This page is an archive of a community-wide discussion. This page is no longer live. Further comments or questions on this topic should be made in a new Senate Hall page rather than here so that this page is preserved as a historic record. Advanced Jedi Training Droid 6(Talk to my master) 01:05, January 12, 2015 (UTC)
This post is pretty long, so here's a summary:
Some of our articles are too long, and their lengths are slowing down render times and slowing down page load times for mobile users. They also can't be read in one sitting by people with an average concentration span. To fix this, we should split our long articles into sub-articles.
Now, for the long version. Over the last week or so I've been looking through the templates on this wiki, identifying ones that slow down page render times, and converting them to Lua. Lua is great for speeding up complex templates: for example, the new version of Template:Eras that uses Module:Eras is about ten times as fast as the old one. However, for less complex templates Lua doesn't help very much. There is a slight overhead when switching from PHP to Lua, and this means that for templates which only have one or two parameters, a Lua version may actually be slightly slower. To find the complex templates that would benefit from conversion I've been looking at pages with long render times, like Wedge Antilles and Palpatine, and going through all of the templates transcluded on them to see how complex they are. I've already converted a few, and there are some more that would still benefit from being Lua-fied (e.g. the infobox templates), but most of the templates left over are simple and won't benefit from conversion.
However, the page render times on Wedge Antilles and Palpatine are still very long - both are around the ten-second mark. In retrospect, the reason for this is obvious. The pages are simply too large. The Wedge Antilles page is currently 659 kb, and the Palpatine page is currently 475 kb. Compare this to Wikipedia's advice that articles of more than 50kb should be considered for splitting, and articles of more than 100kb almost certainly should be split. There are a few things we could do to speed up page render times with these large pages, for example getting rid of the small images in citation templates, and altering Template:'s so that it doesn't use span tags. But these kinds of things alter article content, albeit slightly, so users here may be wary of them, and they may only save a small amount of time spent rendering the page. To get big savings on rendering time, we will need to make the pages shorter by using summary style.
Long pages are not only bad for rendering time - they are also bad for mobile users, and users with slow connections. For example, on my old iPhone 4, the Palpatine took about 30 seconds to load completely, including the images. And my phone connection is 3G, so for readers in developing countries who have slower internet than I do, some of our pages may be completely inaccessible because of their size.
Another reason to make a switch to shorter articles is concentration spans of readers. At the moment, a long article like Wedge Antilles would take a long time to read (perhaps a couple of hours). Organising this content into different pages would enable readers to get a good overview of the subject before their concentration spans run out (according to the Wikipedia page I linked above, that's about 40-50 minutes), and they could always click on one of the links to the sub-articles if they wanted to know more.
So, to sum up, I think we need to start going through some of our longer articles, and splitting the content out into shorter articles. For example, we could have a Personality of Palpatine article, a Powers and abilities of Palpatine article, and perhaps a few different articles for periods in his biography, all linked from sections of the main Palpatine article. And then we could cut the main article down in size drastically, without losing any of the content. What do other people think about this? Mr. Stradivarius (talk) 16:53, December 7, 2014 (UTC)
This has been discussed occasionally, and always rejected. Wookieepedia's policy is "one subject, one article". That said, I just resurrected an old project I was working on a couple years ago that I never did anything with. You can see it at User:Master Jonathan/Wedge Antilles. This is essentially a mockup, using an old version of Wedge Antilles, of how the "one subject, one article" rule could be preserved, while still allowing the article to be split into multiple, smaller pages. The draft guidelines on how this would be done are at User:Master Jonathan/Sandbox. What does everyone think about this? Obviously, some tweaks would likely be needed, as well as updating things for new policies since 2012, but it's a start. —MJ—Jedi Council Chambers 18:46, December 7, 2014 (UTC)
After looking at your mockup, I still disagree with the idea. Too much effort, not enough reward. Besides, it looks sloppy.AV-6R7User talk:AV-6R7 18:55, December 7, 2014 (UTC)
How does it look "sloppy"? Remember that this is just a mockup, and one that has been deleted for the better part of a year and is using a two-year-old version of the article. So of course there's going to be some redlinks, cite errors, and other broken stuff; none of that will be in a finished product. And as I said, tweaks will be necessary; I have no illusions that the idea would gain approval as-is. I'm open to suggestions. —MJ—War Room 19:01, December 7, 2014 (UTC)
Having huge articles is having serious adverse effects on render times and load times, though. I don't really mind how you do it, but my main point is that to make this wiki reasonably accessible and responsive, page sizes need to come down. Long load times and less accessibility means less readers, and less readers means less potential new editors. Mr. Stradivarius (talk) 10:02, December 8, 2014 (UTC)
For some the load time is slow but for others not a problem. I loaded Wedge in 4 seconds on my laptop and 7 seconds on my mobile device. Fe Nite (talk) 20:19, December 7, 2014 (UTC)
I am 100% for this. Too often we reject ideas around here just because a while back it was decided, "no, let's not do that" and then we just stick with that even though things have changed. The fact is that many of our articles are too long, much too long, and it's a problem for exactly the reasons said-- big turn-off for people using mobile devices and huge page-size in general that might throw the average reader who we as regular editors often don't consider. Remember, probably about 80% if not more of those reading will never make an edit, in many cases never even register an account. We need to keep their needs in mind. ProfessorTofty (talk) 01:56, December 8, 2014 (UTC)
Of course on newer devices and faster connections the article will load without too many problems. But for those with slower connections and older devices the size of the page makes a big difference to how fast it loads. I would think a majority of users would give up after 30 seconds of waiting for the page to load, unless they had a really strong reason for wanting to read it. In fact, seven seconds is still a long time in terms of application response. If you look at the established theory on response times, you can see that a) operations that take longer than one second disrupt the users train of thought, so the user should be shown that the computer is doing something (perhaps by some sort of animation), and b) operations that take longer than 10 seconds are ones that users don't like to wait for, so there should be a progress dialogue and a clear way to cancel the operation. We're failing on both points with the load times here - whether the load time is four seconds or 30 seconds, from a response time perspective we're still treating it as if it should take less than one second. Mr. Stradivarius (talk) 10:02, December 8, 2014 (UTC)
I don't like the idea of breaking up articles. But I do recognize the problem of loading times on slower devices. In this age of technological marvels and computer geniuses, is there no way to increase the load time of large articles? Perhaps we can ask the Tech Gurus at Wikia for help?--Richterbelmont10(come in R2!) 05:42, December 12, 2014 (UTC)
Where is a tub of proverbial lubricant when we need one? - AV-6R7User talk:AV-6R7 05:46, December 12, 2014 (UTC)
Large is just... large. Sure, there are tricks for speeding stuff up, like that Lua thing, but there's no getting around the fact that having large amounts of stuff on one page is going to slow things down, especially when the page has a lot of images. ProfessorTofty (talk) 09:04, December 12, 2014 (UTC)