User talk:Jarry1250/RFC

Semantic markup
Re. the semantic markup point, surely it would be best to have the wikimarkup in place now, right? - Jarry1250 (t, c) 15:29, 31 May 2009 (UTC)


 * Yes, of course, even what is available is much better than no markup, when it comes to making use of the semantic structure of the page (as in accessibility, for example). I merely intended to suggest that the flaws in html weaken the argument for strict adherence to semantic markup vis-a-vis the aesthetic argument. In this respect, I suspect that the WP:LAYOUT guideline may admit to exceptions for the latter reason. --RexxS (talk) 16:06, 31 May 2009 (UTC)


 * Ah, yes, I follow. I actually quite the like the aesthetics of headings threes after headings two, but that's another matter completely! - Jarry1250 (t, c) 16:16, 31 May 2009 (UTC)

Circumstances
Jarry1250 has asked under what circumstances might h4 follow h2. That can occur not only for aesthetic reasons, but also for semantic reasons. Here's a made-up example discussing 3 possible outcomes caused by 2 possible causes, the first of which has 2 sub-categories (for simplicity, I've omitted the text which would follow each of the headings, and used indentation to represent levels):


 * Oxygen toxicity (level 2)
 * CNS toxicity (level 3)
 * Outcome1 (level 4)
 * Pulmonary toxicity (level 3)
 * Outcome2 (level 4)
 * Carbon dioxide narcosis (level 2)
 * Outcome3 (level 4)

You may feel that the "Outcome3" section should be level 3, but both semantically and aesthetically, it should have the same weight as the first 2 outcomes, in my humble opinion. This may not be common, but will probably exist, so I suggest that it needs to be considered. --RexxS (talk) 18:37, 31 May 2009 (UTC)
 * An interesting point. Is that particular example from a page I could have a look at? - Jarry1250 (t, c) 18:39, 31 May 2009 (UTC)
 * I'll learn to read one of these days. Does anything similar actually happen in practice, do you know? - Jarry1250 (t, c) 18:45, 31 May 2009 (UTC)
 * (e/c)I think that it should be level 3. Now, if the sections were actually named "1, 2, 3", then yes, it should be like that... but I don't think that they'd be named "1, 2, 3". Outcome3 is an outcome of "Carbon dioxide narcosis" and not "Oxygen toxicity", so it is a subtopic of the former, not the latter.
 * Alternatively, if more restructuring was wanted, it may be possible to combine Outcomes 1 and 2 with their level 3 headers. If a heading has only one subheading, they can (generally) be combined... subheadings make more sense when there are multiple ones under the same heading. –Drilnoth (T • C • L) 18:43, 31 May 2009 (UTC)


 * With respect, you miss the point that both level 2's are subtopics of a higher level. I tried to keep it as a simplified outline to concentrate on the disjoint between the subtopics. I would not encourage combining of levels if that's not the natural structure of a page, just to meet sensibilities about semantic levels. This becomes far more unnatural when, as you say, there are several subtopics at each level. Do you still feel the same if the structure is:


 *  Effects of breathing elevated levels of oxygen  (level 1)
 * (some text here)
 * Oxygen toxicity (level 2)
 * (some text here)
 * CNS toxicity (level 3)
 * (some text here)
 * Visual disturbances (level 4)
 * (some text here)
 * Seizures (level 4)
 * (some text here)
 * Pulmonary toxicity (level 3)
 * (some text here)
 * Breathing difficulty (level 4)
 * (some text here)
 * Lung damage (level 4)
 * (some text here)
 * Carbon dioxide narcosis (level 2)
 * (some text here)
 * Unconsciousness (level 4)
 * (some text here)


 * Sorry for taking so much space. I'm happy to accept that others may not agree with what I feel is the natural structure, and how it should be marked up. But as far as I can see, the only objection to giving "Unconsciousness" the same weight as "Visual disturbances" is the stricture that level4 should not follow level2. Are there any other reasons? --RexxS (talk) 19:33, 31 May 2009 (UTC)
 * I still feel that "Unconsciousness" should be level 3. It is related to "Carbon dioxide narcosis" more than it is to "lung damage" or "Visual disturbances"... sure they talk about the same sort of thing, but think about this structure (with what I consider to be incorrect header levels):


 *  Dungeons & Dragons creatures  (level 1)
 * (some text here)
 * Humanoids (level 2)
 * (some text here)
 * Reptilians (level 3)
 * (some text here)
 * Troglodytes (level 4)
 * (some text here)
 * Sahuagin (level 4)
 * (some text here)
 * Goblinoids (level 3)
 * (some text here)
 * Bugbears (level 4)
 * (some text here)
 * Minotaurs (level 4)
 * (some text here)
 * Dragons (level 2)
 * (some text here)
 * Dragon turtles (level 4)
 * (some text here)


 * In this case, "Humanoids" and "Dragons" have the same weight, appropriately. Humanoids is further divided into subgroups: Reptilians and Goblinoids (each level 3), which both have further subgroups on individual creatures. However, "Minotaurs" are not "Goblinoids", but they are technically in a subsection of "Goblinoids" so that they are given the same weight as the other monsters. Troglodytes and Minotaurs should have the same weight... but using identical sections would mean putting Minotaurs under "Goblinoids" incorrectly. So, Minotaurs should obviously be level 3, not 4. But this gives them a different weight. Then Dragon Turtles should also just be level 3, not level 4, since they are a direct subtopic of "Dragons". But only the Minotaurs are level 3... should the Dragon Turtles be level 4 because they should have the same weight as Sahuagin and Bugbears? Or should they be level 3 like Minotaurs? It seems to me that having the headings technically correct (Minotaurs and Dragon Turtles both at level 3) makes more sense, even though that gives them "more weight" than other creatures which should have the same weight. My feeling is that this sort of thing would come up much more than your example, which I also feel would be better with the technically correct heading levels. Headers are designed to separate subtopics from their parent topics... the size just serves to indicate subsections, not "importance" to the topic. Besides, the TOC interpretes level 4 headers immediately below level 2 headers to be level 3 headers when determining its layout... so the TOC would then be different from the article's actual structure.
 * But that's just my opinion. Sorry for choosing such an arcane topic for my example; it's what I'm most familar with that would serve as a good example :) –Drilnoth (T • C • L) 22:45, 31 May 2009 (UTC)

(o/d) It took me a couple of reads, but that's a good example. The problem again is that there is are bottom-level objects which have different numbers of antecedents - "Creatures.Humanoids.Goblinoids.Bugbears" is the same sort of thing as "Creatures.Humanoids.Minotaurs" and "Creatures.Dragons.Dragon_Turtles". I sometimes come across that when trying to parse a hierarchical classification into a relational database. Since the bottom level objects will have analogous properties, my instinct is always to assign them to the same level. As you can see, your solution to your example would show "Bugbears" in the TOC at a lower semantic level than "Minotaurs", which doesn't seem right to me. Whereas my suggestion would show it as as a sub-category of "Goblinoids", which is clearly wrong. Hehe, that's what you get when you work in html which has no way of delineating the end of a semantic level. My usual solution would effectively be to place "Minotaurs (level 4)" before "Reptilians (level 3)" - in other words, put all the level4's that have no level3 antecedent before all those that have do have - but I do understand that the order of subtopics in an article may be dictated by more important considerations. Plenty of food for thought here. --RexxS (talk) 13:03, 1 June 2009 (UTC)
 * If people are that concerned with the bottom-level items being at the same TOC level, something like this could always be done:


 *  Dungeons & Dragons creatures  (level 1)
 * (some text here)
 * Humanoids (level 2)
 * (some text here)
 * Reptilians (level 3)
 * (some text here)
 * Troglodytes (level 4)
 * (some text here)
 * Sahuagin (level 4)
 * (some text here)
 * Goblinoids (level 3)
 * (some text here)
 * Bugbears (level 4)
 * (some text here)
 * Other (level 3)
 * Minotaurs (level 4)
 * (some text here)
 * Dragons (level 2)
 * (some text here)
 * Other (level 3)
 * Dragon turtles (level 4)
 * (some text here)


 * Anomie⚔ 16:15, 1 June 2009 (UTC)


 * Yes indeed, insertion of an "empty" level 3 section solves the problem most of the time, but (as I suspect you have anticipated) it leaves the level 3 under Dragons in a somewhat inappropriate state: there is nothing that "Other" is another to. In my example, I can't see what level 3 header I could usefully put under "Carbon dioxide narcosis" as a supertopic of "Unconsciousness", while I remain convinced that "Unconsciousness" should carry the same weight as all the other outcomes in the example. Oh well, I guess the moral is that editors will have different views of the best structure for an article. As long as the bot doesn't enforce one view over another, it should still be useful in clearing genuine mistakes or highlighting where thought is needed. --RexxS (talk) 17:23, 1 June 2009 (UTC)
 * "Neurological effects"? At any rate, skipping a level for subject-level alignment purposes seems rather pointless as long as MediaWiki doesn't pay attention to that when generating the TOC. Anomie⚔ 19:48, 7 June 2009 (UTC)

Centralised discussion
I'm tentatively thinking about adding this to WP:CENT, given how constructive the discussion has been so far regarding improvements and so forth. I'd try to add it once a couple had been removed first, but does anyone have any major complaints? - Jarry1250 (t, c) 13:58, 2 June 2009 (UTC)
 * Sounds good to me. –Drilnoth (T • C • L) 14:13, 2 June 2009 (UTC)


 * Don't you think the discussion here (or there) might be moot, given the activities of Docu's (inappropriately named) bot, who seems much less conciliatory than this group? -- Michael Bednarek (talk) 14:30, 2 June 2009 (UTC)
 * No, I don't think that this discussion can be declared moot. There room in the world for both; we can discuss this separately from discussion about what Docu chooses (not) to do, so that we know where community consensus lies on the topic at large. - Jarry1250 (t, c) 14:38, 2 June 2009 (UTC)
 * The changes done by D6 are somewhat basic, more simple than the ones LivingBot is planning to do. BTW point 1 is unrelated to D6. -- User:Docu 06:31, 3 June 2009 (UTC)

Section blanking
Note that the actions of this 'bot will actively hide some forms of section blanking vandalism, making it less immediately apparent that such vandalism has occurred. Uncle G (talk) 16:18, 5 June 2009 (UTC)
 * (This was going to be longer, but my browser died:) That sounds interesting. Have you got any example? Presumably section deleters delete the entire tree... Anyhow, worth looking into. - Jarry1250 (t, c) 16:30, 5 June 2009 (UTC)
 * It's not section blanking, but the vandal might just wipe out the section header at the top, leaving the subsections to be "fixed" by the bot. I wouldn't call that "actively hiding" the vandalism any more than any non-fixing edit after vandalism "hides" it, though (any editor who thinks they're watching for vandalism on an article without looking over the entire history since the last known-good edit is IMO mistaken) . But to avoid even that, the bot could load the TOC as of some time T previous to the current time and refuse to edit if the current version's TOC is different to give human editors a chance to see it. Anomie⚔ 16:52, 5 June 2009 (UTC)
 * Ah yes, I saw your code for the basic OTC grabbing Anomie. Very helpful, I'll be sure to try it out before any future edits are made. As for this... well, what has to be done has to be done - Jarry1250 (t, c) 16:54, 5 June 2009 (UTC)
 * Here is an example that I reverted last month. Longer versions of the same thing occur.  Uncle G (talk) 19:15, 5 June 2009 (UTC)

The manual alternative
I thought that I'd give my own opinions on the manual alternative as it were. A bot is the epitome of cold, hard logic. It follows the rules you give it to the letter. Evidently, if not well thought through, this can cause problems. But what about turning the list of articles to be fixed over to humans?

It probably depends on the outcome of this RfC. If it is decided that the bot idea should be shelved on the grounds that sometimes level four really does look better after level two, what position does that put a human reviewer in? Do they change every one, or try to judge when it does look best and when it doesn't? The RfC would presumably rule the former out, leaving the human 16,000 style judgements to make. If it's shelved on ground of no net benefit, then does that rule out any human intervention?

And so the list of options goes on, but none of them looks a sure bet to me; perhaps the alternatives here are automated editing or no editing at all. Maybe that's a polarised view though. Thoughts? - Jarry1250 (t, c) 19:32, 7 June 2009 (UTC)
 * I for one just can't see how "make a list for people to check manually" could possibly work. Even if people were to be willing to go through those 15000 articles, if we accept the premise that "I don't like how H3 looks in the Monobook skin, so I'm going to make everyone everywhere have to deal with an inconsistent H4 instead" (which is the only reason given by almost all the opposers) then it'll just lead to edit wars between those who dislike how the H3 looks and those who find it more appropriate. Anomie⚔ 19:46, 7 June 2009 (UTC)
 * That, however, will be the case regardless of whether a human or a bot does it. ~ Amory (talk) 20:20, 7 June 2009 (UTC)
 * What will? I for one don't accept the premise mentioned and think it should be roundly rejected, if that's what you're referring to. As I mentioned elsewhere, if people have issues with the Monobook CSS styling they can take it to Special:MyPage/monobook.css, MediaWiki talk:Monobook.css, and/or WP:VPR instead of arbitrarily making articles structurally inconsistent. Anomie⚔ 20:57, 7 June 2009 (UTC)
 * I can't quite figure out what you're saying but to answer your question, I simply meant that the inevitable disagreement will take place whether it's human-human or bot-human (or, God forbid, bot-bot!). ~ Amory (talk) 21:15, 7 June 2009 (UTC)

I don't think that a list of 15,000 will ever be done without automated tools like AWB or bots. Just look at CAT:BACKLOG already... how fast are any of those being fixed? Who would be willing to work through such a list? Even if fifteen people got together to work on it (unlikely, IMO), that's still over a 1,000 articles per person. If AWB were made to fix this, it might as well be done by bot since for the most part the structure of each and every edited article wouldn't be checked... the user would just look at the proposed changes (which would be identical to those done by this bot), see if they were inline with "H4 after H3 after H2", and save, the same as a bot (it's a sad fact, but I think that for the most part editors using AWB just glance over the changes... it's actually kind of hard not to if you keep looking for the same thing over and over). So my feeling is that either A) This is going to be done in a bot-like manner, either by a bot or by a human who looks at each articles for something like 5 seconds using AWB, or B) The backlog will never be cleared. Who here wants to work on a backlog of this size doing the same thing over and over... and over and over... and over? After editing ten or twenty articles at a stretch, would you really continue to study the entire structure of each article to see what is appropriate? And if that started to happen, wouldn't it be easier to do with an automated tool that can determine what to do for you? And why not just make that automated tool into a bot so that you can spend time working on backlogs that can't be fixed by bot? –Drilnoth (T • C • L) 21:25, 7 June 2009 (UTC)