Wikipedia:Footnote3/Lessons Learned

Footnote3 and has been used successfully in a number of articles. I'd like to list here the experiences from this.

Lessons from Experience

 * 1) Notes are needed even if we have a common citations database.
 * A book is a very big thing; even a single page can be somewhat ambiguous. Ideally each citation should show exactly which point in the source it is drawing on.  This means that almost every citation is unique, even if many citations use the same source document and can have bibliographic data separated out.
 * Example: in the Bombing of Dresden in World War II article, Philip Baird Shearer has been questioned about references and told that they were needed down to the page level.
 * Comments on the interpretation and use of a source may be needed. For example, Neo-Nazi sources tend distort the material the provide, but may also have to be referenced.  Without countext this type of reference can be misinterpreted.
 * 1) Tables need notes and need multiple references to those notes.
 * Comparison of file systems cannot be reasonably converted to my footnoting system as it now stands (I started trying then gave up). The fact that note 26 is referenced in the same way makes it easy to see which file systems have that common feature.
 * 1) It should be possible to have multiple sequences of notes.
 * The notes at the bottom of a table, which are about understanding, are not the same as citation notes. They should be kept separate if both citations
 * 1) Notes should be separate from references.
 * It does happen in a number of articles (e.g. Bombing of Dresden in World War II)
 * 1) multiple references may cover the same point.  It may be important to have all of them.
 * This happens, again in the Dresden article, but also in Islamophobia and some others. Where different opposing sources agree on a point that may be a very valuable fact.  This could cause serious problems for systems that use hilighting of text for their reference point.  Numbers, little symbols or
 * 1) I haven't yet found a ligitimate justification for using footnotes as a digression.  I always thought that a subsection in some article could be used better.
 * 2) Nobody can agree how notes should be formatted, but it would not be a problem to change later.
 * 3) It's very useful to be able to see all the notes together
 * if all the notes can be grouped together, then they can all be checked for consistency together and errors spotted more easily. This would still be mostly true if the grouping only happened in the displayed article and not the wiki text
 * 1) we can build a system which doesn't deviate much from simple wikitext
 * I'm a bit worried that systems proposed which use a citations database and so on will make it more difficult for editors to contribute. Anything which makes citations require skill will create a barrier for new users.  This system works, so future systems should also be sure to also be as easy or we should run two systems in parallel and more experienced editors should move information from one to the other.
 * 1) Numbered cross references are a nightmare
 * in fixing Bombing of Dresden in World War II it took me several hours to decipher the cross references since they didn't use simbolic cross references and needed to be verified properly. There is still no way I can be sure I did it right since I wasn't involved in the original writing of those notes.
 * 1) Almost any notes are better than none.
 * the person who first writes an article is normally in the best position to give sources. Hunting for sources is very difficult and watching how slowly the work of the WikiProject Fact and Reference Check goes even with a very strong collaboration, e.g. on Titan, shows that putting footnotes into use for citation is a matter of urgency.

Implications for future systems
A clear lesson is that in a future system there has to be support for note text not always being inline.

My proposal for the "Future Footnote System Which is to Come" is as follows


 * three commands (overloaded according to arguments)


 * ref(1) - one argument, a symbolic reference may be used multiple times
 * ref(2) - argument one, a sequence number, argument two as artument one in reference(1)
 * note(2) - two arguments, argument one, a symbolic reference matching the ref, argument 2 the footnote may be used only one time, second use is ignored (warning?)
 * note(3) - three arguments, argument one, a symbolic reference matching the ref, argument 2 the footnote may be used only one time, second use is ignored (warning?)
 * lnote(n) - note and reference all information together.

The first reference found should bind the footnote number to the symbolic reference.

Display modes for references should be as follows


 * invisible
 * small inline symbols
 * small numbers (default)
 * small numbers with square brackets
 * big numbers with square brackets and a symbol (like current URL) - for

And these should be directly configurable on the users wikipedia preferences.