User talk:The Duke of Waltham/Auto-formatting is evil

Endorsement

I'm not sure of the propriety of this, but I wanted to record my endorsement of this essay. Please feel free to refactor or delete or whatever. --Elliskev 00:23, 15 August 2008 (UTC)[reply]

Endorse I came into the discussion with full support of auto-formatting (check my edit history). After reading the arguments against, I challenged myself to turn off date preferences. I quickly (in about two weeks?) realized that auto-formatting is like the rug under which dust is swept. It hides the inconsistencies that are obvious to most readers (unregistered), without providing any benefit to those readers. --Elliskev 00:23, 15 August 2008 (UTC)[reply]
Endorse In addition to Engvar, there is the analogous decision to allow both AD/BC and CE/BCE. Septentrionalis PMAnderson 00:55, 15 August 2008 (UTC)[reply]
- I do not agree that we should now ban autoformatting, or even linking dates. Autoformatting was a bad idea, when we had not yet adopted the practice of learning to live with diversity; but some dates ought to be linked: 5 November 1605 and July 4, 1776 come to mind - and the practice on peerage articles of linking the years of creation is harmless and useful. Septentrionalis PMAnderson 02:12, 15 August 2008 (UTC)[reply]
  - I am not saying that we should outright ban auto-formatting, but I do support its (gradual) deprecation. There should be room for exceptions, of course, as with most things; this is something to be discussed, probably in the future. Waltham, The Duke of 07:07, 15 August 2008 (UTC)[reply]
Bleh -- SatyrTN (talk / contribs) 02:34, 15 August 2008 (UTC)[reply]
I, too, wish to congratulate His Grace on a fine essay, which I'm happy to endorse. Anderson, yes, I agree that earlier year pages, because they're shorter, are less likely to be a total waste of a click from a focused article. But where to draw the line (by rule of thumb, I mean, not strict boundary). Has anyone done a survey to see where year pages become rather shorter? Tony (talk) 09:08, 15 August 2008 (UTC)[reply]
Decline to endorse - Although well-written, this essay does not take into account other obvious (and previously expressed) ways to address the issue. In particular, the first point made -- regarding autoformatting being "useless to most readers" -- does not take into account the possibility of changing the behavior of the mediawiki software used for enwiki. The proposed change, which would allow all readers to see autoformatted dates, would completely address this complaint regarding autoformatting. This substantially weakens the case for describing date linking as "evil". (sdsds - talk) 17:55, 15 August 2008 (UTC)[reply]
- The purpose of this essay is to demonstrate that auto-formatting is of no substantial benefit in any form, and a problem in this form especially. Point Two makes it clear that the problem auto-formatting is set to solve is so minor as to be a non-issue; readers can understand dates in both major formats used in text, and it is bad for the mentality of the community to start applying artificial methods of changing the way articles are displayed in different areas—this is a slippery slope, and there have been attempts to compromise the unity of Wikipedia. Not to mention the inevitable side-effects in the accuracy of articles that any but the most sophisticated system will cause. This is an encyclopaedia, and such sacrifices for the sake of a minor convenience are unacceptable. Point Three further elaborates on how extremely advanced any global auto-formatting system would have to be before it could be even considered; I doubt we are anywhere near this level of sophistication. And all this has been said without taking the obvious reluctance of developers into consideration, or the extent of the resources a system of this type and scale might require to function. And even then, I still find it better to combat the problem at its root (ensure consistency within articles) rather than disguise our own incompetence or unwillingness to do so. We are a project relying on volunteers, and we shall always need them. If a reader stumbles upon an inconsistent date, we should encourage them to fix it instead of pretending to be a finished encyclopaedia. Waltham, The Duke of 21:04, 15 August 2008 (UTC)[reply]
  - And point 6 is fundamentally not soluble without a program capable of construing English, to determine whether a comma results from the date format or is required by the construction of the rest of the sentence. That's a long way off. Septentrionalis PMAnderson 22:19, 15 August 2008 (UTC)[reply]

PMAnderson: I think His Grace rather likes a practice I deprecate, the formulaic insertion of a comma after the year in the so-called US date format. Your conception—my preferred usage—involves grammatical analysis. His Grace's doesn't. Tony (talk) 03:25, 16 August 2008 (UTC)[reply]

The problem exists both ways: if the date in edit space is in American format, the autoformatter must decide, equally, whether to retain the comma in international format.
I suppose that if we ordained a rule that editors should use the comma in American style only when meaningful, and accepted that autoformatting American readers would always see the comma, we wouldn't need AI. But this makes Point #3 worse. In such non-autoformattable cases, autoformatting and non-autoformatting readers would have different expectations for the comma, resulting in conflicts; and it all seems like going around Robin Hood's barn. Septentrionalis PMAnderson 16:52, 16 August 2008 (UTC)[reply]

Unless, that is, we abolish auto-formatting altogether. :-) Waltham, The Duke of 09:43, 17 August 2008 (UTC)[reply]

Sdsds: Has my post on your talk page a few hours influenced your take on this issue? Tony (talk) 03:25, 16 August 2008 (UTC)[reply]

PMAnderson: Linking July 4, 1776—so you have to click on either "July 4" or "1776". The "July 4" page tells you, right down the list, "1776 - American Revolution: the United States Declaration of Independence is adopted by the Second Continental Congress". Where would that be useful? Wouldn't we already know it from an article from which you'd link to that page?
- Not necessarily. Like any other link, it will get the reader more information; xe'll have to follow the links on that line, to the Revolution, the Continental Congress, or the Declaration of Independence. Unfortunately, in some cases, that will be the only link to the information. Many articles on American history have used "has existed since July 4, 1776," or some such phrase, as a reference which will, in itself, be clear to the reader. Ideally, most of these should be rewritten; but simple delinking will leave the non-American readers who don't get the reference lost. Septentrionalis PMAnderson 16:52, 16 August 2008 (UTC)[reply]
  - Well, I suppose a year link could be retained in the absence of a better one. But the better one will be added, sooner or later, and then the year link must go. Well-written articles must have the information anyway; links, navboxes, and categories, if handled well, should take care of the rest. Waltham, The Duke of 09:43, 17 August 2008 (UTC)[reply]
    - Perhaps, without autoformatting, July 4 would have a section on the Declaration of Independence, and 4 July would have a section on colonial insurrection? ... I wrote this fairly tongue-in-cheek but thinking on it, it seems likely twin histories would develop, one US-centric, the other Euro-centric Bazj (talk) 13:24, 26 August 2008 (UTC)[reply]

Related point: I've done a quick and dirty survey of year pages back through the ages. Hmmm ... big problems there. Either too much info (past few centuries) or not enough (beofre then); lack of citations an issue. Tony (talk) 04:15, 16 August 2008 (UTC)[reply]
- I would take year pages, like disambiguation pages, as falling under an exemption for navigation aids; citation on the page linked to is enough. Septentrionalis PMAnderson 17:03, 16 August 2008 (UTC)[reply]
  - Perhaps. But there are hundreds of them, and they don't receive nearly enough attention by editors. I think disambiguation pages are much better regulated, as a matter of fact. Waltham, The Duke of 09:43, 17 August 2008 (UTC)[reply]

While it's easy to make the argument that autoformatting has minimal benefits, I have yet to see a compelling case that it's genuinely harmful. None of the issues presented in the essay are genuinely significant. Inconsistency between two easily readable formats is unlikely to impact the reader experience in any meaningful way. etc. Further, delinking all these dates involves destroying a great deal of potentially useful metadata that editors have contributed through the years, and technical improvements to autoformatting could perhaps address many of these concerns without destroying those contributions, which would probably require a great deal of effort to rebuild. I have a hard time believing that widespread application of delinking is the best solution to the complaints about autoformatting as it stands. Christopher Parham (talk) 06:27, 16 August 2008 (UTC)[reply]
- I think it's easier to see it like this: its benefits are not enough to outweigh its shortcomings. And even if you don't find each argument compelling on its own, putting them all together might be more persuasive. In any case, auto-formatting is rather ineffective even for the editors that use it (Point Three), and it represents a mentality damaging to the project, namely that we should distinguish between American and European editors on a global scale (Point Two). And even the style arguments have more weight if you consider that even FAs can fall prey to the confusion caused by auto-formatting (they aren't protected, see); the feature speeds up rot. Finally, I am opposed to auto-formatting in any form; I've already mentioned Point Two, and I'd also suggest taking a look at my reply to sdsds higher on this page. Waltham, The Duke of 09:43, 17 August 2008 (UTC)[reply]

Comment. Point #8 is not entirely true: autoformatting exposes errors, which makes them more likely to be fixed. Here's an example:
- Decmeber 31, 2005 (no autoformatting)
- Decmeber 31, 2005 (with autoformatting - redlinked!) GregorB (talk) 15:29, 16 August 2008 (UTC)[reply]
  - Until somebody puts in a redirect for common misspellings.... Septentrionalis PMAnderson 16:52, 16 August 2008 (UTC)[reply]

Endorse: Well said. I also changed my mind about autoformatting. Teemu Leisti (talk) 01:05, 21 August 2008 (UTC)[reply]

Comment. If the functionality is available it will be used, regardless of any ban or deprecation. I will agree the current situation is not a desirable one.
Taking your example of the date-range problem as a challenge I tried writing a template which would express a range as concisely as possible (d1-d2 m1 y1, d1 m1-d2 m2 y1, or d1 m1 y1-d2 m2 y2) and in the user's chosen preference (m1 d1-d2 y1,...). The bulk of it was straightforward and simple, except the user's preference is imposed AFTER the template's run, so there's no way to take it into account.
It now seems obvious to me that the problem lies in the wiki software.
Either the autoformat should be removed (which will make linked dates less attractive), or should be implemented more completely (assume a preference for users who haven't expressed one, and expose the user's preference for templates to work with). Bazj (talk) 13:09, 26 August 2008 (UTC)[reply]

Coverage

Was looking at this, following your (The Duke of Waltham's) comment (probably tongue-in-cheek) that no-one is promoting the essay. Tony has linked it, but I suppose you (or others) could link it a little bit more - especially given the endorses above. I know it can feel wrong promoting your own essay, but that is the only way to increase coverage sometimes. Carcharoth (talk) 06:55, 8 October 2008 (UTC)[reply]

Maybe you are right. Self-promotion is not all bad; I do have some support here. Thank you for the encouragement.

And I know just how to do it... (prepares plans to vandalise the Signpost) Waltham, The Duke of 17:56, 8 October 2008 (UTC)[reply]

Out-of-date

Oops. You've linked to a section of the MoS that now says something different. Might want to link to the old version of the MoS, and/or update the essay? Carcharoth (talk) 06:56, 8 October 2008 (UTC)[reply]