I'm not talking about the technical difficulties we've been experiencing lately, although those are extremely frustrating (props to Nichi for dealing with all of this and investing his personal time and funds on it ;_;). I'm talking about the system that AFF uses to sort, rank, and both indirectly and directly promote fics. In particular, this focuses on ease of finding fics to read and which authors/stories will get exposure, and to what extent.

This post is not really about the quality of fics. This is mostly about biases that exist regardless of the quality of the stories. I will address the implications and effects of these biases on the quality of fics some other time.

Definitions: Viewing a story means you actually clicked into the story to look at the description and foreword. To be exposed to a fic means to see it in its most basic form with title, author, tags, the first 200 characters of the description, etc.

Assumptions: Quality of writing is the same across the board or doesn’t matter when people are choosing fics to view/subscribe to. Time factors (when it was started, when it was completed, the time between starting and finishing, how frequently it is updated) and length of the fic don’t directly affect readers decisions to view or subscribe to a fic. The more likely you are to have exposure to a fic the more likely you are to click on it to view it. Exposure is the most important factor because none of the others (Rated M or not, who’s in the fic, personal tastes, etc.) matter if you don’t ever even see the fic in the first place.

Sorting Fics

Tags: The problems with tags are numerous. They are unused, misused, and/or abused, and they are limited in how helpful they are in the search for a good fic.

Unused: People don't tag their stories with tags that could help people who are looking for those things find them. Decreases exposure certain fics should get.

Misused: People tag their stories with things that are irrelevant and unimportant or not what people normally tag (usually it's idols, groups, group members, and genre), such as "sad," or "mind." There is little point in tagging things that nobody would search in the first place. Does nothing to overexpose or unexpose any fics.

Abused: People will tag their stories with every little thing that pops up in their stories, going as far as to tag members of groups who show up in all of one scene. I'm guessing this is done to lure in people who are looking for those groups/idols. Overexposes certain fics.

Limitations: Inconsistencies in tags mean searching a tag won't necessarily get you all of the fics that fall in that category or label. At a certain point, there are so many fics under a certain tag that it does very little to narrow your search down. Having a certain idol/group and falling into a certain genre do not guarantee anything.

Alphabetical: I honestly don't see the point of this. Unless you know the title of something and are looking for a specific fic, it doesn't really do much. I hardly think what letter the title starts with will have an impact on whether you want to read a story.

Rated: There aren't strict standards and guidelines for when to mark a story rated, and it doesn’t allow for different degrees of ratings. Different authors have different ideas of what is considered "mature content." In addition, it makes no distinction between violence, language/cursing, and ual content (the three most common things that people will rate stuff for), which have different levels of appeal or repulsion for different people independent of one another. Going off whether something is rated is not a good way to sort fics because it doesn't say a whole ton about the story's content.

Latest: Anyone who updates more frequently is going to have a higher chance of being checked out on any given day. When I browse through the latest fics, I'll see a lot of the same fics pop up again and again. I check the Latest section pretty frequently for the heck of it, but for anyone who's taking a look around for the first time or checks infrequently, you're bound to miss out on a lot of the less frequently updated fics. Oneshots will basically get overlooked completely.

Newest, Completed, and Oldest: You could argue that the bias from the Newest (oh, I'm going to think of Nu'est because of this lol) section is counteracted by the Oldest or Completed sections and vice-versa, but that leaves you with a bunch of fics that get lost because they're not at any of those extremes. The Completed fics are still further sorted by how recently they were completed, and most people don’t search all the way through completed fics or sort by oldest fics, so there is still a bias toward more recent fics.

Ranking Fics

Views: The factors in the sorting section influence exposure and therefore how many views fics get. How many views doesn’t say anything about whether the people who viewed the story enjoyed it, only that they looked at it. I’ve looked at more stories that I didn’t like/want to read than I can count, and I’m sure everyone else has. Nobody subscribes to every fic they look at. Also, a view from refreshing the page because of a server problem is different from a view from rereading the fic, which is also different from a view that’s due to a new chapter being added, but there’s no way to distinguish between those. Fics that have been around longer will naturally collect more views if the bias toward recent fics doesn’t win out completely. Ranking by views will simply help the fics with many views get more exposure while leaving nothing to be gained for fics with fewer views.

Subscribers: Number of subscribers perhaps says more about a fic than views, but it is influenced by number of views and therefore the biases from views will carry over. Also, just because a lot of people subscribe doesn’t mean that you personally will enjoy it. Different people have different tastes. Also, people are less likely to subscribe to fics that are completed because for the most part, subscribing is done to get updates. This once again leaves oneshots or completed stories with multiple chapters that are posted all at once in the dust.

I think maybe a better way to do rankings is to create a measure that tells you the proportion of subscribers to views. In particular, the subscribers should be total subscriptions ever minus the unsubscriptions that happened while the fic was still going to account for unsubscribers but still distinguish between unsubscribing because it’s finished and won’t update anymore and unsubscribing because you didn’t like where the fic was going or something. Also, it should only count the first views (which I don’t think there’s a way you can keep track of that except for registered users), perhaps only the first views on the foreword page. That would basically tell you what fraction of people who looked at this story liked it enough to subscribe, read it and liked it enough to stay subscribed, or subscribed even though it was finished because they wanted to keep a bookmark on it, that’s how much they liked it.

The flaw I can see in this is that different people have different ways of deciding whether or not they will subscribe to a story, if they use the subscribe function at all. For some, the foreword page pretty much seals or breaks the deal. For others, they give that first chapter a try. For still others, they read more before making that decision. Counting only the first view by each person would miss out on views where the decision to subscribe was made later. In short, ideally, we want to count subscriptions in a way that accurately measures how many people liked the story and count views that actually factored into the decision to subscribe.

Popular Authors: This is determined by total number of subscribers. The more fics you’ve written, the more likely you are to have more subscribers. It skews the ranks in favor of people who have written a lot of stories, regardless of how good those stories actually are. A better way to do it would be to have a ranking based on subscribers per fic. The problem with that is that it still doesn’t account for bias that comes from people reading fics because they’re by an author who wrote another fic that they liked. Quality of writing across an author’s fics is bound to vary. It’s possible the author will write one really good fic but then run out of steam, but loyal readers will read and subscribe to their other works anyway.

Promoting Fics

Although the exact formula for determining who gets recommended and featured is unknown, I think it’s safe to say that views, subscribers, and comments factor into this. I already discussed the biases from views and subscriptions. Number of comments is largely dependent on how many chapters there are. Even assuming that the number of chapters contributes no bias to how stories are viewed and subscribed to, which would mean the number of people who subscribe to longer fics and shorter fics and therefore comment on those fics would be pretty much the same, number of chapters still plays a factor in number of comments. Say you have twenty-five people reading a short fic with four chapters, and they comment for each time the fic is updated. That makes 100 comments. Take a longer fic also with twenty-five people reading who comment for each time the fic is updated. If it has twenty chapters, that’s 500 comments. Longer stories aren’t necessarily better, but there is a further bias toward longer stories.

In order to eliminate that bias, comments should be factored in based on number of comments per chapter, per fic.

Recommended fics: These are determined by overall, total traffic for fics. Fics that have been around longer are more likely to have more traffic.

Featured: These are determined by amount of traffic proportional to a certain time frame. Meaning there’s a difference between getting twenty subscribers over the course of a month and twenty in the course of a day.

People who get featured already accumulated popularity, whether based on their actual writing talent or the biases I’ve discussed, and then the feature merely multiplies the popularity, which often creates what you might liberally term "fandoms" for the featured authors. These fans then religiously subscribe and read their works, increasing the likelihood that those same authors will get featured again. In the same way that rookies will find it hard to compete with sunbaes for awards in the kpop industry, new writers will have a hard time getting featured when they’re competing with formerly featured writers.

Frankly, I think there should be some sort of limit on how many times a person can get featured. I think that a lot of these famous AFF writers already get plenty of attention and aren't about to fade out of the spotlight because, even if they won’t have their newer fics all over the front page, they have the AFF site promoting them already in the Past Featured fics section, not to mention as many as or more than a thousand or so people out there acting as their advertisers and promoters.

The AFF Loop

Here's a diagram of how it works. Arrows show direct relation, whether by cause-effect or association.

1. Number of views is dependent on amount of exposure.
2. The number of views is a way stories are ranked.
3. Number of subscribers is dependent on number of views.
4. The number of subscribers is a way stories are ranked.
5. The number of views contributes to whether a story is featured.
6. The number of subscribers contributes to whether a story is featured.
7. The number of subscribers contributes to the author’s popularity (referring to the "Popular Authors" section, i.e. total number of subscribers an author has).
8. The total number of subscribers is a way an author is ranked.
9. Ranking stories affects their exposure.
10. Featured fics will gain more exposure.

At certain points the looping effects can and will apply to the author’s other stories rather than the same story you started the loop with, and obviously the same story will not be featured twice, but aside from that, you could go through the loop pretty much forever.

In conclusion, the problem isn't that it puts people in the spotlight, and it isn't even whether or not these people are overrated. The bigger problem with the AFF system is that it can create a pretty monstrous self-reinforcing feedback loop. There exists an unfair bias toward giving exposure to stories with a certain traits, and that it can and does feature an author more than once, making it harder for other writers to get a chance to be featured. In order to gain headway in this system, people will write stories that cater to this system, thus furthering the existing bias.