Wikisource:Requests for comment/Annotation policy

Request for comment about the annotation policy

Discussion to gather the community's opinion about the annotation policy.

All users are free to add new topics or points for discussion. These should be added as second-level headers as with the existing points (or just click the "Add topic" link).

All users are free to comment on or discuss each and every point. Please comment on each point in the appropriate "Comment" section.

First part (policy scope and deannotation)Edit


Policy scope

  • This policy applies to the addition, removal, or alteration of material in Wikisource works.
  • This policy applies to annotations by Wikisource users, not to annotations already present in published works (although it applies to their removal).
  • This policy does not apply to original translations, which are hosted in the Translation namespace and subject to the translation policy.
  • This policy does not apply to comparison pages (i.e., pages where different versions of the same work, either whole works or extracts, are placed alongside each other, either in series or in parallel, to provide a comparison between the different versions), which are banned.
  • This policy does not apply to the handling of published errata.


Deannotation (stripping the annotation out of a published version) is allowed in very limited situations. All of the following must apply:

  • No unannotated version of the work is known to exist. The work has never been published without annotations or the unannotated version is unavailable for other some reason.
  • The complete original version (with the original annotations) is hosted separately on Wikisource.

The following rules must be followed when deannotating works:

  • Any layer of annotations must be either entirely included or entirely dropped. Wikisource editors must not exercise an editorial voice appraising the appropriateness or validity of different parts of the same layer.
  • Errata must not be removed unless they refer to one of the layers of annotations that has been removed.

Wikisource users can choose to transcribe or not transcribe advertisements that are part of a larger publication. Failing to transcribe them does not count as deannotation.

Overall approach and structureEdit

The current draft of the annotation policy has the following structure:

  • excludes published annotated works and translations from policy scope
  • specifies that "Wikisource is under no obligation to host any type of annotation, based solely on the argument that it is out of scope for all other Wikimedia projects"
  • defines what counts as an annotation and what does not
  • bans comparison pages and purely decorative images
  • outlines three requirements for annotated works
    • annotated works must be labeled in the title and categorized as such
    • the unannotated version must exist before the annotated one is created
    • annotations must be factual, not interpretative
  • gives carte blanche about methods of annotating
  • explains what "deannotation" is and the limited situations in which it is allowed

Since this draft recently received general opposition at the Scriptorium, it seems that this approach is not ideal.

Personally, I do not like it. I would like a "Show/Hide annotations" button rather than the annotated version hosted separately. But this would not be easily attainable since it is difficult to agree on the methods of annotating—e.g., floating templates, footnotes, tooltips (for example, I would ban tooltips since they cannot be printed nor do we have proof that they display correctly on e-readers).

What would the ideal approach for an annotation policy be?--Erasmo Barresi (talk) 08:44, 16 July 2014 (UTC)

Comments (overall)Edit

  • I think the easiest places to begin are (1) the general scope of this policy, and (2) deannotation, as those are probably the least controversial and most easily handled sections of this draft. Those two sections are also unlikely to bleed over into the specifics of other parts of this policy. Then we need to decide what the overall goals we wish to achieve, and what we wish to avoid. Once we've agreed on that, the next step in my mind is to proceed with definitions. I have mixed feelings about how the definitions are currently handled, and suggestions to modify that, but I'll hold off opening that can of worms for now. --EncycloPetey (talk) 09:28, 16 July 2014 (UTC)
  • I wrote a lot of the current draft, so I'm not really in a position to comment on it much. However, it was my best attempt to codify the consensus reached in the last RfC, which was itself necessary as we were having little edit wars every so often due to different ideas of what the project "should" be doing; we needed an actual discussion and some written policies so everyone would be on the same page. Part of the reason both annotation and wikilink policies are drafts is that I was never 100% happy with my work, although I think they are a good-enough interpretation of the dicussion. A general theme from the previous RfC was a desire for purity and fidelity to the original source. That's why user translations were moved out of the main namespace (or "are being moved"; it's still on my to-do list). Restricting (and edging towards banning) annotations and wikilinks came out of that. - AdamBMorgan (talk) 13:55, 17 July 2014 (UTC)
  • I've added a link to this page to Wikibooks, as per my opinion that we should work with them to cover the range of annotated books that Wikimedia should cover.--Prosfilaes (talk) 19:42, 18 July 2014 (UTC)
  • I am opening six specific points now. Two are about policy scope and deannotation respectively, as suggested by EncycloPetey. One is about which types of annotation belong here and which belong in Wikibooks. Three are about things we want to ban (comparison pages, purely decorative images, and interpretative annotations), which I think are also pretty easy and uncontroversial.--Erasmo Barresi (talk) 10:34, 19 July 2014 (UTC)
  • Just a comment on the eBook question above with respect to tooltips. On my Sony eReader using ePub or pdf the display word(s) show up as highlighted links but there's no link behind them. Whether this is due to the ePub creation process here or just happens anyway I'm not sure. Beeswaxcandle (talk) 09:01, 27 July 2014 (UTC)

Policy scopeEdit

The current draft of the annotation policy says that published annotated works and translations are outside the scope of the annotation policy. However, I think the latter is inaccurate because we may also want to annotate translations.--Erasmo Barresi (talk) 10:34, 19 July 2014 (UTC)

Comments (scope)Edit

  • The policy, as currently worded, specifically omits translations original to Wikisource from THIS policy. Any restrictions to annotations of user-created translations will be covered by the Translations policy and NOT by this one. There is no need to try to cover user-created translations with this policy. --EncycloPetey (talk) 10:52, 19 July 2014 (UTC)
I disagree. I think that, once defined, rules for annotations should apply to all works, including translations (both published and user-made, without distinction).--Erasmo Barresi (talk) 12:36, 24 July 2014 (UTC)
It could be argued that, as an original work, user translations (in the Translation: namespace) are entirely annotated. It may also be useful to add translation notes to explain certain passages and what thinking or choices went into that. It could be useful to put any translation-specific instructions into the translation policy. That said, I am not really opposed to the annotation policy appying to the Translation namespace as well as the mainspace. - AdamBMorgan (talk) 16:01, 24 July 2014 (UTC)
If the rules for annotations are to be applied in the same way to both sorts of works, we're going to have a real uphill battle getting this policy finished. The general principle for already published works is that there must be a "clean" copy separate from the annotated version. If we try to apply that principle to user-created translations, then we're effectively requiring every user-created translation to be created in two copies in cases where the person intends to annotate the translated work. It would be a huge roadblock to translators, and would inhibit the very kinds of translations we've mostly seen to date. A user-created translation is fundamentally a different kind of hosted work from a previously published one. I just can't see making the same rules apply in both situations. --EncycloPetey (talk) 15:06, 25 July 2014 (UTC)
Really, I am thinking of annotations as templates that are added to the text during proofreading but do not show up in the transclusion unless one clicks a "Show" button. If you imagine things this way, having to comply with the same requirements would not be such a burden :-)--Erasmo Barresi (talk) 10:51, 26 July 2014 (UTC)
That's not what annotations are. Annotations include anything from links, to images, to additional commentary, translations of passages, and ANY changes to the original text. And the transclusion process does not work the way you seem to think it does. Your view would require additional technical changes that do not currently exist. Take a look at Translation:Romance of the Three Kingdoms/Chapter 1; this is an example of an annotated translation. It includes: added illustrations, added maps, parallel text sections, links to wikipedia articles, links to wiktionary entries, footnotes by the translator, and much formatting not present in the source text. All of these things are considered "annotations" and there is no way one could simply "show" or "hide" all of this with the click of a button. --EncycloPetey (talk) 06:38, 27 July 2014 (UTC)
I am pretty sure that there is a way to do so, maybe with the exception of parallel texts. However, I get your point. If my approach would prevent agreement, I am ready to renounce it.--Erasmo Barresi (talk) 13:04, 31 July 2014 (UTC)
  • In a separate issue of scope: we have already had a lengthy debate in which, broadly, the general concept of annotation was allowed by consensus but it must be light and clearly separate from the original material. I am assuming this remains the case? - AdamBMorgan (talk) 16:01, 24 July 2014 (UTC)
Yes, I think so.--Erasmo Barresi (talk) 14:34, 25 July 2014 (UTC)

Which types of annotation belong here and which belong in WikibooksEdit

Ban on comparison pagesEdit

There was no disagreement on banning comparison pages in the previous RfC.--Erasmo Barresi (talk) 10:34, 19 July 2014 (UTC)

Comments (comparisons)Edit

  • ...

Ban on purely decorative imagesEdit

Ban on interpretative annotationsEdit


The current draft of the annotation policy says:

Deannotation (stripping the annotation out of a published version) is allowed in very limited situations. All of the following must apply:

  1. No unannotated version of the work is known to exist. The work has never been published without annotations or the unannotated version is unavailable for other some reason.
  2. The complete original version (with the original annotations) is hosted separately on Wikisource.
  3. All annotations are removed, leaving only the unannotated work. Partial removal of annotations is not allowed.

--Erasmo Barresi (talk) 10:34, 19 July 2014 (UTC)

Comments (deannotation)Edit

George Orwell III and I had a protracted public discussion about this, and we were of the opinion that in a work with several layers of notes, one or more of the layers of notes should be removable for a particular version. For examples of layers: 1. a layer which discusses difficulties from the fact that the work is derived from different manuscripts which do not agree. 2. A layer discussing difficulties in determining the meaning of words written in an ancient language. 3. A layer summarizing the ideas or information advanced in the manuscript. 4. Cross references placed by the editor where the author of the work refers back or forwards in the text to the same ideas or information. 5. A layer explaining unclear ideas or words presented the work. 6. A layer of commentary by the editor.

We were of the opinion that either the whole layer or none of the layer should be included and that Wikisource editors should not exercise an editorial voice appraising the appropriateness or validity of different parts of the same layer.

I would just like to add that common sense should be applied to prevent the useless multiplication of versions simply because it would be possible to place various layers in different combinations. ResScholar (talk) 04:56, 30 July 2014 (UTC)

With the greatest of respect I understand the concept described, but yet have absolutely no idea from the above discussion as to how layers are intended to be implemented. If this question is out of scope please defer it until such time as appropriate. In any case I would appreciate an (eventual) answer (or at least perhaps, some pointers?) AuFCL (talk) 00:52, 31 July 2014 (UTC)
You just don't include the layers when you type them in. If you're planning to do various combinations, if you're smart, you do one with the fewest layers first, and then have a bot copy the less-layered version into a duplicate index, and you have a head start with which to compose the more-layered version. ResScholar (talk) 11:09, 31 July 2014 (UTC)
We were also of the opinion that duplicate .djvu (or similar) files should be stored at Wikisource, not Commons, and marked like "File:xxxxx (deannotated)" for example. ResScholar (talk) 20:42, 31 July 2014 (UTC)
Thank you. I'm a bit clearer on the idea now. So the "layers" may only be revealed by taking the differences between alternate publications? Has any thought been put into tracking which versions of a work relate to alternately-layer-combination ones (isn't this getting close to alternate versioning?) I suppose the point I am making is that at a sufficiently abstract level there is always going to some kind of entity which incorporates all available alternate layers. Sacrificing that would result in complete anarchy wouldn't it? AuFCL (talk) 23:03, 31 July 2014 (UTC)
Yes, it was assumed the de-annotations would be subtractions from a scanned source. One of my contributions from long ago deannotated a biography from a collected work when the biography's notes referred to page numbers of poems later in the work (and it wasn't an excerpt as it didn't just appear in that work), so on the rare occasion where scans aren't available, we might want to allow deannotations from non-scanned sources (very rare as typing in a non-OCRed work takes as much as 4-8 times as long as OCR).
As for Versions, the very origin of this de-annotation section was its status as an exception to Versions policy. Unfortunately for the policy as a whole, I only agreed with it, and I doubt I am alone, with the expectation that transcluded pages of works at Wikisource would be renderable with the "create a book" application or quickly placeable into an PDF file. I've tried the Wikisource ePub file tool, but that doesn't seem to render much of the formatting of the text, for instance, small capitals. ResScholar (talk) 02:37, 1 August 2014 (UTC)
So far tracking different de-annotations has been limited to the notes section of the headers of the works or the "translations" pages. They could easily go on the Author: pages as well. ResScholar (talk) 03:09, 1 August 2014 (UTC)
Agree as to the concept of layers. Sometimes, especially with ancient works, even the subdivision into chapters is an annotation (i.e., not part of the original work), but it is still useful to include it in a deannotated version. Is someone able to reword requirement #3 accordingly?
However, it seems that you are challenging requirement #2 ("The complete original version (with the original annotations) is hosted separately on Wikisource") by suggesting to do the deannotated version first and then the annotated one. I do not agree as to this.--Erasmo Barresi (talk) 08:49, 2 August 2014 (UTC)
This last raises an important point which I admit I had forgotten/glossed over/ignored in the discussion, and that is how are the various layerings chosen themselves going to be justified? O.K. so everything scan-based is legitimated; perhaps some footnotes are obvious from context/bibliography etc. (if present) But then everything beyond that has to be backed only by the editor's personal status, and haven't we more or less blanket ruled out self-publishing as justification?

I hope I have not over-simplified the situation? AuFCL (talk) 19:41, 2 August 2014 (UTC)

To couch this last more as an example, imagine if a scan-backed document contains an Errata section (I have a couple of examples in mind—let me know if you want links.) Ought the final transcription be presented as:
  • errata applied; or
  • Errata section present but uninterpreted so that a user may not even notice the text contains (legitimate/deliberate) errors?
I personally lean toward the first choice, but would like to know where others stand and perhaps a little on why they might so. AuFCL (talk) 19:56, 2 August 2014 (UTC)
The question of handling errata is a good one, but I'd be inclined to have a separate page that discusses the specifics of Errata. Perhaps a Help:Errata page that describes best practices. I do think that where published Errata exist, the fact ought to be mentioned on the Talk page of the Index (or main page of the work when there is no Index). Removing published errata from a work seems ill-advised, and whether applied or kept separate, it's part of the original work, and so wouldn't count as a "annotation" in my mind for purposes of this policy page. --EncycloPetey (talk) 21:18, 2 August 2014 (UTC)
Is hould clarify. I did not intend to imply the Errata section should be removed, rather I feel attention ought to be drawn to its presence through the application of judicious annotations and am interested in whether this attitude was true in others as well. For instance see this sort of situation A_History_of_Mathematics/Middle_Ages/Europe_During_the_Middle_Ages#errata.MA.E.134. AuFCL (talk) 21:54, 2 August 2014 (UTC)
Granted. However, I still think this should be treated separately from our Annotation policy, and that this separation should be explicit. That is, the handling of Errata should not count for "annotation" when the errata have been published, and only the removal of errata (as a Bad Idea) should explicitly and concisely be covered under the Deannotation section here. --EncycloPetey (talk) 08:06, 3 August 2014 (UTC)

Ban OnEdit

Second part (what to achieve and what to avoid)Edit

Goals to achieveEdit

What are annotations for? What should be their content? How should they be displayed?--Erasmo Barresi (talk) 10:30, 10 September 2014 (UTC)

Comments (achieve)Edit

  • ...

Drawbacks to avoidEdit

How can annotations be misused? What should they not include? How can we avoid that they look ugly?--Erasmo Barresi (talk) 10:30, 10 September 2014 (UTC)

Comments (avoid)Edit

  • We should avoid that original links (like the ones here) and user-added links become indistinguishable.--Erasmo Barresi (talk) 07:54, 13 September 2014 (UTC)
  • ...