metioned in this essay, and not simply saying that the statement in the article is sourced and therefore correct. The lay person could then not contribute to the debate, but you would then hope that the regular editors would start to discuss the issue among themselves. Count Iblis (talk) 00:43, 14 November 2009 (UTC)

If you notice an error in an article, I submit it would be more productive for everyone if you just fixed it yourself (WP:SOFIXIT) rather than complaining that WP:V makes science articles inaccurate and that everyone else is doing it all wrong. You admit that WP:V did not prevent fixing the error you note, but you make the unsupported and I think unsupportable claim that WP:V and WP:OR cause simple errors to not be corrected quickly and that using your proposal would prevent those errors. I don't see any reason to believe that is true. Errors are not corrected quickly because there are not enough knowledgeable editors to fix 3 million articles, and I don't think any policy change can fix that. It is my opinion that wikipedia has many more "expert" pop culture editors than it needs, but not nearly enough true experts in mathematics, the sciences, philosophy, art, literature, etc. Good science articles are often much harder to write than articles on cartoons or pop songs too, so there's even more work to be done by that much smaller group of experts. I appreciate your expertise, but experience shows many science experts can write good articles within the requirements of the policies and guidelines that apply to all of wikipedia, and I think you can do this too if you care to.

You have to do what you want to do, but given your knowledge of physics I would have thought that you would find it more fulfilling to focus more on writing physics articles and less on argument about core policies like WP:V. I also think you do a better job writing about the science you know and I think love than you do wrestling with the wikipedia policies and guidelines which you seem to not hold in high regard. While I support every editor's right to be heard, I don't think that arguing policy or wading into wikidrama really plays to your strengths. Quale (talk) 10:51, 14 November 2009 (UTC)

(deindent) To quale: all the arguments I have included are obvious to experts, anyone who disputed them is, almost by definition, not an expert. I did not meet any actual experts who disputed the correctness of the content of any additions I have made, at least not after patient explanation and a rewrite or two. If you think you have an example of such a case, please bring it up, and I will explain what was going on.

Following this guideline does not weaken OR, it removes a bad interpretation of OR which allows sourced misinformation to propagate, because of misinterpretation. For example, the equation "dU < TdS - PdV" is correctly intepreted in the following form: "dS > 1/T dU + P/T dV", which is correct in a certain sense, it says that the entropy can go up even if there is no energy flowing into a system nor any volume change. But it is incorrect when applied to a system like a gas which is already in equilibrium, which is the automatic interpretation found in the article. For such a system, the inequality is an equality. While this statement is easy to source, it is also obvious to all knowledgable editors, and should not be disputed. When a mistake such as this creeps into an article, the only way to fix it is for discussions to go from first principles.

An example of a simplification for thermodynamics articles which is a trivial rewrite is to replace all "dU = ..." forms of thermodynamic identities with the equivalent "dS = ...." forms, as above. While you can find sources that use "dU = ..." and sources that use "dS =....", most often appearing as "d(F/T) = ...." where (F/T) is the entropy-like quantity which is the free-energy divided by the temperature, or the logarithm of the canonical partition function. To put the entropy-like quantity on the left is the better representation, as understood from statistical mechanics, because this way makes the Maxwell relations obvious. But for historical reasons, we are saddled with historical forms. In this case, there is no reason for us to use the most transparent form. We now know that energy is more microscopically fundamental than entropy.

In order to determine which form is clearest, sources are of absolutely no help. Although there are a few sources that make the argument that entropy-like terms should be put on the left, these sources are useless, because we aren't writing a page called "how to write thermodynamic identities". The thing to do it to verify the material from the sources, then put the sources aside and think hard about how to present the material, and think hard about whether to follow the advice about presentation in the other sources.Likebox (talk) 03:20, 14 November 2009 (UTC)

I appreciate your very clear explanation of the mistake in the thermodynamics articles and I hope that explanation (in encyclopedic form) has a home somewhere in wikipedia. I still don't agree that argument from first principles was required to fix the error. I don't know how the error was noticed and fixed, but Count Iblis admitted that the repair could be made within the bounds of WP:V and WP:NOR. (If I understand correctly, the accurate formula is found in textbooks and is easily sourced.) The high quality of the explanation you just gave of the physics (I would guess that you teach it) makes me wish that you directed more energy to physics articles and less energy trying so persistently to force your unique views and presentation into mathematical logic articles. There are plenty of physics articles that need work, and you write about physics with a clarity I do not find in your writing about mathematical logic.

The issue of what to do when material is presented different ways in different sources is not unique to science and comes up constantly in many areas. Even so, this problem does not require weakening WP:OR. Your advice is sound, and can be followed in accordance with the key policy requirements: WP:V and WP:RS require that the approach used must be found in a reliable source, and WP:UNDUE requires that undue weight should not be given to minority views or rarely used formulations. It is always the case for any article that the sources do not determine precisely how the material is presented and organized, and it is here that subject matter experts and skilled writers can elevate an article beyond the routine work of those who can only summarize by rote. Quale (talk) 10:51, 14 November 2009 (UTC)

Quale: I'm sure you know there is no reason to think WP policies and guidelines are perfect or even as good as they could be. To suggest that Likebox butt out because everything is OK is to avoid the opportunity to discuss improvements. The right thing to do is to distill the problem Likebox sees and solve it.

As I understand the problem in general terms it is this: many textbooks present arguments that follow traditional lines that are not as clear as a more modern attack would suggest. In some cases, a statement of the modern approach can easily be constructed and is more readily understood than a traditional development based on an outdated framework. For whatever reason, however, this desirable approach is not found simply stated all in one place in some source. The issue is then: why not provide the simple modern development that leads to the results? Is there a way to pose this dilemma so its solution fits WP?

I hope I have got the point right. A general statement of the dilemma divorced from particulars like thermodynamics might help in finding a solution. Brews ohare (talk) 15:05, 14 November 2009 (UTC)

Documenting a logical bridge

One proposal is to source certain key premises and results but allow the logical bridge between the results to be unsourced. If there are challenges raised to some unsourced intermediate results, one hopes they could be sourced as well. Thus, the development becomes more tightly verified as time goes on and challenges or clarifications are made. It is not essential that the final product appear all in one sitting: time will allow evolution and challenges always can be raised, just as happens with a normal article. The challenges should not be of a vague nature claiming the entire presentation as WP:NOR or WP:SYN, but to specific items like intermediate results that should be sourced. The request that only premises, challenged intermediate results, and final results need be sourced allows flexibility in the development. Brews ohare (talk) 15:20, 14 November 2009 (UTC)

In an extreme case, every logical step might be challenged and require reference to a simple syllogism to support it. The result may resemble a proof from Euclid. Hopefully that will happen only rarely. Brews ohare (talk) 15:42, 14 November 2009 (UTC)

Carl has raised the point (I paraphrase) that apart from correctness, inclusion of an argument should satisfy other standard criteria such as WP:Weight, WP:Notable.

We are writing for Wikipedia readers, not for University students

Brews, on his talk page, mentioned the audience we're writing for, when arguing in favor of different presentations. This is a very important point to consider. I think Likebox has made this point also.

The readers of a Wikipedia article are not the same students who are following a university course on that topic. But textbooks are written for these university students. This can mean that you have emphasize certain points in a Wiki article that textbooks pay little attention to. You may have to avoid certain jargon that almost all textbooks use. You may have to explain a certain mathematical formalism in much more detail than the textbooks presentations do. It may be advantageous to use a mathematical formalism that textbooks do not use in the context of the topic.

So, textbook presentations should not necessary be the norm of how we present material here on Wikipedia. And editors shouldn't just present material in Wikipedia in their favorite ideosyncratic way either. Count Iblis (talk) 17:45, 14 November 2009 (UTC)

Maybe one way round this would be to use the lead to eplain the topic in a more user-friendly way, keeping the in-depth maths till later. Abtract (talk) 17:59, 14 November 2009 (UTC)

Count Iblis: Very good points. They argue for some room for flexible presentation that goes beyond finding a verbatim source. The preceding proposal is my suggestion for one way to achieve flexibility. Brews ohare (talk) 18:01, 14 November 2009 (UTC)

I recommend that some form of Count Iblis' comment above be placed into the essay. Brews ohare (talk) 18:06, 14 November 2009 (UTC)

Could you explain what you mean when you say we need to emphasize things not emphasized in undergraduate textbooks? It seems to me that our fundamental job, as article writers, is to give a summary of what is in the literature, using the same sort of terminology and jargon that the literature does. Moreover, it seems to me that inventing a new mathematical formalism is exactly the sort of thing that the OR policy is intended to prevent. — Carl (CBM · talk) 18:20, 14 November 2009 (UTC)

My take is that a WP reader has not necessarily the same background as a physics major, and yet a greater sophistication than a fifth grader, so the approach has to be tailored, or multiple approaches presented, for different reader backgrounds. Brews ohare (talk) 19:33, 14 November 2009 (UTC)

How does that relate to what I asked? — Carl (CBM · talk) 19:40, 14 November 2009 (UTC)

The point is that textbooks are a lousy replica of the literature--- the literature is a big thing, and textbooks are one author's sketch of the contents. Textbook writers are not a good guide for which content should be presented, the only guidelines are WP:NOTE, WP:V and WP:NPOV. There is no guideline that says "Wikipedia cannot fix misstatements found in elementary textbooks which are corrected in more advanced literature". If textbooks all say something false, which the literature does not say (for example, textbooks all say that Planck was motivated by the ultraviolet catastrophe when formulating his distribution, but that's actually Einstein's ex-post facto explanation of the law. This is correctly explained here on ultraviolet catastrophe, and correctly sourced to Popper, I believe).Likebox (talk) 01:15, 15 November 2009 (UTC) (originally anonymous by accident--- sorry that was me)

I agree with Anon that sometimes textbooks can be misleading and then you have to do nontrivial literature research. But what I meant to say and what I think Brews' and Likebox' opinion is, is that in many cases the way things are explained in literature is not a good way to present the topic in Wikipedia. It really boils down to what we think we should do here at Wikipedia. Explain the topic as best as we can, or just give a summary of the literature. The two things need not be equivalent. Of course, whatever we do must be consistent with wiki-policies. So, we are not going to do original research.

If a different presentation than that given in the literature is seen to be more appropriate, then why not include that in a Wiki-article? Of course, the editors should agree that it is indeed better and that it is not original research. Here "not original research" is to be understood in the sense that an expert reading the article would not learn anything new (or see that something is deeply wrong). All that the expert may notice is that the text explains the topic in a different way than usual.

It is not to be understood in the way Likebox has often described it: A Wikilawyer who does not understand the topic coming to the article's talk page and then objecting to the text because the letter of WP:OR has been violated in his/her opinion. A valid ground to object would be if an editor is not conviced that the proposed text is indeed a valid alternative presentation of the topic relative to that given in textbooks. Count Iblis (talk) 22:39, 14 November 2009 (UTC)

We have gone over this and over this and over this in the context of your friend Likebox's tendentious attempts to rewrite halting problem to be how he thinks it should be presented rather than how the textbooks present it. Everyone else agrees that this approach constitutes original research, and I find your continued defense of it troubling. Your implication that anyone who objects must be someone who doesn't understand the subject is false, uncivil, and a violation of WP:AGF. —David Eppstein (talk) 23:54, 14 November 2009 (UTC)

You don't get it (and I mean what I just wrote, not the halting problem). If you as an involved editor object to Likebox's edit then the ball is in Likebox's court. Likebox then cannot edit in his proof until agreement is reached. Likebox also has to take into account the finite patience of his fellow editors.

So, all I'm saying is that presenting a topic in a different way than in textbooks is not necessarily original research. You need to evaluate this on a case by case basis. If you don't agree with this, then you arrive at the silly conclusion that you always need to exactly stick to how textbooks present the topic. I think Carl has pointed out a few times that many math articles violate a strict reading of WP:OR, simply because they include examples that are not taken from text books. He has also argued that it is difficult to legislate where the boundary should be. What I'm saying is that the boundary should be determined by the editors in any particular case. We simply tell the editors to stick to the theory as presented in the literature, not necessarily to the exact way the theory is presented in some source. Count Iblis (talk) 00:15, 15 November 2009 (UTC)

I agree with you; OR is a problem mainly when it's challenged and the editor won't back down. But I think you do us all a disservice to characterize those who push back on OR as requiring editors to "always need to exactly stick to how textbooks present the topic." Brews used to try to tell me that's what I was doing when I pushed back on his OR, but that's never been an actual problem on wikipedia as far as I have been able to tell. Dicklyon (talk) 00:25, 15 November 2009 (UTC)

Yes, I'm sure no one would demand we exactly stick to textbooks. But didn't you edit out the statement about writing for scientific articles, just because you couldn't find a similar frase in this essay? :) Anyway, we don't need to focus only on editing disputes, because in most articles, there are no editing disputes. The focus of this essay is to have recommendations that will make the chance of errors smaller. Also, we can have recommendations that will make articles accessible to a wider audience.

Then, it may look like those recommendations may violate the OR policy. That is then something that one has to discuss, like we're doing here now. So, Eppstein gives the example of Likebox's alternative proof that was supposed to be more accessible. However, that was disputed by other math editors, so it did not appear in the article. I'm now not saying that a recommendation that would encourage more accessible proofs, would allow Likebox to edit in such proofs over the objection of his fellow editors. Count Iblis (talk) 00:46, 15 November 2009 (UTC)

(deindent) To be clear about what these contested proofs are:

Halting problem

For contradiction, suppose there is a computer program HALT(X) which tells you whether X halts or not.

Write SPITE to do the following:

1. Print its own code into a variable R
2. calculate HALT(R)
3. do the opposite: meaning, if HALT(R) says "halts", go into an infinite loop, and if HALT(R) says "doesn't halt" to halt.

This is the proof the caused so much controversy.

Godel's theorem

For Godel's theorem, the proof is as follows: suppose an axiom system S is computable, meaning there is an algorithm to list the theorems, and capable of making statements about a computer. Then consider the program GODEL which does the following:

1. Prints its own code into a variable R
2. deduces all theorems of S looking for "R does not halt"
3. if it finds this theorem, it halts

This is the proof of Godel's theorem. The Godel sentence is "Godel does not halt". It is true, but unprovable.

Rosser's theorem

To prove Rosser's theorem, with the same assumption as Godel's theorem, consider the following program ROSSER:

1. it prints its code into R
2. deduces all theorems of S looking for a) "R prints to the screen" or b) "R does not print"
3. if a) halt without printing anything. If b) print "hello" to the screen and halt

The Rosser sentence is "ROSSER does not print to the screen". Neither this statement nor its negation can be printed out.

Godel's proof length theorem

Godel's proof length theorem of 1934 states that for any computable function "f", there is a provable theorem T, whose statement is of length N bytes, but whose proof is longer than f(N).

To prove this, you do the following:

1. print your code into R
2. calculate the length of the statement "L prints to the screen", and call it N.
3. compute f(N)
4. deduce all theorems of length less than f(N) looking for "R prints to the screen". If you find this theorem, halt without printing.
5. print "finished" to the screen, and halt.

This construction provides the proof. This theorem was controversial for decades.

Injury/priority

Post's work in the 1940s led to the injury/priority method, which involves very complicated computer programs. The essence of injury methods is programs similar to those above. The goal of my edits was to build up to a presentation of the injury/priority method here, which would be extremely useful.

What happened?

The point of my contributions was to see how amenable the encyclopedia was to presenting injury methods in a natural way. Instead, I got arguments that the trivial review proofs of Godel's theorem/Halting problem was OR! That's complete nonsense. This proof is equivalent to Kleene 1940's paper. The place where OR (arguably) starts is with Rosser's theorem, which is proved here in an ever-so-slightly different way than Rosser's 1936 paper.

The proof-length method is essentially equivalent to Godel 1936, but takes more liberties (still not enough to be independent), while the injury/priority argument does not exist, because I got discouraged by the amount of flak that this type of presentation got.

Having gotten bogged down in arguments over trivialities with exceptionally ignorant people demanding that I show sources, I have little patience for editors who say "show me the sources". These people are not helpful to the encyclopedia, since they are not contesting material they believe is wrong, just material that they find unusual. This is not a battle for accuracy, but a battle against innovative exposition.Likebox (talk) 01:43, 15 November 2009 (UTC)

Yes, I agree that material should only be contested based on the content. A fellow editor could say that it would take some time to study your proof. It is also reasonable for the fellow editors to say that because they cannot a priori assume that there are no problems with the proof, that it not be included until they had the time to study it in detail and discuss any problems with you. In that discussions any "real" original research problems could be raised.

This then does mean that you could be bogged down because of a "we don't have time" problem, instead of an artificial OR issue. But that problem can be solved on the long run by getting more experts interested in Wikipedia. Count Iblis (talk) 02:07, 15 November 2009 (UTC)

I am not sure if you realize how many "experts" commented on this matter. For example, in this conversation about the incompleteness theorem page, at least Arthur Rubin and Trovatore have doctorates in mathematical logic. Also, David Eppstein and Hans Adler have commented on previous discussions about the same material. To be clear: I am not arguing that experts are always right or have any privileged position, and I don't want to put any of those four editors on the spot. I am only saying that there is no lack of expert attention in that article.

We are fortunate that mathematics is one area where there are lots of content experts around to look at things. Indeed, I think one reason that so few editors of math articles wave their credentials around is that there are so many other PhD mathematicians that it would come across as sophomoric to do so. — Carl (CBM · talk) 02:35, 15 November 2009 (UTC)

(inserted later) Trovatore always believed that the proof was correct, but he felt the language was too quirky for Wikipedia. That's a reasonable criticism. The unreasonable criticism came from others, who said that this material violates OR policy.Likebox (talk) 20:54, 15 November 2009 (UTC)

Yes, I agree that there are quite a few math experts here. Also, we all know that even after a review a dispute may exist about the fairness of the review, like in this case. Sticking to the points of this essay would only allow a discussion of Likebox's proof to go forward. It doesn't mean that the others have to accept the proof. If, in their opinion, including the proof does not improve the article, then it cannot be included. All that the essay would say is that one should not simply reject the proof, just because you cannot find it in a textbook. Count Iblis (talk) 03:16, 15 November 2009 (UTC)

When this page says, "experts often do not consider that providing the logical connection between sourced premises and sourced conclusions as original research", I read that as "providing the connection between sourced premises and sourced conclusions in the way that this connection is typically made in the literature". I cannot agree that editors are free to invent their own methods, unrelated to the literature, to "connect" things. My main context for evaluating this in the in the context of mathematical theorems, where "sourced premises and sourced conclusions" is the same thing as "a theorem whose statement can be sourced", and "the logical connection" means "a proof". To the extent that we provide proofs on wikipedia, the proofs should match those that are actually in the literature. — Carl (CBM · talk) 15:04, 15 November 2009 (UTC)

We need to clarify that statement then. The methods used must themselves be well known methods in the scientific community. The way the whole proof is presented must not be out of line with the thinking in the scientific community.

There may be a difference in culture between math and theoretical physics. In math you often see new proofs published in journals, while in physics that is not done that frequently. So, if some physicsist is able to give a more streamlined proof of some result it often does not make it into the official literature. It stays within the realm of college notes, Summer School lectures etc. Then, when the new generation of physicsts start publishing they may use that technique in their articles. It may sometimes be necessary to explain that the used technique is not originally due to the author. E.g. Ryder, in his textbook on quantum field theory, when deriving the S-matrix from the path integral formalism, writes in a footnote that he draws freely from unpublished lectures by M. Veltman given at Basko-Polje school in 1974. Count Iblis (talk) 15:47, 15 November 2009 (UTC)

I'm afraid I now still disagree that the proofs given are simpler than the "standard" proof, and I don't see what the sentence for the "proof length theorem" is. Nonetheless, the "concept" of a computer program placing its code in a variable (in context, "printing" is wrong) is equivalent to a non-trivial formula coding and diagonalization technique. Hiding that technique in such a simple statement might be acceptable if someone had done it in the published literature.

Again, as I've noted, matrix calculus tries to maintain a single notation, which may or may not have appeared in the literature. Even if the notation has occured in the literature, it becomes difficult to verify whether certain matrix transposes belong in the formulas. If the formulae had actually occured in a reference book, then we (at least those who have copies of the book) could easily maintain the formulas. As it stands, I, at least, have to rederive the formulae each time a — perhaps, well-intentioned — editor changes them.

(In this paragraph, I've decided to alternate formulas and formulae. Perhaps this should be the new Wikipedia standard?) — Arthur Rubin (talk) 16:19, 15 November 2009 (UTC)

(deindent) I think Count Iblis hit the nail on the head about physics culture: I always just assumed that reworking and streamlining of old proofs is not publishable work.

About the proof length theorem: this is a great theorem of Godel, an underappreciated follow-up to his 1931 paper, and I wanted to explain the result and the proof here, because it has an ignominious history. Godel published the theorem in 1934 with a one-page proof, nobody understood it at the time, as far as I can see, and I can see why: I couldn't make heads or tails out of the paper.

But it passed review, probably because Godel's reputation was godlike back then, but I think that nobody really understood it. Every once in a while somebody presents a simplified proof. Because the original is so short, and the proof has been controversial for so many decades, a lot of people don't believe that Godel had a real proof.

The proof length theorem is about a (consistent, computable) axiomatic system S which is sufficiently strong that the incompleteness theorem applies. It has three parts:

1. There are short theorems with very, very long proofs: The precise form of this statement is that for any computable function "F" there are theorems whose statement is of length "L" but whose proof is of length greater than F(L) (think of F(L) as exp(exp(exp(L))) ).
2. Going to stronger systems shortens proofs: If you pass to the system S+consis(S), the proof of these same theorems collapses to a constant length. So that a very long proof in S becomes a short proof is S+consis(S)
3. The same holds for any system T which is computationally stronger than S. There are theorems provable in both S and T, but the proof in T is shorter by an arbitarily huge amount.

Godel proved 1 and 2 by a method which I couldn't follow very well, because it is tied up to his ideosyncratic notation. But the proof is extremely short, less than a page, indicating that the essential idea is already present in the incompleteness theorem.

The actual proof can be reconstructed ex-post-facto from the statement and Godel's previous result as follows: consider the computer program PROOFLENGTH which:

1. prints its code into R
2. calculate the length of the statement "R prints to the screen", and calls it N.
3. computes F(N)
4. deduce all theorems of S of length less than F(N) looking for "R prints to the screen".
5. If it finds this theorem, it halts without printing.
6. Otherwise, it prints "finished" to the screen, and halt.

Since S is consistent, PROOFLENGTH will eventually halt and print "finished". Since PROOFLENGTH halts, S will eventually prove that it halts, and also that it prints "finished" to the screen.

But by construction S will not prove this with a short proof. The proof will necessarily have length greater than F(L), where L is the length of the statement "PROOFLENGTH prints to the screen". This is obviously just a finitistic version of Godel's original method.

To prove 2, note that S+consis(S) will prove that PROOFLENGTH halts and prints "finished" with a proof which doesn't depend on the function F. This is Godel's proof shortening.

To prove point 3 is a little trickier, and I showed how on the talk page of Godel's incompleteness theorems a few months ago. It involves finding one program which T proves does not halt but S does not, and following the operation of this program for a long time, to see if it halts.

The controversy about this theorem came from what exactly the "length of a proof" is. In the preceding, it is the length of the proof in bytes. But there is a different definition of length, which is the number of applications of the axioms and deduction rules, which is not the same as the length in bytes because the axioms can get unbounded. In this interpretation, the proof shortening theorem depends on something called "Kreisel's conjecture", which says that proof-length is equivalent to proof-byte-size for the purposes of this theorem. But, complications aside, this gloss helped me make sense of Godel's paper.Likebox (talk) 21:35, 15 November 2009 (UTC)

Thanks for the explanation. The material still seems better suited for WikiBooks than for Wikipedia. We have enough trouble with published, but disputed, proofs, such as that of Fermat's last theorem or the four color theorem, to worry about unpublished variants of published proofs. — Arthur Rubin (talk) 18:15, 16 November 2009 (UTC)

Maybe you are right. But what are we going to do about the proof length theorem and injury/priority? Are we just not going to cover them? This would be a big gap.

I agree that there is a certain amount of novelty in this type of exposition, but I am not sure exactly how much. I found one theorem so far which is easy to prove by this type of method which doesn't appear in the literature, so there might be something new here. This theorem might not be considered difficult by logicians however. But, regardless of the degree of novelty, the issue of presenting Godel's proof length theorem remains. I don't know any other way to cover the theorem clearly.

If you want to stick to Godel's original exposition, in his language, the original paper of Godel focuses on statements of the form "Forall N (computational statement on N )". Godel assumes that the stronger theory will prove one of these statements, and the weaker theory will not. The computational statement could be "there is no proof of a contradiction in S of length <N bytes", for example, and then each instance of the theorem is provable in S, but the forall N version is not provable in S, and the forall version becomes provable in S+consis(S). If you think of the program that tests the computational statement looking for a contradiction, then this is equivalent to saying that the stronger theory proves that the contradiction-searching program will not halt.

Then Godel points out that given such a statement, the proof of each instance of this statement for larger and larger N requires a longer and longer proof, since the only way to prove the statement once and for all is to prove the "forall N" version. That's equivalent to truncating the program after a finite long running time. If this running time is a large computable function, then you can make a statement which is provable in the weaker theory with a very long proof, but whose proof collapses in the stronger theory. The construction is then essentially equivalent to the one that I wrote above.

But the chain of reasoning to show the equivalence is sufficiently complex in this case that I think there is a case to be made that it is OR. I am not sure, since I am familiar enough with arguments like this that it has become obvious to me, but that doesn't mean anything.

I think we have to be mature about this issue. We should find a cutoff for originality which is somewhere near the cutoff for originality in a journal. That way, if good material would be rejected from journals for lack of originality, if it is all review of previous results, then it isn't OR. I think that the streamlined proof of the proof-length theorem is on the borderline. It might have a home in a journal. But then this can be a guide for where the cutoff should be.

Maybe someone else would put the cutoff somewhere else. But it's important to place the cutoff in a place where exposition of mathematics is not needlessly made more difficult than it needs to be.Likebox (talk) 20:36, 18 November 2009 (UTC)

Analogies

The use of analogies is being questioned here which has relevance to this essay. See what you think. --Michael C. Price ^talk 23:48, 21 November 2009 (UTC)

Failed or what?

The fight over showing the "failed" tag, is really a problem with the use of general templates. I think we need to replace the "failed" template with a one-off template that more clearly describes what has gone on. I do not think that it failed as a policy, because it very clearly would never fit into the rather small set of wikipedia policies. It might have become a guideline, but it failed to reach consensus on that. It evolved to be an essay that some science editors might find useful. The template needs to state something along those lines and not just raise the temperature about this. --Bduke (Discussion) 22:36, 2 December 2009 (UTC)

Agreed. I removed the OR-busting statements that some people didn't like; to state that it still is a failed policy is false.--Michael C. Price ^talk 23:34, 2 December 2009 (UTC)

What you have done is not agreeing with me. I was suggesting a specifically written tag that is a compromise. As a proposal to create a guideline, this proposal failed, but it has made a useful essay. It wants something that explains that without the drama of a big read cross. --Bduke (Discussion) 06:06, 3 December 2009 (UTC)

I agree none of the current tags are appropriate and that a new one needs creating. My point above is saying why I think the last tag failed, as indeed the current one fails. --Michael C. Price ^talk 09:45, 3 December 2009 (UTC)

The essay "actually editing scientific articles"

In the summary it says: "This page in a nutshell: WP:ESCA is claptrap from people who have the WP:TRUTH"

When I tried to change this sentence to make it more neutral, I was told "edit your own essay". I don't think it is a good idea to stray away from constructive criticisms and resort to such language. Of course, my edits don't have to be accepted in that essay, but then they could have changed the wording in some other way. Also, they should address the choice of words in their essay. It now contains literal copies of entire sections of this essay.

As long as this situation remains the way it is now (i.e. not only does that essay contain these statements, but there is no willingness to address the problematic issues), one cannot leave out these facts. I.e. the essay is plagiarized from this essay and it uses the word "claptrap" when describing this essay. If people don't like it, they are free to edit that essay and rewrite it in their own words. Count Iblis (talk) 23:28, 23 April 2010 (UTC)

I reverted your change to this article that contained a personal attack against the authors of the other essay. Your comment about plagiarism here comes close to the same thing. What we write in Wikipedia, including in essays such as this one, is common intellectual property for others to make use of as they see fit, so calling it plagiarism is inappropriate. —David Eppstein (talk) 00:24, 24 April 2010 (UTC)

Facts are facts. The other essay does contain a personal attack and it is clearly plagiarized from this essay. Mentioning that here can hardly be called a personal attack. Count Iblis (talk) 00:28, 24 April 2010 (UTC)

Saying that the other essay is copied from this one is a fact, verifiable by looking at its edit history. Saying that it is plagiarized is a value judgement: it means that you think the authors of the other essay dishonestly intended to hide their copying. And it is a personal attack. —David Eppstein (talk) 00:39, 24 April 2010 (UTC)

Ok, I agree that it should be hanged to "copying". Plagiarism is too strong as it also means implies hiding this. Count Iblis (talk) 00:41, 24 April 2010 (UTC)