Talk:Beta Features/New Features

Latest comment: 6 years ago by Quiddity (WMF) in topic Historical?
edit

I don't like this implication that wikifying is about (over)linking. --Nemo 13:13, 18 March 2014 (UTC)Reply

I'd added a note previously, which said: "Quick note: This would have to keep w:WP:Wikipedia doesn't use Allwiki in mind (a very old decision, connected to WP:Build the web c.2002)"
Then there was an edit-summary comment from Jaredzimmerman (WMF) when it was removed: "Interesting note, but unsigned, and references a 10 year old decision, if its an issue it will be brought up on the feature page when/if work starts on it."
So, there's that for reference.
Also, en:Wikipedia:MOS#Links and en:WP:MOSLINK are the relevant Enwiki styleguides.
Possibly we're both misinterpreting exactly how Jared envisions this feature working? As I understand it, there would be a problem with inaccurate targets getting linked, and the reader having to intuit the difference between editor-selected wikilinks and automated n-gram links. Eg. in en:Flag of North Korea, the phrase "The white stripes symbolize purity" might (would?) get mis-linked to en:The White Stripes.
There would also be: problems with anyone trying to select-and-copy text, as the hidden wikilinks would get copied to any richtext clipboard, or get clicked-on depending on the browser.
Possibly a design mockup, or more feature details, would help us understand how he (or anyone) sees it differently. HTH. –Quiddity (talk) 19:02, 18 March 2014 (UTC)Reply
Linking all pages is clearly adding a lot of noise and something wikignomes wouldn't like, but some people like noise and as long as those links are clearly distinguishable from actual, manually curated links I suppose the editors won't feel insulted. However, calling that "wikification" is an oxymoron. Call it "link invasion", or "link everything", or "click all the things", or whatever, but don't use the word "wikification". Thanks. --Nemo 19:14, 18 March 2014 (UTC)Reply
@Quiddity and Nemo bis: Hi guys. The concept of "wikify everything" should be a subheading under a topic of "content is not its presentation", and reenvisioned as part of a campaign to properly abstract content from presentation in Mediawiki, or at least Wikipedia's implementation of it.
See also: Model view controller
This is relevant to both the proliferation of wikilinks and of citations and their page numbers. See also the clunky and inherently obsolete Sfn vs. the clunky but clean and forward-compatible Rp, both of which James_F dislikes due to their use of secondary templates but we have to do something.
From a cosmetic or presentation viewpoint, there is generally no reason why any editor should care how many wikilinks or citations (including page numbers) there are, or how often they could appear. As seen in w:WP:Wikipedia doesn't use Allwiki, Wikilinks and citations should be as prolific as possible, with a user- and article-specific and wiki-specific default setting of how they are presented. This rich metadata undercurrent should be the story beneath the story.
As far as I'm concerned, there are two definitions of "overlinking" (same as overcitations), where one is real and one is imaginary. Wikipedia's predominant policy is based on the imaginary one, seemingly to the level of aesthetic superstition.
  • The real one is where an editor is in the wikicode, making wikilinks, splitting up the atomic ideas of the article's copy text, or linking to articles whose target context is not completely solidly aligned with its source. That's just overly aggressive or careless metadata cultivation.
  • The imaginary one is the idea that overlinking is totally defined by the prolific occurance of wikilinks inside the wikicode, period. Because content and presentation are actually fused as one.
I wasn't around for the discussions mentioned here, and I'm not a software engineer. But as I know it, this particular concept of software or content being able to fall under its own weight is one of antiquity. Even the idea that wikicode could become unreadable is not real, when given a properly abstracted, modal, editing interface.
The fusion of content and presentation is an idea that seems to have been inherited from the legacy days of the inception of the first wiki software decades ago. Heck, back then, we had the abomination of CamelCase, in order to deliberately fuse content and presentation, for detrimentally infectious levels of editorial convenience, where wikilinks automatically grew like weeds. Since then, we have slain the Camel but Wikipedia's policy admonishes us to manually hack each article's down to an inkling of wikilinking!
In reality, what the reader sees of wikilinking is ultimately supposed to be defined by the presentation layer, which Mediawiki (as I have ever seen it, so please inform me) does nothing to aid in cultivation and configuration by the editor or the reader. It should have a slider bar on the article, or at least some radio buttons in preferences.
By default, that wikilink and citation display slider could be slid down. Maybe you don't want to see the page numbers. Maybe you don't want to see the citation for every major statement or sentence. Maybe you don't want to see citations and other metadata at all. Maybe a large article would burden the render servers too hard; so for performance reasons, if Mediawiki benchmarks a given page poorly, it could adjust its default slider, or notify someone. James_F in IRC said that there is a new design idea for Cite.php to incorporate page numbers into the reference like this: <ref name="reuse">...</ref> foo bar<ref name="reuse" page=2/> bav baz<ref name="reuse" page=40-42/>
Quiddity (an impressively literate person who I just met today) and I (a 1.5 year Wikipedia editor) have been discussing this in #wikimedia-tech. He cites these examples given by Encyclopedia of Life: specifically 3:19 to 4:01 for the preferences panel and 2:48 to 3:20 for a glimpse of the slider. Quiddity said, "Basically, the complexity-slider would emphasize/re-arrange different aspects of the article (eg. making the Common Name of the animal bigger, or the Taxonomic Name), and it would trim or add complete sections. I thought it was brilliant, at the time. But I can understand how it wasn't a successful enough feature to warrant the prime-placement that it had." Presentation. Those whole videos are great, and they also illustrate the value of having a social hierarchy of vetted expert editors which is a whole other far more necessary topic.
So to summarize the subject of MVC-like abstraction of content and presentation as I see it, we need to upgrade Cite.php to infuse page numbers, and various other abstractions and GUI presentations. Leave any notion of noise, likes and dislikes, and people's feelings, up to some rational defaults and a slider or radio. Thanks. Smuckola (talk) 06:04, 2 July 2014 (UTC)Reply
Furthermore, this is just one example of how you can't apply a social solution to a technological problem. Wikipedia has some technological limitations which lead to politicking and superstition, and continuous social strife and chaos. Some of these issues just shouldn't exist! There should be a new era of wikignoming where people add in as much accurate metadata as possible and let the renderer sort it out.Smuckola (talk) 21:23, 4 July 2014 (UTC)Reply
@Smuckola: Also mentioned (the next day) in IRC, was m:Grants:IdeaLab/Reform of citation structure for all Wikimedia projects (which leads to most of the discussions and ideas in the area of centralized citations), and mw:Citoid which is now testable at Enwiki via this user script.
Re: Linking everything, one way to take the idea further, would be to create 2 examples of how it might look/work (perhaps annotated screenshots, or sandbox wikipages), in the hopes of finding someone capable of writing the javascript necessary to turn this idea into a userscript/gadget. Eg. make a copy of a short article, and add wikilinks to every single currently-unlinked article that you envision would be added (perhaps with a template, that makes the link text black until rolled-over).
I fear it's going to be remarkably resource intensive though, at the user-end (by adding hundreds or even thousands of links into long pages), and/or at the server-end (by making the renderer (or API or whatnot) search for "potential articles to link to" for every word or string. I.e. searching for 4 million+ articles in every page. And then also have to decide what to use when there's more than one match, eg. "King George" vs "King" and "George").
(Personally, I'm not really interested, as I use this firefox extension to make searching Wikipedia very easy, so I can already get extra lost/distracted without additional assistance ;)
HTH. –Quiddity (talk) 20:51, 3 August 2014 (UTC)Reply

Historical?

edit

@Quiddity (WMF): - It looks like this page hasn't been used at all for the last couple years (despite at least a dozen Beta Features being released in that time period). Should this page be marked as historical? Kaldari (talk) 23:09, 18 October 2018 (UTC)Reply

@Kaldari: - Yes, iff there's a replacement process that can be described in the docs. This page is currently linked from these 3 locations, which should be updated with instructions or pointers to the new/current ideal process for proposing a new Beta Feature.
HTH. Quiddity (WMF) (talk) 23:18, 18 October 2018 (UTC)Reply
Return to "Beta Features/New Features" page.