Follow TV Tropes

Following

Trope Overdosed in need of a second revamp?

Go To

TompaDompa from Sweden
Jun 23rd 2017 at 10:23:21 AM

In 2012, Trope Overdosed saw its categories change from 1000-2000-4000 wick intervals to 1500-3000-6000-12000 following a discussion on this board. Before that change, the page was very cluttered (unfortunately I can't show you just how cluttered it was, as the page history doesn't go that far back), and now it's getting that way again. Consequently, I propose raising the thresholds a second time.

I initially brought this up at the discussion page, before bringing it here. I have created two sandboxes for different possible wick intervals: 3000-6000-12000-24000 and 3000-6000-9000-12000, respectively. 2000-4000-8000-16000 was also suggested, but I haven't made any sandbox for that suggestion (feel free to do so if you wish).

edited 16th Jul '17 9:05:37 AM by TompaDompa

Ceterum censeo Morbillivirum esse eradicandum.
Madrugada Zzzzzzzzzz Relationship Status: In season
Zzzzzzzzzz
Jun 23rd 2017 at 10:38:56 AM

Ok, you've made a valid point and presented possible solutions Good OP.

I would suggest that rather than incrementing arbitrarily, we start with "how many trope pages do we have?"

  • The Main namespace contains 51690 pages.
  • Assuming that perhaps a quarter of those are redirects, or pages that aren't tropes, or are tropes that do not allow examples. (that's not based on any hard research; just a gut feeling — many tropes have no redirects, quite a few tropes have several. There are Main disambiuguation pages for franchises, and for names that have several works.) It may be a bit off, but I think it's a safe number to work with.

That means that there are approximately 37,500 different tropes.

I would suggest that the top tier should start closer to the total number of tropes than the midpoint. We'll be revamping it again fairly soon if we set the number too low.

A work or franchise that has only 12,000 wicks is on fewer than a third of the trope pages. I'd suggest 25,000 as the minimum floor for "Saturated".

5,000, 10,000, 17,500, 27,500 has a pleasing not-quite-symmetry, to me. It also makes it progressively more difficult to move up a step, the higher you start.

edited 23rd Jun '17 10:59:34 AM by Madrugada

...if you don’t love you’re dead, and if you do, they’ll kill you for it.
Madrugada Zzzzzzzzzz Relationship Status: In season
Zzzzzzzzzz
Jun 23rd 2017 at 11:04:50 AM

It also appears that the entries have been alphabetized, meaning that the note about how to determine the ranking is incorrect. I would suggest that we go back to listing them in rank order, rather than alphabetically, with each entry noting how many wicks it has as a whole ( so a franchise entry gives the total count for all works listed as part of that franchise.)

...if you don’t love you’re dead, and if you do, they’ll kill you for it.
thatboy Relationship Status: Dating Catwoman
Jun 24th 2017 at 10:26:29 AM

While we are discussing about a revamp for Trope Overdosed, what are we going to do with the younger Trope Kilowick page? I personally suggest we merge works that have wicks between 1000-3000 together.

Madrugada Zzzzzzzzzz Relationship Status: In season
Zzzzzzzzzz
Jun 24th 2017 at 10:52:42 AM

The kilowick page is not a problem. You've suggested several times that we change it,, but failed to give any reason why it should be changed.

...if you don’t love you’re dead, and if you do, they’ll kill you for it.
TompaDompa from Sweden
Jul 16th 2017 at 9:06:15 AM

Somebody else made a 2000-4000-8000-16000 sandbox, and I made a 5000-10000-17500-27500 one.

edited 16th Jul '17 10:30:53 AM by TompaDompa

Ceterum censeo Morbillivirum esse eradicandum.
Jul 21st 2017 at 3:51:43 AM

If you look at the long April/May 2010 discussion on the discussion page, you'll see Fast Eddie himself removed the old rankings; while I haven't looked for any from-the-horse's-mouth explanation, I can see the argument that it turned the page into a dick-measuring contest for fanbases (potentially encouraging shoehorning), not to mention said long discussion was partly about how difficult the page was to maintain. The note about how to determine rankings is still relevant to determining what category, if any, each work falls in (not to mention that most if not all works on the page have commented-out notes listing links to the hundreds place) and in fact the bit about the Python script was added after the rankings were removed. In my experience said script doesn't work with modern versions of Python anyway; I suggest an alternative on the discussion page, but my experience with it suggests it's no longer entirely clear how combined link counts even work, or how they should work, or whether the counts are accurate for how they should work. (And one thing I don't mention is that even with the commented-out notes indicating what to include and what not to include, it wasn't always clear to me whether or not certain works should even count in a given franchise, especially with crossovers, and especially since different franchises may have different standards of what counts.)

In any case, if you look at to the 2012 "status reports" on the discussion page, you see that the way the existing thresholds work, typically each level has half or less the number of works as the one below. I don't know if this still holds; you can see that by the time of the last status report that November, the Overdosed category is approaching three times the entries of Overloaded, suggesting even by that point changing the thresholds was looking like a good idea.

Since that long 2010 discussion, both the 1000-2000-4000 and 1500-3000-6000-12000 thresholds have involved each level requiring twice the number of links as the one below (and before then the jump from Overdosed to Overloaded went from 600 to 2000, a jump of more than triple), and I would prefer to keep that. 5000-10000-17500-27500 means nothing to me - I only even realized what Mads was going for when I typed it out - while 5000-10000-20000 involves more pleasing, round numbers, but doesn't meet Mads' "top level over 25k" threshold, while adding one more level over 20000 results in requiring more links (40000) than there are tropes on the wiki. (The ceiling for the top level's threshold is 33,800, the amount Star Wars gets credit for... based on the last time it and many of the Smothered entries were checked in 2015. Any threshold higher than that would result in a category with no entries.)

If we start with 25000 for the top category as a base, cutting it in half results in 12500, and cutting it in half again results in 6250, which are even worse numbers than Mads' suggestion. 7500-15000-30000 makes a reasonably attractive set of numbers, but the next level down would be the uglier 3750. I personally prefer 5000-10000-20000, even if the upper threshold is less than what Mads would prefer, with the possibility of a 2500 level below it (establishing a stepping stone between 5000 and the Kilo Wick page), but I'd be fine with 7500-15000-30000. (That's if we have to go by Mads' criteria. 2000-4000-8000-16000 is most extensible, at least in the downwards direction, while maintaining pretty numbers - Trope Kilowicked effectively already is the next level down - and maintains the relatively small 500-link increase of the bottom tier of the last revamp, while 3000-6000-12000-24000 falls just short of Mads' mark and has the advantage of simply moving everything in Overloaded and above down a level, except those over 24000.)

In any case, based on what people said during the last revamp we should probably avoid any set of thresholds that include one at 9000. The temptation to meme would be too great.

TompaDompa from Sweden
Jul 23rd 2017 at 9:11:46 AM

It might be possible to go with 4000-8000-16000-32000 thresholds. We'd have to update the counts for a few entries (Batman, Final Fantasy, and Star Trek – Star Wars is above 32000 already), however.

Ceterum censeo Morbillivirum esse eradicandum.
Madrugada Zzzzzzzzzz Relationship Status: In season
Zzzzzzzzzz
Jul 23rd 2017 at 10:14:28 AM

Good points, Morgan, but frankly, I'm far more concerned with setting thresholds that we aren't going to have to readjust again in couple years than I am with aesthetically-pleasing ones. If we could have both, that would be great, but I don't think we can.

From a different point of view, the top tier is "Saturated"; by any definition in common use, something is "saturated" when it's at it's absolute capacity (a cloth is saturated when it can't absorb any more liquid, a solution is saturated when the liquid content can't dissolve or suspend anymore of the solid content, "market saturation" means that you have reached all the possible customers; you can't expand your market any more.) That's why I think that the top tier needs to be something over half the possible tropes; If those works saturate the wiki, then it should be statistically more probable that anyone clicking on the 'Random Trope' button will land on a page that contains an example from those works than that they'll land on a page that doesn't.

edited 23rd Jul '17 10:23:59 AM by Madrugada

...if you don’t love you’re dead, and if you do, they’ll kill you for it.
Jul 23rd 2017 at 11:19:43 PM

On the one hand, we don't have to have thresholds that result in a fourth tier that already exists, though the wiki itself may well be "trope saturated" enough that 40,000 is a bit on the high side. On the other, when Trope Overdosed had only three levels, the top level, if memory serves, was Saturated, not Smothered.

I can get behind Tompa Dompa's 4000-8000-16000-32000 suggestion, but the prospect of recalculating the numbers brings me back to the question of how to calculate counts for franchises spanning multiple work pages. I'm not going to rule out that I did something wrong, and I won't be able to get to the file I used to calculate these things until late Friday night at the earliest (and that's because I'm flying up to Seattle for a wedding on Saturday, so if I don't say anything Friday I won't say anything until late Saturday or even Sunday), but when I started to recalculate the numbers using the method I describe on the discussion page (in the days after I described it), Star Wars and Star Trek were the only franchises I recall being over 24000. Again, the provenance of the tropers involved in the history of the page makes me hesitant to jump to conclusions (and it's possible I should have included some pages that I didn't), but my suspicion is that at some point, people just stopped bothering with avoiding double-counting pages that link to multiple works in the same franchise.

TompaDompa from Sweden
Jul 24th 2017 at 3:49:04 AM

I can't speak for anybody else, but when I did wick counts, I used the naive method of adding up the different pages' figures. I suspect that most people did.

Ceterum censeo Morbillivirum esse eradicandum.
Jul 27th 2017 at 12:18:12 AM

I mean, I fully expect most people to do that sort of naive adding; the page itself presents the Python script as a more advanced option, and I wouldn't even have brought it up if the variance from one to the other wasn't considerable. Personally I think my approach is easier to work with and involves a program everyone has instead of requiring you to download a Python parser that the script won't even work with if you get one for the most recent version of Python, but that doesn't necessarily mean it's easy, especially for these big, huge franchises spanning over a hundred pages.

But what gives me pause is when tropers of good, long standing, who you would expect to be using the Python script if anyone was, who are taking it upon themselves to correct other tropers' link counts, come up with results that look an awful lot like those produced by the naive adding method (presumably only weeding out double-counted redirects). That's the sort of thing that makes me wonder if I'm supposed to be using the naive-adding method (even though there's no evidence on the discussion page that I should), or if my approach has a flaw that undercounts things for some reason (even though the only flaws I can think of would overcount things).

TompaDompa from Sweden
Jul 27th 2017 at 12:27:05 AM

Naive adding does inflate wick counts. That much is known. I would argue that getting too low counts is better than getting too high ones, so even if your method has that drawback, it would still be preferable to naive adding.

Ceterum censeo Morbillivirum esse eradicandum.
Aug 2nd 2017 at 12:16:28 PM

Okay, we might have a problem. I didn't get access to my old files until Monday, and when I did I found that I hadn't gotten as far along as I had thought; I had only just finished determining what pages would be encompassed in all the franchises that were listed as Saturated at the time after my early analysis of works around the current cutoff for the page itself. So I'd have to spend a long time calculating the numbers to even begin to come up with corrected counts for the top franchises.

In lieu of that, I did go through the Final Fantasy pages and came up with a quick estimate for that to get a sense of how overstated the current numbers are, since it encompasses fewer pages than most of its peer franchises (meaning I could get it done quicker) and it's listed as the largest franchise that's not one of the big "Star" franchises. After adding up all the links for all its pages, it came out to 36,121 links, which is a significant increase from the 30,600 it gets credit for now. After weeding out the duplicates, though? It had only 12,858.

Now, Final Fantasy is a franchise that spans multiple popular works, almost all in the same medium and mostly in the same genre using the same tropes (to the extent it's those tropes more than the actual plot or universe that links the games together), so there are a lot of pages that show up multiple times in the Related searches (you have no idea how many times Zettai Ryouiki was the last entry in the Related list), so the degree to which naive adding inflates its count may be higher than for other franchises. A franchise like the superhero franchises that spans multiple media and multiple works that don't necessarily use the same tropes might have a count closer to what they're being credited for, and the same goes for something like Doctor Who where a lot of the links can be attributed to a single work (that show has 14,490 intrawiki inbound links all by itself). So I decided to take on Star Trek, where there's at least a bit more variation in the tropes used across various series and films, where there are popular works on both TV and film allowing for a bit more diversity of tropes used, and which, just in terms of its appeal to the troper demographic and overall impact, it's hard to imagine more than a handful of other franchises beating. The total number of links I copied over amounted to more than 37,600. After weeding out duplicates? 17,324.

If those sorts of numbers hold up, it's not clear that any change to the current thresholds are needed; at best maybe a bump up to 2000-4000-8000-16000. Certainly Mads' 25000 minimum for the top tier would seem to be completely unworkable, at least without leaving the top tier completely empty. So many tropes are specific to certain genres or mediums, or are mutually exclusive with one another, or are highly specific or not thriving for some reason. If even Star Trek across all its series and films, not to mention its Expanded Universe, can't be truly saturated in the sense Mads has in mind, you have to wonder if a show reaching those heights is even possible. It's possible Star Wars or Batman, the other two franchises credited with more than 25,000, could at least come close, but looking at these numbers it's hard to imagine them getting much beyond 20,000. And keep in mind that a lot of these links are coming from other works (including work pages in the same franchise, including the franchise page itself, linking to each other), creator pages, and other things that aren't trope pages or their example pages, so the true count of tropes exhibited by the franchise or just trope pages mentioning it is even lower than these counts.

But if we were to decide to de-inflate all these counts, it would take a long time to do just at the start and would be a major pain to maintain thereafter, perhaps especially for the biggest franchises. I said that my method was easier than the old Python script, but it still involves copying over large numbers of very large bulleted lists, although Star Wars and Batman are the only franchises other than Trek for which I count over 100 work pages (with Star Wars Legends hitting that mark all by itself); for pages with as few as 500 links I find it easier to copy the whole page to another program, where I can shift-click to select the list, before copying into Excel. I may look into a way to adapt parts of how the Python script worked into my method to at least accomodate my inability to get it to work, but I'm not sure how much time it would actually save. It would be nice if there were a wiki function to combine several pages' link counts automatically, but that might be a major strain on the server and I doubt it would be a high priority for the devs anyway (though I imagine 2.0 could make it easier in several ways). I could see things very easily backsliding back to naive adding as people are tempted to inflate their franchises' counts and make things easier for themselves at the same time, possibly not even realizing how they're supposed to go about things, and creating a lot of extra work for those trying to keep the counts accurate, all for a Just for Fun page that so easily turns into a franchise dick-measuring contest.

edited 2nd Aug '17 12:16:54 PM by MorganWick

SeptimusHeap from Zurich, Switzerland Relationship Status: Mu
Aug 2nd 2017 at 12:31:28 PM

For the record, at the time of this writing there are about 25257 tropes (i.e pages with pagetype "trope" in the Main/ namespace).

Madrugada Zzzzzzzzzz Relationship Status: In season
Zzzzzzzzzz
Aug 2nd 2017 at 2:27:24 PM

Ok, I rather underestimated how many of the Man namespace pages were something other than tropes. It does sound like the overhaul needs to be getting better numbers, not necessarily changing the tiers.

Which makes me wonder if it's really worth doing at this point, since the devs are still in the planning phase for a complete database overhaul, which may give us an easier way to get accurate counts, once it's completed. But that won't be for a while.

...if you don’t love you’re dead, and if you do, they’ll kill you for it.
TompaDompa from Sweden
Aug 16th 2017 at 11:51:00 AM

I just noticed that my original sandbox (3000-6000-12000-24000) has been deleted, which is somewhat annoying, but whatever. I agree that we need to focus on fixing the numbers first of all. I'm afraid I don't have any suggestions as to how to do that in a way that isn't hugely labour-intensive.

Anyway, I created Sandbox.Trope Overdosed Corrected Counts and added the Star Trek and Final Fantasy counts above as well as a count for Whoniverse which I just did (20,288 by a naive count, 15,877 after removing duplicates).

Ceterum censeo Morbillivirum esse eradicandum.
Aug 17th 2017 at 2:41:57 AM

I'm working on my own corrected counts, though I'm not always certain what pages to include in what franchises or sub-franchises (though that matters less for this since even overdosed individual works usually won't affect the counts much unless it truly dominates everything else in the franchise). I just got 12,230 for The Simpsons. Star Wars actually came out to 21,195, more than I expected given the others. Most of the rest of what I've done is lower down just because franchises spanning fewer pages are easiest, but I had been planning to start my own sandbox once I'd done enough Saturated entries, and I've kicked several franchises out of Saturated already. (Among them is Pokémon at 10,128, which makes me think there are very few remaining franchises that'll stay in Saturated when all is said and done.)

edited 17th Aug '17 2:48:59 AM by MorganWick

TompaDompa from Sweden
Aug 17th 2017 at 4:13:20 AM

Right. We'll just have to see if the thresholds need to be changed when the corrected counts are done. I'm thinking maybe we should go back to three categories to keep it from getting too cluttered.

Ceterum censeo Morbillivirum esse eradicandum.
Aug 17th 2017 at 9:52:03 PM

Well, I think Maddie would be fine with keeping the threshold for Saturated as-is. Based on her "over half the possible tropes" standard, 12000 is still kind of on the low end as it stands.

So are we doing a full-fledged run-through correcting the counts, not waiting for 2.0 to potentially make things easier like Maddie suggested (which could well increase the currently-inflated counts if we decide each example counts separately)? If so, if I were to add all my findings to your sandbox page, should I keep the simple bulleted list you have, convert it to a full-fledged mock Trope Overdosed page like the other sandboxes but keep it in the order it's currently in, or do a full-fledged mock page in the alphabetical order used by the current page? Also, if you intend to check any more franchises yourself I would recommend checking the sub-franchises along the way; I didn't do so for Star Trek or Final Fantasy because I wanted one simple number for each to back up my point, but now one of us will have to do so for those two and the Whoniverse.

On the other hand, there's a change coming as soon as 1.8 that could prompt us to re-do all the counts anyway:

The Related view for articles will show works and tropes in distinct sections to improve readability.
That could raise the prospect of restricting the counts to just tropes, and not the various miscellaneous pages that happen to link to various works for whatever reason (and could well affect Overdosed Tropes similarly).

TompaDompa from Sweden
Aug 18th 2017 at 4:09:28 AM

Honestly, I don't know what to do. The current version is frankly pretty much garbage. I don't know how long we'll have to wait for 2.0 (or 1.8, for that matter). I think that makes a big difference when deciding what to do.

I probably won't be able to help much with the heavy lifting, because after next week I won't have much free time. If you have an idea about what would be best to do with the sandbox and the recounts, I say go with that idea.

I do think that when we get 1.8, we should restrict Trope Overdosed to just tropes and Overdosed Tropes to just works. That's what they're for, right? To see how many tropes a particular work uses and how many works use a particular trope, respectively?

If I knew Python (or another programming language), this is what I would do (maybe, as I have precisely zero insight as to what would be feasible):

  • Make a list of everything in the Franchise/ namespace (currently 352 works + 12 indices) or alternatively use the Franchise Index
  • Manually remove stuff from the list that doesn't belong (misclassified non-franchises and franchises that clearly won't be Trope Overdosed, for instance) and add stuff that's missing (large enough works that are in some other namespace, for instance)
  • Give each entry in the list subentries of everything that's indexed on the page
  • Manually assign correct indentations for subfranchises and their entries (and designate subfranchises as such)
  • Retrieve the "related lists" for each entry and subentry on the list
  • Weed out duplicates within each entry (including subentries)
  • Count the number of unique values within each entry (including subentries)
  • Filter the results based on number of unique values (so you don't have to wade through 100 entries whose counts are only like 700)
I don't know if that would work. Ideally, I should want a list whose entries and their levels of indentations are automatically generated but manually editable, and that performs the calculations automatically. Something like this (notice that the numbers don't add up because of overlap – Work 1 and Work 2 have several thousand tropes in common in this example):
  • Franchise 1: 14,000 unique values
    • Sub-franchise 1: 10,000 unique values
      • Work 1: 8,000 unique values
      • Work 2: 7,000 unique values
    • Sub-franchise 2: 5,000 unique values
    • Work 3: 3,000 unique values
  • Franchise 2: 6,000 unique values
  • Work 4: 5,000 unique values
  • [...]
Does that make sense?

edited 18th Aug '17 4:25:45 AM by TompaDompa

Ceterum censeo Morbillivirum esse eradicandum.
Aug 19th 2017 at 1:06:14 AM

Well, I have an Excel file that I use to keep track of what works fall under what franchises, but a) it doesn't necessarily match what is or should be used and b) it would probably be more readable by me than by a machine without a lot of editing (I use very narrow columns to simulate indented lists).

Madrugada Zzzzzzzzzz Relationship Status: In season
Zzzzzzzzzz
Aug 22nd 2017 at 10:24:59 AM

2.0 is a long way off.

...if you don’t love you’re dead, and if you do, they’ll kill you for it.
Add Post

Total posts: 23
Top

How well does it match the trope?

Example of:

/

Media sources:

/

Report