Wikipedia talk:No original research

Numerical Comparisons in relation to SYNTH (in conjunction WP:NOTCOM, WP:CK, WP:BUNDLING)

Asking for clarity because I'm currently involved in a dispute where I'm seeing SYNTH applied to an extremely literal level that I've never seen before. Essentially I want to put forward an edit that says this in regards to viewing figures for the long running TV show Doctor Who:

"In consolidated ratings Lux received 3 million viewers (RadioTimes citation), this is lower than the previously recorded lowest-rating of 3.1 million recorded for Battlefield Part One (Doctor Who Complete History citation)(BARB data citation(s))".

For context the 3 million viewers figure is explicitly cited in Radio Times, while the previous low of 3.1 million is explicitly established by the latter two citations (DWCH up to time of publication in 2015, BARB cite(s) to encompass viewing data of years since). We also have two list articles on Wikipedia itself that contain in detail every comparable viewing figure for the show using these same sources (List of Doctor Who episodes (2005–present), List of Doctor Who episodes (1963–1989)).

However this edit has been bluntly opposed as a SYNTH violation because while it is accepted all figures themselves have valid citations, I have no citation saying that 3 million is less than 3.1 million ("You are stating Lux's ratings with a source, you are stating Battlefield's ratings with a source, and then comparing the two with no source that actually compares the two").

Now as far as I understand the guidelines and essays around SYNTH (per the three in the article title as examples) I don't need a source for such a trivial numerical comparison but this is being completely rejected on grounds that SYNTH is policy and those other items are not policy and therefore have no say on the matter.

Would it be possible to get some wider views on this strict interpretation of SYNTH please and whether it is acceptable or not? Rambling Rambler (talk) 23:43, 29 April 2025 (UTC)

Making comparisons that no reliable sources have made is WP:OR. If someone inserted a reference to Avengers: Endgame and said "Doctor Who is less popular than Marvel", that would really be an original observation from an editor. It's not a summary of accepted knowledge about the topic, which is the real purpose of an encyclopedia. Wikipedia isn't a place for editors for insert original analysis or comparisons, and it's best to follow what independent reliable sources say. Shooterwalker (talk) 13:02, 1 May 2025 (UTC)

If someone inserted a reference to Avengers: Endgame and said "Doctor Who is less popular than Marvel"

@Shooterwalker in that instance I'd agree, but the case here is that I have reliable sources for both statistics. So the only "original observation" is saying that 3 is less than 3.1.

Going by WP:BUNDLING, which shows this exact situation by using one source for the size of the Sun and a different source for the size of the Moon and then comparing them without a source explicitly doing that, and WP:CK, which states it is customary to accept "Mathematical or logical truisms ("1 + 1 = 2")" without a citation, this would therefore be an acceptable situation no? Rambling Rambler (talk) 14:13, 1 May 2025 (UTC)

When juxtaposing figures from diverse sources, I'd say there's a difference between the case where the numbers are objective, constant, enduring, tone-independent and pretty much source-independent and the case where two sources present figures that are inherently imprecise, time-dependent (as of when we're viewers-to-date counted?), possibly counted using different methodologies, that are, for those reasons, not necessarily comparable. Largoplazo (talk) 14:33, 1 May 2025 (UTC)

In this case the statistics are as consistent as can be understood as they are all derived from BARB, which is the UK's recognised body for measuring viewer statistics from broadcast media, and their standard reported viewership figures. So while there are three different sources cited, all three sources cited are reporting with statistics from the same authoritative body. Rambling Rambler (talk) 14:57, 1 May 2025 (UTC)

Also, for concerns regarding methodologies and potential inaccuracies, I'd say that's not really important here as that is an issue of assessing source reliability which isn't a factor as all sources are considered reliable. So essentially I'm taking two reliably sourced figures and saying "the figure reported here is less than the figure reported here". Rambling Rambler (talk) 15:05, 1 May 2025 (UTC)

Chipping in here, but the reason I'm a bit opposed to it is because there's no source directly comparing them, primarily because, if you aren't familiar with the subject area, it's wholly possible an episode could have aired between Battlefield and Lux that had lower viewing figures. As those familiar with the show, we know that's not the case, but for verifiability's sake, we need that comparison so we aren't making any original rationalizations that could potentially be incorrect (Both per above and for the uninitiated observer) Magneton Considerer: Pokelego999 (Talk) (Contribs) 02:32, 2 May 2025 (UTC)

I think that comparisons are actually important, but they need to be reasonably obvious and they need to be reasonably sound (e.g., not saying "Paul Politician polled 55% approval before his big speech and 65% after, so the speech improved his popularity" if the first poll was an internet survey and the second one only surveyed people in-person as they were leaving the event).

Rambler, my impression of the facts is that you have the same (Barb Audiences) survey system/tool for both numbers. In the spirit of WP:Let the Wookiee win, can you find a source that mentions both numbers on the same page? It might be faster to find a source than to argue about it. WhatamIdoing (talk) 03:42, 2 May 2025 (UTC)

@WhatamIdoing there is one source that used to be used a lot that contained every single rating including just aired ones, but that has since been deemed a blog and therefore unreliable. It'd be of course helpful if the published source (Doctor Who The Complete History) was still going but that finished in 2018 and hasn't been revived.

The best we can do is cite BARB statistics but using multiple publications for this. Rambling Rambler (talk) 10:10, 2 May 2025 (UTC)

Was Doctor Who The Complete History a website? If so, did it contain this information, and is it available in an archived form? WhatamIdoing (talk) 16:28, 2 May 2025 (UTC)

It was a printed publication, pretty in-depth. But yes it did indeed maintain historical broadcast ratings data (which was derived from BARB). |For example here is the link to the IA version of the publication, it's on page 133 of issue 45 that states Battlefield Part One was the lowest broadcast ratings at time of publication.

It's what is now used to establish most listed ratings in the List of Doctor Who Episodes pages. Rambling Rambler (talk) 17:53, 2 May 2025 (UTC)

Ah. The problem is that the magazine making the comparisons between viewership of different episodes was published in 2015, and Lux (Doctor Who) was broadcast in 2025. WhatamIdoing (talk) 22:18, 2 May 2025 (UTC)

That's why the BARB database is cited. Basically you could even avoid the RT source and just have the two sources. The Magazine is cited to support the statement that according to BARB statistics Battlefield Part One was the least watched episode, and then the second source for BARB carries all the ratings since 2015. Given both are citing the same ratings authority it isn't mix and matching sources. Rambling Rambler (talk) 23:05, 2 May 2025 (UTC)

@Pokelego999 is that not the case with any subject or subject area though? What really matters here though according to the guidelines is that would we expect an average person, even without knowledge of the subject, if presented with these sources to be able to understand that 3 is less than 3.1 even if neither explicitly says so? And here I think the answer would be a resounding yes. Rambling Rambler (talk) 10:02, 2 May 2025 (UTC)

I will not be responding further to replies after this one, as I've explained this at the episode talk page as well, but what is also worth noting is that for "Battlefield", this was a broadcast episode with a first-aired rating on television from 1989. With "Lux", the episode's first release was its release on iPlayer at 8am. Its second release is its broadcast on BBC eleven hours later at 7:15pm. Consolidated viewers refer to the airing of the episode, as per BARB itself. BARB lists "Doctor Who" as "DOCTOR WHO (SAT 19:15)", which is from its second release at 7:15pm, and takes into account all viewers within the seven days from that airing - that does not include all viewers who viewed the episode between its first and second airing. BARB stating that "Viewing figures for programmes on SVOD services are calculated based on the average audience achieved during the first seven days of their availability" does not indicate that SVOD (iPlayer) is included within BBC One's rating or whether they're counted separately outside of the top 50; that appears to be assumed. It is completely inappropriate to compare two differently measured statistics. -- Alex_21 _TALK 04:19, 2 May 2025 (UTC)

This is just flat out wrong, as has been explained to you multiple times now.

Per your own link to BARB Viewing figures for programmes on broadcast channels are consolidated 7-day viewing figures with pre-broadcast viewing and viewing on tablets, PCs and smartphones included.

So yes, it does include iPlayer viewing from the moment it becomes available as part of the consolidated rating.

And iPlayer is not SVOD. SVOD means subscription services (Netflix, Disney+ etc), BBC iPlayer is not a subscription service in that sense. Rambling Rambler (talk) 09:50, 2 May 2025 (UTC)

You know what this sounds like to me? A great opportunity for the two of you to find some reliable sources about what BARB 'counts', and to expand Barb Audiences with that information. WhatamIdoing (talk) 22:22, 2 May 2025 (UTC)

That's a great idea. A lot of assumption going around here, as to what viewer ratings counts, and what our readers already assume (below). As you yourself said, if a source was given that compares the two on the same page, then this need for this whole discussion ceases to be. Unfortunately, one doesn't seem to be able to be found, in the face of a consensus of the SYNTH application here. -- Alex_21 _TALK 02:05, 3 May 2025 (UTC)

Also as to BARB lists "Doctor Who" as "DOCTOR WHO (SAT 19:15)", which is from its second release at 7:15pm. This is only done as a descriptor using the linear airing day and time to allow differentiation between ratings for shows where multiple episodes air every week or even on the same day (namely soaps like Emmerdale or Eastenders). Rambling Rambler (talk) 10:06, 2 May 2025 (UTC)

IMO this is deep into wp:or / wp:synth. It contains implied statements / deprecation of the subject which is controversial; (the latter being an unwritten part of the working definition) created solely by the Wikipedia editor. I said "created" because one can select and use statistics to say or imply anything that one wishes to). Sincerely, North8000 (talk) 21:13, 2 May 2025 (UTC)

I wonder if that's really true in this case. We have a source that compares most of the television episodes. Is continuing a simple and obvious series really something "created solely by the Wikipedia editor"? WhatamIdoing (talk) 22:21, 2 May 2025 (UTC)

Not only that but I don't see how this use of the data is inappropriate when it's accepted on the List of Doctor Who episode pages to average out the ratings data for each series despite no explicit source existing to support that average viewership claim.

If it's acceptable to take a range of sourced values and average them yourself, how is it inappropriate to say which is the lowest one? Rambling Rambler (talk) 23:08, 2 May 2025 (UTC)

It's quite literally taking the figures from the ratings authority in the UK and saying which verified value is the lowest. I'm not altering or manipulating them via any sort of mathematical operation (averaging, dividing etc), it's just going "this is less than that". Rambling Rambler (talk) 23:12, 2 May 2025 (UTC)

Regarding inconsistencies and flaws in the rule

Hi there, I know this is hard to comprehend but in regards to original research, I would like to address some very real and very major concerns regarding this. I'm not sure if you are aware but in various communities, there are times where certain topics aren't really covered by many sources due to either general ignorance or hatred of certain topics. Essentially, wanting people to ignore stuff. The job of an editor should be, in my opinion, to find information on sources no matter if they do the research themselves or use a source that is reliable. Using a source has its perks but sometimes, some topics that journalists cover like in the professional wrestling industry, like various promotions, are often given stub articles or are ignored. I speak on behalf of the fans of Juggalo Championship Wrestling who feel as though they have been ignored in the industry and the pro wrestling media despite the promotion's many accomplishments. Some people like me often go out of their way to find information from people who are very familiar with various topics and know what they are talking about. Please, if any admins see this, I would like for you to take this into careful consideration. Unknownuser45266 (talk) 06:50, 10 June 2025 (UTC)

• What you asking is never going to happen. The basic premise of Wikipedia is not just not original research, but WP:Verifiability and WP:Neutral point of view. NPOV "means representing fairly, proportionately, and, as far as possible, without editorial bias, all the significant views that have been published by reliable sources on a topic", while verifiability means that "content is determined by published information rather than editors' beliefs, experiences, or previously unpublished ideas or information." Original research isn't allowed because it goes against the basic principles of the encyclopedia. If you want somewhere to publish such information Fandom doesn't have the constraints that come with Wikipedia. -- LCU ActivelyDisinterested «@» °∆t° 09:48, 10 June 2025 (UTC)

  Plus, pro wrestling is stupid. EEng 10:25, 10 June 2025 (UTC)

And editors themselves come with their own biases. Take the notion of reliability in sources. Like most other editors, you're anonymous. We have no idea whether your research is reliable. We've no idea what credentials you may or may not have. We've no way of knowing whether or not you're making it all up. (Heck, take a look at my own user page. I make a number of assertions about my own background and experience. And you've nothing other than my own word backing any of that up.) This is why we rely on reliable, independent, third party sources with reputations for fact checking and accuracy. This doesn't often apply to professional wrestling, an industry famous for exaggeration, mythmaking, storytelling and putting themselves over. Ravenswing 11:29, 10 June 2025 (UTC)

Not to mention that pro wrestling is stupid. EEng 16:00, 10 June 2025 (UTC)

and here you go, admitting your own bias against professional wrestling. Unknownuser45266 (talk) 20:03, 10 June 2025 (UTC)

Guilty as charged. EEng 21:21, 10 June 2025 (UTC)

You might find something interesting at Wikipedia:Alternative outlets. Gråbergs Gråa Sång (talk) 11:33, 11 June 2025 (UTC)

One more shortcut

"Synthesis of published material" has one more shortcut: WP:SYN. This is currently not named in the shortcut(s) box of the section. Should it be added? --Himbeerbläuling (talk) 14:09, 13 June 2025 (UTC)

SYN doesn't seem as immediately intuitive. CMD (talk) 16:24, 13 June 2025 (UTC)

Additional context needed, from WP:SHORTCUT, to explain why a given shortcut might possibly not be added: The point of these template boxes is not to list every single redirect for any given page (that's what Special:WhatLinksHere is for). Instead, they generally should list only the most common and easily remembered redirects. On the other hand, I'd think that with only one entry in the box, "SYNTH", and with "SYN" being intuitive (in my opinion) and easily remembered, it would make sense to add it. On second thought, a shortcut is best if its name already communicates to the reader of the text in which it's included what its target will be about. "SYNTH" is more obviously going to be about synthesis than "SYN" is, which could also be about synchronization or synergy. "SYNTH" is the better of the two, so there's really no good reason to advertise "SYN" as well. Largoplazo (talk) 20:01, 13 June 2025 (UTC)

Is this an unreliable source?

Using a source directly created by the subject of an article about a dead person for that article. I know this is not allowed for living people. Faster than Thunder (talk | contributions) _{Tamil speakers: Contribute here} 23:25, 27 June 2025 (UTC)

Is this the right place to ask this?

Aren't there more relevant pages to discuss it, like the article itself, or the reliable sources page, or the biographies of living people page?

Moreover, I just joined, but from what I can see you can use sources created by the subject only as evidence that they created it and might hold such views. The content itself is not considered authoritative unless it is verified by a trusted third party. Though I could easily be wrong. Mara.Namuci (talk) 23:41, 26 July 2025 (UTC)

Original Research is unavoidable

The rule as stated makes no sense. I believe I understand the intent: to prohibit people from writing opinion pieces, making political arguments, or advancing controversial ideas, which would transform wikipedia from an encyclopedia into a debate platform.

However, logically, the process of finding verified sources and weaving an accurate description of a topic is itself original research.

I feel that this contradiction arises from the use of the term in academia to refer to a study that uses new data the authors have compiled. The authenticity of the data and the collection or analysis methods are unverified until they undergo peer review. While unoriginal research is then considered to be work that only cites other material.

But these definitions only apply to scientific trials, not to data in general. The problem is obvious when sources, links, and citations are considered data.

The examples given about Paris and routine calculations are demonstrative of common knowledge, in other words consensus or authorised positions. From this it is clear that the true meaning of "No Original Research" is actually "No Subjective Research".

Standard logical deductions or easily verified statements that any rational adult would agree with are not subjective.

Whereas claims based on inaccessible resources or flawed reasoning and spark controversy are and don't belong in an encyclopedia.

I'm under no illusions, I know this post won't inspire any changes, but I felt it was worth writing for the sake of those who notice the contradiction and were confused by the inconsistent and arbitrary application of this rule, and its cousin original synthesis.

I've seen a lot of discussion and criticism of the rule, but nobody seemed to concisely explain the problem.

For an illustrative example, suppose there was an easily accessible source for a 100,000 page book that takes 3 years to read. The person who reads it all and summarises it is conducting subjective research. Nobody else is likely to verify their summary because nobody has that much time.

In conclusion, finding, reading and summarising sources is research. Mara.Namuci (talk) 01:49, 27 July 2025 (UTC)

You seem to be confused regarding the scope of the policy, though it's made quite explicit in its first sentence. Articles ideally contain exactly what information we glean and balance from our best sources, and nothing else. Unless your point is that avoiding novel claims when synthesizing existing material is literally impossible, and every contribution has phantoms of our own inexpert opinions attached, then not much of the above has much meaning. Remsense 🌈 论 01:55, 27 July 2025 (UTC)

Though, I want to emphasize you're off-base as regards WP:BLUE, in my opinion. The point is that idealized rational adults in your or my imagination aren't the audience we're writing for, we're writing what's meant to be a resource of the highest quality possible for everyone on Earth. It's not about what's obvious to you personally—it's about making sure that absolutely every claim of substance can be verified by any readers who would need or want to do so. Remsense 🌈 论 01:58, 27 July 2025 (UTC)

Thank you for taking the time to respond.

I will clarify my position.

I am not suggesting that it is about what is obvious to me personally, rather its the opposite. "Subjective Research" refers to conclusions or beliefs that cannot be verified that may be based on inaccessible resources.

The people responsible for verifying are the editors not me.

So to describe it in practical terms, if most editors think that something is either wrong or not obvious to them, then it is subjective and needs citation.

If certain editors disagree with the way the citations are arranged, interpreted or summarised, then it is subjective.

So it must be reorganised until it is agreeable with the majority of the editors overall belief systems and world views.

Calling the problem "original research" or "original synthesis" obscures the reality, which is simply that the editors as a collective don't believe the conclusions follow from the available data.

It is comparable to a situation with a scientific journal refusing to publish anything that their scientists cannot independently reproduce.

Replace the scientists with editors, and the lab with a home computer. The only tools they are allowed to use are the internet and their brain.

Furthermore, it isn't about whether the readers can verify the statements. Since many readers could well disagree with the sources on the wiki page. It is only about what the editing team feels is appropriate, neutral and authoritative.

I hope that clarifies it. Mara.Namuci (talk) 02:43, 27 July 2025 (UTC)

Wikipedia contains (and should contain) a lot of subjective information. Strictly speaking, subjective is about how a view depends on your situation. I know a young man who can lift 600 lbs (270 kg). A school-age child is "light" according to him, and "heavy" according to me. Neither of us is wrong; we just have different perspectives on the effort needed to lift that amount of weight.

I often find that when people talk about "subjectivity", they actually mean opinions. Wikipedia needs information about opinions, too. That is most obviously necessary when writing about cultural works. The article about David needs to have some universally agreed-upon facts, such as "5.17 metres (17 ft 0 in) tall" or "made from Carrara marble". But it also needs statements that are opinions, like "feeling of energy" and "a political symbol", even "the figure's erotic aura". You haven't properly described a piece of art if you leave out the viewers' emotional response to it.

"Subjective Research" is most certainly not anything that is "based on inaccessible resources". You may wish to read Wikipedia:Reliable sources/Cost and the WP:PAYWALLED policy. WhatamIdoing (talk) 02:48, 27 July 2025 (UTC)

Right, I meant to state this but got tangled in my reply. Objective statements (the precise value of the fine structure constant) and subjective statements (Huckleberry Finn is an anti-slavery novel) are often equally verifiable. In fact, the latter subjective example is easily verifiable, while the former will probably never be. Remsense 🌈 论 03:07, 27 July 2025 (UTC)

I agree that articles contain subjective statements.

But as you said, subjective things depend on or change according to who considers them.

Whereas objective claims are matters where nearly everyone agrees about it and have reached a consensus.

In most cases where emotional or subjective claims are found on wikipedia, they are supported by an authorised soruce. It doesn't matter that nobody can independently verify it if the government or a large media group is making the claim.

If the BBC claims that there is a war in some foreign country, we can't actually verify it personally. We have to trust someone.

In short, authorities are able to make subjective claims about events. Contributors must stick to what is available on the internet.

And your example of quoting other's opinions is not itself subjective, if it is merely reporting that someone had an emotional reaction.

"He thought it was wonderful" is different from, "It was wonderful". Mara.Namuci (talk) 03:11, 27 July 2025 (UTC)

Editors are not required to "stick to what is available on the internet". Whatever gave you the idea that books and other "dead tree" sources are unacceptable? We even have a specialized template for {{cite sign}}, for the purpose of citing real-world signs (historical markers, museum placards, etc). WhatamIdoing (talk) 03:23, 27 July 2025 (UTC)

If the BBC claims that there is a war in some foreign country, we can't actually verify it personally. We have to trust someone. we won't necessarily report it. You seem to forget that our policies don't work in isolation. M.Bitton (talk) 03:23, 27 July 2025 (UTC)

There is no trivial path to reliable truth and there never will be. Our core content policies essentially split up the process of getting there the best we can as an encyclopedia anyone can edit.

You're imposing this dichotomy of subjective/objective that doesn't reflect either policy or how those terms are generally used to my knowledge.

"Subjective Research" refers to conclusions or beliefs that cannot be verified that may be based on inaccessible resources.

If it helps, that's what we mean by "original research", when it makes its way into live articles.

The people responsible for verifying are the editors not me.

This is a false dichotomy, and it's just not close to true. Editors are expected to provide clear, complete citations. They don't have to unless challenged, from which point some level of consensus building greater than the individual will help decide, or take over the question entirely (e.g. with an RfC). The point of editors citing their sources is that others in turn can check their work and continue their study of the subject if they so choose.. That is why we require reliable sources to be published publicly.

Calling the problem "original research" or "original synthesis" obscures the reality, which is simply that the editors as a collective don't believe the conclusions follow from the available data.

We define the term clearly. We didn't pull it out of thin air, it is pretty consistent with what professional scholars consider "original research". We are not necessarily experts, and the consensus building process is indeed fundamental to how this project is capable of functioning to begin with. I'm not sure what the alternative is—are you under the impression we could somehow operate as if human language has precise, definite, objective meaning? Plato already knew this was a dead end.

Your latter statements were admittedly not at all clear to me. Your concerns seem to concern WP:NPOV and WP:RS as well to a significant degree, so maybe getting a sense what each policy says and how they work together would help you. Remsense 🌈 论 03:04, 27 July 2025 (UTC)

Note However, logically, the process of finding verified sources and weaving an accurate description of a topic is itself original research. It's research, not original research. The sources we cite (or the sources they rely on) did the original research. We aren't reporting news stories live, reporting on our own scientific or sociological research, or polling movie audiences about how much they liked a film. Largoplazo (talk) 02:38, 27 July 2025 (UTC)

suppose there was an easily accessible source for a 100,000 page book... because nobody has that much time. they will summarise the various claims within the book and provide the page numbers (especially when challenged) for each one (so that others don't have to read all of it in order to verify those claims). M.Bitton (talk) 03:17, 27 July 2025 (UTC)

Thank you for addressing that thought experiment. However, I have to point out that individual pages are likely not enoguh to prove that a summary is accurate.

Especially for a story, treatise or report.

I suspect you would need multiple quotes from hundreds or even thousands of places to reliably confirm that the text is consistent with the description.

Which definitely becomes unwieldly. Mara.Namuci (talk) 03:34, 27 July 2025 (UTC)

This is where WP:V comes into play. M.Bitton (talk) 03:38, 27 July 2025 (UTC)

If it's an accurate summary of the book's contents, it's not original research (=stuff made up by editors that isn't in a reliable source). If it's not an accurate summary, then it might be original research (only might be, because what someone accidentally pasted the wrong source at the end of that sentence, and the information is merely misattributed rather than not being in any reliable source at all?).

But your questions suggest that you are expecting source checking to be good, fast, and cheap. In reality, we don't care whether it's fast or cheap. If it costs you months of reading or thousands of dollars (or both) to check whether the Wikipedia article says the same thing as the cited source, then that's okay with us. WhatamIdoing (talk) 06:13, 27 July 2025 (UTC)

This thread brings up some interesting topics - but will unlikely change the long standing practices. Perhaps my personal approach which I tell my students in classes helps a bit here to disentangle the different types of original research here. I tell my students there are 3 types of new information they can bring into a paper and each of them needs backing.

(1) Selected topic relevant information based on reliable sources. Of course the selection of sources, and which sources get emphasis remains subjective and to some extent an original approach to the paper (but that is part of what students should learn). Important here is that the information is verifiable (WP:V) from reliable sources (WP:RS) but need not necessarily be true. Well established laws of nature (such as Newton's laws) are gross oversimplifications and hence not true (when achieving speeds close to light speed), yet they can and should be verifiably cited from reliable sources if relevant in the context.

(2) Self collected new data (for which the data files can be made available). This is truly original (primary) research and something Wikipedia explicitly excludes from inclusion (for good reasons)

(3) (New) Conclusions based on logical reasoning in combining information from sources and/or own data. This is synthesis and also something Wikipedia excludes (WP:synth for good reasons). We should not combine information to come up with new ideas posed as original conclusions, speculations etc., as that is not the purpose of an encyclopaedia. However, the selection, ordering and narrative flow of any written text always implies some kind of synthesis (this is unavoidable).

When reading this thread it seems that the discussion alternates between the verifiability of facts from reliable sources (which is in my view fairly straightforward), the selection of sources (which is unavoidably subjective), and to some extent the implied or explicit synthesis of these sources. The subjectivity and selecting sources and the implied synthesis by that selection and the order of argument in a narrative flow is unavoidable but hopefully the consensus model of Wikipedia takes off the sharp edges (the existence of templates like template:Globalize indicates this sometimes needs additional attention but we do our best). Arnoutf (talk) 07:40, 27 July 2025 (UTC)

Thank you for your thoughtful message.

This is the kind of reply I hoped for and your breakdown of the different types of OR is very helpful.

I agree with everything you have said. However, I would argue there is a fourth implicit OR, which is knowledge based logical deduction and synthesis from zero or more sources in the construction of a narrative.

Although you may have been including this as part of your 3rd type.

A simple example being recognition of the chronological flow of time. Sources which discuss events in no particular order are composed to form something readable.

With more convoluted examples being translation of texts written in obscure languages, interpretation of complex or abstract concepts, and knowledge of jargon or contextual implication in legal documents or historical works. There is also the example already given of an immensely long text that a committed reader has transformed into a mental representation before offering a summary of their impression. Mara.Namuci (talk) 16:57, 27 July 2025 (UTC)

You may find it helpful to read the explanation at Wikipedia:These are not original research.

Keep in mind that we are not concerned about the One True™ Meaning of the words original research. We are concerned with what we choose to ban from articles, and what we choose to not ban from articles. For our own convenience, we call the banned stuff "original research", but we could equally well call it "Policy 3" or "egoirh worgi" or "the banned foo". What matters is the thing itself, not the label. WhatamIdoing (talk) 17:34, 27 July 2025 (UTC)

I don't think the selection of RS nor their presentation is that subjective. Both are objective to the extent that they attempt to apply standards, not of the editor's internal feelings, but standards that are laid out for them externally. Indeed, it is that sense of objectivity that means that anyone can come along and write "your" article. And they will. -- Alanscottwalker (talk) 17:35, 27 July 2025 (UTC)

It is perhaps not an opinionated choice, but it is subjective in the narrow sense that it matters what language(s) you're searching in, whether your search engine is giving you personalized results, what resources are available to you, etc. You and I could apply exactly the same standards, and still end up choosing somewhat different sources because of differences in location, past search history, etc. WhatamIdoing (talk) 18:54, 27 July 2025 (UTC)

Well sure, you can say in some limited way. But that's not at all responsive to what I said, which was "that subjective". And it's not that I am trying to adress what I feel, it's that I am trying to address the external standard. I assume you are too. And it is the external standard on which it will be judged, not how you or I feel subjectively. And the external standards are both broad and narrow. It's as broad as 'this is an encylopeidia', if that's so, than you better not be writing an impressionistic tonal poem, and the external standards get narrower and narrower. (And much of what you mention is not about subjectivity, it is about how capable you or I are in meeting the standard. True, one may not be as good at it as others, but that's beside the point. Still, if you do sufficiently poorly, it will be excluded.) --Alanscottwalker (talk) 19:58, 27 July 2025 (UTC)

The detail that I think is frequently lost is that subjectivity isn't about "how you or I feel". Subjectivity is about legitimate, logical differences based on how each 'subject' is situated. A statement like "The Treaty of Versailles stated that Germany would pay reparations to France" is universal and therefore non-subjective. A statement like "Many Germans saw reparations as a national humiliation" is an encyclopedic, source-supported description of subjective facts. The reasons that this is subjective is because some people (Germans) saw reparations as a national humiliation, and some other people (e.g., French people) saw it as justice. It's not just "opinions"; it's seeing the same facts from opposite sides of the border. WhatamIdoing (talk) 00:48, 28 July 2025 (UTC)

OK. Let's backup. I assume you don't mean that the definition of subjective is subjective. That the definition is the same for you and me and it's something we don't construct internally, at least at base, it is what we read in a book not authored by either of us. And what we read there shows that subjective has different facets but those facets are not infinite, nor constructed by us. Regardless, of whether that definition of subjective references internal or personal feelings, opinions, or tastes or whether your "isn't about" was overstatement, perhaps it is telling that you chose "humiliation" as your example (which I assume we would agree is a feeling).

You and I have no idea what the Germans or French thought or felt or why. Nor do you or I know whether it was logical for them. The best we can both do is read about it from the external sources and try to find the best sources for that, and then be faithful to those sources in our writing (not faithful to ourselves). But perhaps a more useful way to discuss what you describe is "point-of-view", and we have external standards for talking about, and the sourcing of, others' points of view (NPOV). Note first, they belong to others not to us and they are the same for both of us, then too, the standards are the same for you and me. We also have external standard about writing up our own points of view: don't. So in those senses, it is not subjective to you nor to me - it is objective for both of us. Alanscottwalker (talk) 11:54, 28 July 2025 (UTC)

Maybe the example I used in User:WhatamIdoing/Subjectivity in Wikipedia articles (unfinished) would be simpler. Shall we take this to my talk page? WhatamIdoing (talk) 16:01, 28 July 2025 (UTC)

Ok, but if your point is a scientifically supported cause and effect can be treated as fact, I would note that neither the support, nor the cause, nor the effect, nor how those are put together is subjective to me or you - they are objectively the same for both of us. Alanscottwalker (talk) 16:28, 28 July 2025 (UTC)

Unless you have turned into a gray-haired old lady without telling me, I doubt that "How cold is too cold to go swimming?" is "objectively the same for both of us". WhatamIdoing (talk) 18:06, 28 July 2025 (UTC)

That question is exactly the same for both of us, no matter how old you are or what gender/sex. If you are suggesting the answer is not the same, you'll have first convince me my answer matters to any article (and you won't be doing so without sources). And that's before you even get out of me my age, hair color, or sex. -- Alanscottwalker (talk) 18:19, 28 July 2025 (UTC)

OP is completely correct that what we call "original research" is not the same as what that phrase would mean in academia. There are plenty of similar examples in Wikipedia policy. An outstanding example is "verifiable", which in plain English means that its truth can be checked, which is different from our meaning of appearing in a "reliable source". And "reliable" is another example. In all these cases, our meaning is related but not exactly the same as the general meaning outside Wikipedia. We can't do anything about this except to try to write policy pages to explain what these words mean to us. In this case the policy attempts to do that right at the beginning. Zero^talk 02:20, 28 July 2025 (UTC)

I am not so sure the Wikipedia view is that different from the view in academia. In academia, original research (be it new data, new synthesis or new opinion) is key to everything we write - why else write it. In Wikipedia we should not do original research but create an overview of reliably sourced and/or verifiable findings (the sky is blue need no sources as it can be verified by all by simply looking up). So in my view it is not so much the definition that differs, but whether original research is desirable or not that makes the big difference. Arnoutf (talk) 07:13, 28 July 2025 (UTC)

As I see it.

The similarity between them is that the scientist conducts original research when their work uses unpublished and thus unverified data, and the wikipedian is doing OR when they use unverified sources.

However the first difference is that when a scientist wants new data he can mine it from experimental results.

Whereas the wikipedian's data is reliable sources, which they mine from the internet.

Furthermore, journals consider the submitted body of work as OR if the data is new to them. They make no claims about whether it is new in any other sense which is unknown and irrelevant.

However, when a wikipedian proposes that new sources be added to the body of verified and reliable sources, despite being new to wikipedia, this is not considered OR because wikipedia makes the illogical and circular argument that OR only refers to data that is new to their body of verified sources including that which they just verified.

Such that OR only refers to those sources which are not accepted and thus deemed unverified.

Alternatively, some may claim that it is because the source data was already known to the relevant field of study that it doesn't count as original. But as mentioned, this isn't a reasonable definition as it is was previously unknown that it was well known, and it all hinges on the claim of reliability. Which is why nobody defines it that way. It is original if it is original in the eyes of the adjudicating body.

Moreover through this process, sources are retroactively implied to have always been considered reliable and that no change has ever occurred. Which is pathologically insane.

The equivalent would be if a journal verified a paper and then made the absurd claim that since the data is verified and published that it is not and has never been Original Research.

So we see that the concept is being used inappropriately in order to maintain the fiction of objective reliability and avoid the reality of appeal to consensus and appeal to authority.

To provide an illustrative example. Suppose that in an alternate world, wikipedia is based in china. And an american attempts to update a page about his country according to news sources in his area, they may deem these sources unreliable and flag it as original research. Which makes no sense. Mara.Namuci (talk) 14:17, 28 July 2025 (UTC)

I'm finding it extremely hard to follow you. What is it exactly that you're trying to achieve? M.Bitton (talk) 14:45, 28 July 2025 (UTC)

It is indeed hard to follow. For publishing academic papers it is always important to claim a relevant knowledge gap / novel contribution to literature - i.e. to claim an original contribution. Whether this is self collected data, a newly develop method to analyse data, or even synthesis (by creatively combining existing sources in e.g. a meta-analysis or theory construction paper) is irrelevant in academia (I can claim some experience there with well over 100 scientific papers to my name). What Wikipedia should not do is make novel contributions and that is basically how we define original research. Arnoutf (talk) 14:52, 28 July 2025 (UTC)

I think, OP, you have misunderstood WP:OR. It's nothing to do with: the wikipedian is doing OR when they use unverified sources. Nor is it true that wikipedia makes the illogical and circular argument that OR only refers to data that is new to their body of verified sources. We talk about it being "verifiable" (in WP:V) not "verified". There isn't a body of "verified" sources that can be added to. The key issue is whether or not it has been published (whether or not on the internet. I don't know why you say the wikipedian's data is reliable sources, which they mine from the internet. It can be "mined" from anywhere.) If someone wants to make a claim that has not been published - that's OR. If they want to make a claim that has been published then that's potentially not OR, although then the question of the reliability of that published source has to also be addressed. Because you seem confused about the difference between published and unpublished I can't relate anything you have said to Wikipedia in any practical way. DeCausa (talk) 15:02, 28 July 2025 (UTC)

I would add that sources are not data, if they were just that, we could not write anything at all, at least not anything coherent; they are context, analysis and interpretation, the very things we can't do originally (along with collecting data). -- Alanscottwalker (talk) 15:37, 28 July 2025 (UTC)

I would suggest, it's adapted to our situation rather than something very different. Alanscottwalker (talk) 12:34, 28 July 2025 (UTC)

WP:OR and wp:Ver simply place a requirement on material which excludes material which doesn't meet certain requirements. The OP correctly points out that this still allows for a lot of creation and creative spinning by editors via other methods. But this is not technically WP:OR and thus not something that would be excluded by WP:OR/WP:Ver policies. North8000 (talk) 14:51, 28 July 2025 (UTC)