User:Xiong/tagging

This is a semi-open page for presentation of the arguments against template tagging. Feel free to edit or ask questions, but I reserve the right to preserve intent and to move lengthy and redundant argument to User talk:Xiong/tagging. The purpose of this page is not to fight out the point, but to concentrate in one place all the information needed to educate members interested in the topic. Please assume I have made any unsigned comment. &mdash; Xiong&#29066; talk* 10:23, 2005 May 14 (UTC)


 * Tagging is the practice of applying a small piece of markup, usually by way of template transclusion, into the markup body of a page, for an administrative purpose. Tags are generally inserted at the very top of a page; it is possible for a page to accumulate two or more tags, in which case the order of precedence may become contentious.


 * Tags come in a wide range of styles, forms, and intents; generally, they span the full page width and appear within a box, often colored red or amber, frequently bearing an icon or other image. Tags are distinguished from other elements added to a page in that the tag does not contain content relevant to the topic or function of the page itself; rather, it contains metainformation -- information about the page.


 * For instance, the template  generates the following tag:


 * Often, tags are used to deprecate or elevate a page's status within the Project. A flag is another name for a tag and flagging is a synonym for tagging.

Position
It is my position that tagging templates, under most circumstances, constitutes vandalism. (See opinion at Transclusion costs and benefits). To summarize in a word my arguments, template tagging destroys instantly a portion of the value of any template; in some cases, it renders the template useless. As this action takes place before process completes or even begins, it is a violation of our core principle of concensus, as prominently stated in Policies and guidelines. That is, the petty policy of whether a given template be permitted to exist must established by process of concensus and not by unilateral act. Obviously, the principle of consensus dominates any minor policy evident at any given time on Templates for deletion; to the extent that such minor policy conflicts with core principles, it is invalid.

Therefore templates ought not be tagged upon their bodies, and instead upon their corresponding talk pages -- prevailing practice or spurious claims of expedience notwithstanding. &mdash; Xiong&#29066; talk* 23:17, 2005 May 11 (UTC)

Of course the actual effect of tagging anything with {tfd} or similar negative tag is to accelerate its demise, and it does so with entirely too much efficiency and too little respect for due process.


 * Tags deprecate templates out of hand, at once, with no discussion whatever. The psychosocial effect of a tag on a given template varies, but can never be positive.


 * Tags damage templates to a greater or lesser degree. Templates, unlike article text, are code -- especially technical templates. In some cases, this damage renders the template immediately unusable -- surely a disincentive to use.


 * Tags on templates confuse the reader who may have no idea that templates even exist; even sophisticated users may be unable to determine what has been tagged.

In short the action is prejudicial and just plain wrong.

Internal Links

 * Templates for deletion is both process and workflow
 * Transclusion backgrounder: Transclusion costs and benefits
 * Another view of transclusion: Avoid using meta-templates
 * A earlier attempt to make this argument: Template talk:Tfd -- notable as the proponent spoke to the issue long before I joined the project

General argument
Shall {tfd} tags be placed on the template page itself, or on the template's Talk page?

It's common practice to place similar tags on article pages themselves. If, for example, the page on Water skiing is up for VfD (because, say, waterskiing is "non-notable"), then this is something we all want to see and know about, right away. The &#123;&#123;vfd&#125;&#125; tag appears in only one place: at the top of Water skiing.

It has been demonstrated clearly that this is a mistake when applied to templates. Templates are used on many pages, and if &#123;&#123;tfd&#125;&#125; is attached to the template itself, it is replicated on every instance of its use, disrupting pages which have nothing to do with the nomination for deletion. This use of {tfd}, when intentional, is actually hostile: it begs the question, anticipates the outcome of debate, by trashing every appearance of the template, possibly rendering it useless. Thus, it's an attempt to bypass the TfD process itself and usurp consensus.

There are times, I agree, that {tfd} may appear on a template page. When the template to be deleted is a series box, {tfd} should be inserted not only on the template page, but within the series box. This makes it clear what is being considered for deletion; and brings the matter to the attention of those most likely to care to discuss it.

tfd

The actual text of the {tfd} tag just makes the matter worse. It has gone through several versions, but always represents the most recent editor's limited conception of what sort of template might be so tagged.

Some templates generate no text at all, only formatting markup; others have no visible effect whatever. It is even possible to create a template that inserts nothing but a standard HTML comment. Technically advanced templates may act over an utterly unanticpated range; {tfd} will always confuse and distract.

We need not tag templates, debate them, and remove them with a blare of trumpets; nor must we do so under shade of night. It's perfectly possible that we tag templates on their Talk pages, and make a polite notice to affected users.

This is not nearly so hard or burdensome as it appears. I see which kinds of templates are nominated frequently. They fall into four general groups:


 * Foolish or vanity templates, such as mine own 01:

 The text below is generated by a template which has been proposed for deletion. Please see its entry on Templates for deletion for comments and voting. It's all just zeros and ones!.


 * Only the creator is likely ever to make any defense of such, and can probably be reasoned with directly, leading to immediate speedy of the template in question and sparing us the drudgery of nomination, tagging, debate, request for admin attention, and logging.


 * "Polish steam locomotive engineer family tree" templates: monstrosities that loom larger on the hundreds of pages on which they appear than any actual page content. These must be dragged through TfD and might discover some support from Polish steam locomotive enthusiasts. These single-minded users haunt Polish steam locomotive pages to the exclusion of all else, and if we notice the deletion on one Talk page within the set -- the template page itself being the natural place for this -- it will show up on every Polish engineer's watchlist and -- with the proper edit summary -- appear as a great red flag. Nobody will be left out; if only one or two Polish engineers take note, they will sound the alarm and every Polish engineer in the project will caucus at length before descending on TfD in mass.


 * "Rain Man" templates, created by obsessive-compulsive semi-autistics (like me) for the purpose of categorizing different shapes of pinto beans, or marking articles as rainforest-endangered-wildlife-film-star-stubs. You may rest assured that the slightest dummy edit to one of their babies will startle them from their dazed counting and re-counting of places within the project where split infinitives fester. I cannot promise they will deign to speak to us, but they will read the notice and have the opportunity to do so.


 * "Frankenstein's Kitbash" templates, which I illustrate (vainly, for which I apologize) again from my own stable: divbox and doctl. If they work, these highly technical templates may be of great interest to many users; if not, nobody will weep over their deletion. The trouble is that placing a tag -- any additional code at all -- within the body of such templates may cause them to break -- in unexpected ways, perhaps. Even if they do not fail outright, their usability is so immediately degraded as to suggest that they were broken before nomination: fait accompli. Or the tag inappropriately points to something that has little to do with the template itself, sowing confusion.

Technical example
Since it was a certain user's adamant insistence on tagging {divbox} that moved me to this debate, an example of this template is appropriate. For all examples, the same source code insertion of {divbox} is assumed:

This is the way the template is meant to be used by a user to do something on a substantive page -- possibly a template, possibly a sidebar, with or without the  atom. The only thing that varies among these examples is how {divbox} is tagged for deletion.

(The actual source of these examples was created by substitution.)

Here is a use of the template as it is intended; obviously, it will render the same way whether it is tagged on its Talk page or not:

 Lorem ipsum Dolor sit amet, consetetur sadipscing elitr, sed diam nonumy eirmod...

But adding {tfd} to Template:Divbox forces every instance to appear thus:

 The text below is generated by a template which has been proposed for deletion. Please see its entry on Templates for deletion for comments and voting.

 Lorem ipsum Dolor sit amet, consetetur sadipscing elitr, sed diam nonumy eirmod...

Perhaps {divbox} stinks and should be carted off with the rest of the rubbish, but the text within the {tfd} tag appears to point to "Lorem ipsum...". FWIW, I copied and pasted that text; it came from no template. Lorem ipsum is not up for deletion. And while you may say "that's obvious", it is only obvious if you have already been over the battleground. A naive user who sees this notice will naturally think {tfd} applies to the contents of the box, not the box itself, which is all the nominated template generates. Further, it damages the appearance, which -- since a colored box is all about appearance anyway -- is again tantamount to strangling the baby in the cradle and fait accompli.

Still, this is not the worst unintended -- or maliciously intended -- consequence possible when fooling with technical templates. {divbox} does some tricky things to make life easier for humans. In particular, I want an easy way for a user to be able to choose colors for a box; that's harder than it looks, because both box border and box background must be set individually, and one cannot be specified as a tint of the other. Besides, that might not be wise, even if technically feasible -- I think one of the most successful styles is "amber", which is a light yellow background and a brown border. I actually expect some user to demand that all boxes, of whatever background color, be bordered in black. Nor will I interfere with the change. I built that robustness into the model.

On another level, I want to be sure that users at a slightly higher level of technical competency can create new styles and extend the set, not be limited by the first dozen things that popped into my head. So, all the style information is contained in one or another subtemplate. But I don't want to create a template named "blue" or "navy"; that's too general. The subtemplates have names like divstylenavy and the calling template, divbox, supplies the first, pseudo-namespacing part of the subtemplate name, allowing users to merely type the style code word "navy".

Now, we should all be glad that a certain user was too lazy to go and tag the template bodies of all 13 subtemplates. If he had, then every call to {divbox} -- no matter where it appeared -- would lead to this:

 The text below is generated by a template which has been proposed for deletion. Please see its entry on Templates for deletion for comments and voting.

 The text below is generated by a template which has been proposed for deletion. Please see its entry on Templates for deletion for comments and voting.

background-color: #AADBE0; border: 1px solid #00477B; margin: 0.5em; padding: 0.5em"> Lorem ipsum Dolor sit amet, consetetur sadipscing elitr, sed diam nonumy eirmod...

Clearly, this is a sort of government-sanctioned vandalism, utterly destroying any usefulness the template might have had. There is no need for a vote; let's just tie the creator to a stake and burn him in public square. As I said, I'm glad he didn't do it.

This last case is difficult to explain and I thank sincerely anyone who has read this far. I hope you will all agree than anyone bold enough to create Frankenstein Kitbash-type technical templates is able to take care of himself; you do not need to tag his templates in order to ensure he comes to the table for discussion of the deletion. You might drop him a line on his Talk page; he'll drag in all the friends he needs or wants.

Meanwhile, though, since the template has not been deleted; since templates, as humans and dogs, are innocent until proven guilty; it is criminal to destroy them to prove a point. To force this premature destruction upon anyone wishing to nominate a questionable template is clerk-mind, the hum of the worker bees. To do so in order to deprecate a comment on a debate page -- which is what led us here -- well, it is beyond my understanding why anyone else would tolerate this, much less endorse it. &mdash; &mdash; Xiong<font color="#997749">&#29066; talk<font color="#009900">* 11:23, 2005 Apr 6 (UTC)

(Edited &mdash; Xiong<font color="#997749">&#29066; talk<font color="#009900">* 06:50, 2005 Apr 29 (UTC) &mdash; Xiong<font color="#997749">&#29066; talk<font color="#009900">*  10:23, 2005 May 14 (UTC) )

Precedent
Users other than myself have removed destructive {tfd} tags from nominated template bodies; but these might all be criticized as individual acts. However, a clear precedent was set in the case of the nomination itself of Template:Tfd for deletion -- quite properly, within TfD process and according to its guidelines.

I made the nomination and, according to "accepted practice", placed the tag, &#123;&#123;tfd&#125;&#125; within the body of the nominated template. Whether due to machine or operator error, this action was doubled.

User:DuKot -- just two minutes later -- removed the offending tag. I restored it with edit summary ''tfd tag restored per stated procedure on Wikipedia:Templates for Deletion. if you disagree with nomination, please vote in the appropriate place.'' This appeal to "accepted practice" failed.

User:Violetriga, User:Korath  , User:Rhobite , User:Hadal , and User:SPUI  all removed the tag thus applied.

While the nomination was itself controversial and I was attacked on grounds of WP:POINT, this series of actions by a number of members did clearly set a precedent.

Consensus is that there must be an exception to this so-called "accepted practice". And if there is one, there must be others. Nothing about Template:Tfd elevates it to the status of a foundation issue; it is in the same class as hundreds of other common templates.

Server Costs
Fears of excessive server costs are generally overblown. The total workload of the WikiMedia server farm is the sum of a very large number of very tiny loads. It is almost impossible for the action of a single user to significantly add to or subtract from the overall burden. It's certain that the service of thousands of pages per hour adds up to a real cost of tens of thousands of dollars per year, but the individual cost of any action is infinitesimal.

As noted at Transclusion costs and benefits, costs do increase rapidly when transcluded elements are the targets of transclusion. Concerned parties note that tagging a nominated template is an instance of double (perhaps triple) transclusion; the more popular the nominated template, the greater the burden. I do not insist that this constitutes a crisis.

In the specific case of the revert war over tagging Template:Tfd, and in other cases of tag reversion, I have been criticized for imposing an unnecessary server load. I did, indeed, restore {tfd} to the nominated template on 6 occasions over the course of 7 days -- and, contrary to "accepted procedure", other users (as noted) removed it; another user restored the tag on one occasion. As {tfd} -- for good or ill -- is a very heavily used template -- worse, the subject of double transclusion -- it can be imagined that this revert war unduly burdened the server.

In contrast, there have been 39 edits to Template:Tfd in the course of the last 5 weeks (as of this writing). Little discussion has taken place anywhere regarding these changes (outside of edit summary!), and the net achievement is that the template's appearance and effect is not very different than it was at the conclusion of its own nomination process, or even from its initial version.

What is more, the template was edited 37 times before I ever touched it; not because it is a common target of anon vandalism of the conventional sort -- but merely as a result of users fiddling with it, with little attempt to test their "improvements" or gather consensus for their edits.

<div class="boilerplate metadata" id="Template:Tfd-1" style="background:#000; border:1px solid #fff; margin:0.5em; padding:0.5em; color:#fff; font-weight:bold;"> (Original version) Template:Tfd (17:48, 2004 Aug 12) This template has been listed at Templates for deletion. Please see its entry on that page for justifications and discussion. If you don't want the template deleted, please read the deletion guidelines and vote against its deletion there; please do not remove this notice or blank this page while the question is being considered.

<div class="boilerplate metadata" id="Template:Tfd-2" style="background:#000; border:1px solid #fff; margin:0.5em; padding:0.5em; color:#fff; font-weight:bold;"> Template:Tfd (08:40, 2005 Apr 10)

<div class="boilerplate metadata" id="vfd" style="background-color: transparent; margin: 0 2.5%; padding: 0; border: 1px solid #aaa;font-size:xx-small;"> The text or formatting below is generated by a template which has been proposed for deletion. Please see its entry on Templates for deletion for comments and voting.

<div class="boilerplate metadata" id="Template:Tfd-3" style="background:#000; border:1px solid #fff; margin:0.5em; padding:0.5em; color:#fff; font-weight:bold;"> Template:Tfd (02:43, 2005 May 14)

<div class="boilerplate metadata tfd" style="background-color: transparent; margin: 0 2.5%; padding: 0; border: 1px solid #aaa;font-size:xx-small;"> The template below has been proposed for deletion. Please see Templates for deletion for comments and voting.

<div class="boilerplate metadata" id="Template:Tfd-3" style="background:#000; border:1px solid #fff; margin:0.5em; padding:0.5em; color:#fff; font-weight:bold;"> (Current version) Template:Tfd (00:00, 22 Jun 2005 (UTC))

<div class="boilerplate metadata plainlinks" id="tfd" style="background-color: transparent; padding: 0; font-size:xx-small; color:#000000; text-align: center; border-bottom:1px solid #AAAAAA;">&lsaquo;The template below has been proposed for deletion. See [/wiki/:Templates_for_deletion#Template: templates for deletion] to comment and vote.&rsaquo;

What is sauce for the goose is sauce for the gander.

Bad faith?

 * Vandalism always implies bad faith. If the tagging is not meant to disturb Wikipedia, it's by definition not vandalism, no matter how disruptive it might be. (Wipe 23:32, 12 May 2005 (UTC) on Wikipedia talk:Transclusion costs and benefits)

"Vandalism" is, among other things, needless damage of common property, per. While I'm aware the word has a private meaning among Wikipedians, I use it in the publicly-accepted sense. It does not actually specify hostile intent. (I can, on request, substantiate "needless", "damage", and "common property".) I shall, however, demonstrate bad faith as well.

I don't doubt that some nominators' intentions are pure and that they merely follow outmoded, inherited practice blindly; but some tagging is clearly done with intent to take the template out of circulation as quickly as possible, by nominators who would really prefer to speedy delete. Other nominationators, in good faith, tag template talk pages and allow consensus to form and process to operate.

An excellent test of good faith in this matter is the recent nomination (with typically piano edit summary) of Template:Tfdnotice. This innocent template inserts entirely neutral text inviting the reader to visit TfD and is directed to a specific nomination; the purpose is to notify directly interested parties (see docs), such as template editors and those who may use the template regularly. I suggest that those who vote to delete a tool for efficiently providing such impartial notice do not want more voices at the table -- especially voices that actually may oppose such deletions.

Must attract attention

 * Using is important to attract attention to the vote. (Wipe, ibid.)

It is not clear to me that tagging brings voices to the table as intended. You may say, "but it's obvious", but many things are "obvious" that are just not so. The most vocal voters on TfD are the regulars -- deletionists and inclusionists both; this is a truism. So far, statistical analysis is inconclusive on this point. There are technical reasons why tag-on-body fails to alert interested parties:


 * It is hoped that some interested party, seeing {tfd} appear somewhere on some target page, will determine which template has been nominated and, thus noticed, proceed to Templates for deletion. But in many cases, this is very likely to fail, as the doubly-transcluded copy of {tfd} does not actually point to any thing except "below" -- and some templates may generate no visible text; others do so in such a way as to displace the copy of {tfd} from its object.


 * It is hoped that placing {tfd} on a template body will cause some page to appear on a number of members' watchlists, again perhaps prompting their visitation of TfD process. However, members who watch a target page A upon which Template:T has been transcluded will not see A on watchlist when Template:T is tagged with {tfd} -- only those who directly watch Template:T. (Those who do watch Template:T also, by design, watch Template talk:T; nothing is lost by tagging there.)

Further absurd effects of template tagging are evident when a template is substituted rather than transcluded:


 * A template is substituted, then tagged. -- The {tfd} "payload" is never delivered to the editor who so used; the substituted content is unaltered.


 * A template is tagged, then substituted. -- The reader seeing {tfd} on a target page is unable to connect it to any template by examining the wikimarkup. The tag "rattles around loose". Whether the template is eventually deleted or not, the tag remains -- pointing to nothing with vague menace.

In any case, this is a straw man argument: "If we don't tag templates, then interested parties will not come to the vote; therefore we must tag, no matter how biased and destructive the act." I don't insist we omit tagging altogether. We can tag on talk; we can notify possibly interested parties directly; in some cases, I might even tolerate tagging on body, if a technical analysis proved the innocence of the act. We can approach the situation from different angles, solve the problem, and still avert tag vandalism.


 * I, for one, would point out that the only reason I started reading TfD was because a template I found useful was so tagged. Anecdotal perhaps, but important to me. In fact, one of the great problems of Categories for Deletion is that it is not evident that a category is being voted on because categorizers do not see that a category they are interested in is up for deletion unless they have the category itself watched, which is rare.  This has resulted at least twice in recent memory in well-used categories being deleted with none of the categorizers knowing about the deletion vote.  At least  spares us from that situation.


 * There are (at least) two interested parties when it comes to template deletion: those who have edited the template, and those who include the template in their articles. Given the stability of most templates as compared to the articles they appear on, I think the attracting of attention is a  reasonable and democratic argument to make for tagging, whether or not particular statistical evidence suggests that those seeing a tag actually end up voting.


 * That said, tagging can blight a large swath of articles, as the recent tagging of Template:Wrongtitle did. In this case, I'd suggest a move to speedy consideration is the right answer, not eliminating the tag (which could well result in the template being deleted without the editors of that large swath of articles having any visibility into the decision). TreyHarris 18:42, 27 May 2005 (UTC)


 * It is a serious concern that interested parties come to the table to discuss templates nominated for deletion. It's just that tagging is the wrong way to do it; in some cases, a disasterous way to do it. Some templates may be so tagged; the problem is figuring out which ones. There are, generally, two alternate routes:


 * Make the effort to discover and notify interested parties in other ways. This requires some thought, which is never a bad thing.


 * Don't try to delete the template. Deletionist bias is not a good excuse for rampant nomination of templates. Templates used by large numbers of editors are on their faces ineligible for deletion. Templates which appear to be unused are ineligible; so are templates that just don't happen to make sense to the nominator. No template should be deleted unless its mere existence is a problem, or it is clearly of no worth.


 * This "attention" requirement is based upon using simple Wikipedia concepts: Viewing an article is the basic action, and having watchlists encourages manual supervision of content which results in editors again viewing an article. There are other ways in which notification could be done.  One way, looking at History and notifications of some editors, is not widely used.  Other possibilities include automated notifications of template editors (public info), of editors of articles who invoked templates (public info but hard to find in History), of editors of articles in which a template is used (public info but list can get large), of editors of articles with similar Categories (public info), of people with articles in Watchlists (private info).  "Notification" might mean Talk page messages, or might involve Watchlist page behavior.  (SEWilco 01:09, 21 August 2005 (UTC))

Cosmetic Fixes
Might it be possible to define the template in such a way that the notice appears somewhere out of the way? Perhaps surrounding it with a suitable DIV tag which the CSS moves to the top of the article. Stoive 02:09, 18 May 2005 (UTC)


 * One issue is that it's not clear what the tag really points to; for that purpose, it should be more obtrusive, not less. Another issue is that the tag politically deprecates the nominee by unilateral act; for that purpose, it should be as neutral as possible.


 * One goal of tagging is to notice interested parties; but (a) it doesn't work and (b) there are other ways to do it. Cosmetic fixes won't really change this.


 * The trump card, though, is that some templates are so constructed that any additional generic code thrown in for any purpose will instantly cause them to break. No amount of cosmetic improvement can fix this -- none at all. &mdash; Xiong<font color="#997749">&#29066; talk<font color="#009900">* 11:13, 2005 May 18 (UTC)


 * Ok, so stuff like breaks whatever you do in the text of the template. Putting it on the Talk page makes it practically invisible to all intents and purposes. The whole thing suffers from the problem that the message doesn't identify which template is being referred to.
 * I would maintain that some kind of tag that propagates through to the article is needed. Otherwise the the formatting or content just breaks or disappears one day, with there having been no prior link to the process that caused it. <font color=#f00000>If this text is red, there seems to be an alternative method Stoive 03:16, 19 May 2005 (UTC)


 * I grant that this is an interesting technical approach; I'm not sure it will not break some templates anyway. But I applaud your effort to find a novel solution; it's the first really new idea in a long time. &mdash; Xiong<font color="#997749">&#29066; talk<font color="#009900">* 02:04, 2005 Jun 2 (UTC)

Conditional tagging
Perhaps template tags could behave differently for possible article editors. As one example, I consider someone who has an article in their watchlist to probably be an editor. If there is a variable which indicates whether the viewer has the article in their watchlist, a template can produce a different display. On one hand, TfD tags might be suppressed if article is not in the watchlist. Or TfD tags may be enhanced (red flashing stars?) for likely editors. I am aware of various issues, but wanted to point out that templates which perform tagging can behave differently based upon what information is available. (SEWilco 01:00, 21 August 2005 (UTC))

Define tag
I think I know what you are referring to, but "tagging" is currently defined by implication and example. An additional sentence is called for in the introduction. (SEWilco 01:00, 21 August 2005 (UTC))


 * Ask, and it shall be done. &mdash; Xiong<font color="#997749">&#29066; talk<font color="#009900">* 09:55, 2005 August 22 (UTC)