3572619
You have the chronology wrong a bit, which makes me think that I really do suck at explaining. Chronologically, the very first event that happens is the appearance of darkfire phoenix. Spitfire was only a Wonderbolt then—it is the death of her comrades that makes her take up poaching for the first time. The darkfire phoenix disappears shortly after the death of the Wonderbolts—in the expanded explanation, the phoenix burned so fiercely, it reached the end of its lifespan every morning (as opposed to an ordinary phoenix, which can take months or years), and then resurrected at dusk, making it hard to detect (for someone unfamiliar with the art of killing phoenixes). Finally, it burned itself out entirely, falling into ash for several years. It's during this time that Spitfire racked up her phoenix kill count and earned experience and her burn scars. The narrator joined her somewhere during this period mostly as a way to get money. Being unfamiliar and disinterested in the history of the Wonderbolts, he did not make the connection between the disappeared captain of the famous flying team and his mentor. He gained respect for her over the years like any loyal apprentice gets for a capable master, but it is only after the events of the story that he truly begins to admire her.

The two find the darkfire phoenix by pure accident (or perhaps fate? Though I admit, this part could use some work.) while on another of their hunts. There is no way to distinguish one pile of phoenix ash from another, so they couldn't know before it resurrected, and nobody even knew if the darkfire phoenix that killed the Wonderbolts would ever come back at all. They found the phoenix just as it was at the end of its long resurrection cycle, before it could begin a second rampage. The guards that captured the narrator were following their trail from a tip left at the town where they last stocked on supplies before leaving, and they found them by flying towards the explosion of the phoenix.

I don't think that they would receive a royal pardon—it's likely that the petrified corpse could be identified as the specific phoenix that caused the rampage years ago, but preventing its second coming hardly excuses their crimes. It would if they have poached just a few phoenixes—but between the two of them, they killed almost a hundred endangered, intelligent animals. Besides that, darkfire phoenixes appear very, very rarely (a certain chain of events must happen to a phoenix egg for it to hatch into a darkfire phoenix), so it's not likely that their expertise could be of much use to the kingdom (even though a lack of that kind of information is exactly what got the Wonderbolts killed, but hindsight's always worse in government officials). Spitfire may have been given leniency or even a pardon in light of her actions and her former status as a Wonderbolt, but she perished in the explosion. The narrator had neither her status nor her character (he was in it purely for the money, after all), and his role in bringing down the darkfire phoenix consisted mostly of triggering the ineffective traps and covering in fear inside the foxhole. Hardly the heroics needed for a pardon.

As you can see, this definitely needed a longer wordcount.

3571372

I think the writers, when voting, would be particularly mindful of the limitations inflicted by that time limit, having just personally experienced them

First, no; most of the variance in voting appears to have come from
people who fundamentally didn't get what the stories were about, because they weren't revised.

But I don't care so much how it affects the voting. I care that we wasted most of our discussion time trying to figure out what was supposed to have happened in different stories because the writers didn't do a second draft, or because they were trying to cram 1800-word stories into 750 words.

Think of it this way: Would anybody hold a speed-art exhibit? A gallery full of paintings that all had to have been completed in one day? Sculptures that hadn't cured yet, whose heads and legs were slipping down towards the floor? Maybe, as a gimmick, but it would be sport, not art.

3572769
I agree, and that's why, back when I and Golden Vision were more active in the group, we made the time limit (for the regular stories, not minifics) into a whole week. I don't know why we returned to the old format, but I'd love to see a return to a longer writing period.

3571836
It is possible to write a story without conflict, but it is extremely difficult to write a good story without conflict. This is as true in the East as it is in the West, which is why virtually all good Eastern stories contain conflict, and why virtually all good Western stories contain conflict.

It is ultimately about being engaging and leaving the reader feeling like they're getting something out of reading your story. I mean, you can write something like One Ordinary Day, with Peanuts, but such stories are few and far between, and honestly, even that story isn't all that great; it is mostly the ending (if you make it that far) which makes it what it is.

3569882

Belated thanks. I sat down and started scribbling, was 300 or so in when I remembered you'd already -done- this bit, then said 'eh, fuckit' it and wrote it anyhow, because I wanted to get something out.

Then last night I realized what I -should- have done was take Queen Twilight's 'Good Ending' and put some sort of twist on it, which would have actually been different.

But the main point was just to do something, since I only found out this existed within 12 hours or so of the contest originally ending!

Lastly - there is an amusing twist that Listen, the fanfic of a fanfic everyone compares this to as the one done right, was written by you.

3572819
Well, twisting her good end is pretty easy. In fact, I thought of it the moment I read her good end, because it ended with Well, that worked.

And of course, it had to.

But if time REALLY had stopped, then she wouldn't have been able to think that, now would she?

As author of Disconsolate, I am deeply disappointed in myself that absolutely no one fully received the message I was trying to send; that being, the terror that lies within what seems to be a logical fault pointing to an obvious and terrible conclusion, excepting that there is an equally probable conclusion assuming that a piece of information is misrepresented.
What I mean by that is, the goal of the story was that at the end you wouldn't come out feeling sure about what Rose did or didn't do. I hoped that with the cutoff at that final line I'd leave someone gasping, "But who's right? Did Rose kill him, or is Twilight misremembering something? Was Rose really just babbling insensibly?" And then the story was over, so the question would never be answered--or maybe, you'd have to answer the question for yourself.
Obviously, this was not what I accomplished. I originally had a longer introduction, which I think helped set the tone of the story, but it had to go for word count. I also readily admit that Twilight's voice is off (which is strange, because normally one of the few things I'm reliably good at is character voices) and her reason for being at the funeral is weak. Beyond that, were I to rewrite, I'd be sure to include a lot more speculation by Pinkie regarding what Twilight is recalling--Pinkie might've disagreed, for example--and just generally create far more tension between the two conclusions one might come regarding just what happened behind the scenes. Because it seems like at present people are far more likely to conclude that Rose killed Goldengrape, and that isn't what I wanted at all.
Anyway, this story was written the same week as I watched four episodes of BBC Sherlock in half as many days, so if nothing else my inspiration is obvious.

3572791 On the contrary, all short stories in literary magazines for about the past 50 years have been required not to have obvious conflicts. The trend, since John Updike started writing for the New Yorker 60 years ago, has been for literary short stories to make a game of seeing how many crucial story elements they can throw out and still be interesting in some way.

If you gather back issues of The New Yorker, The Atlantic Monthly, Tin House, The Kenyon Review, and Narrative, and read fifty stories, you'll find most have no obvious conflicts. They are more likely to have problems, but they aren't conflicts because the characters don't try to solve them. They just fall into despair.

3572853
I have a hard time even reading a lot of stories published in The New Yorker, because I find them so dreadfully dull. They don't hook me, and even when I force myself to read them, my mind asks me, why? Why do I subject myself to this? Reading is supposed to be fun. Reading textbooks is fun. Reading about science is fun. Reading literature should be fun, or at least enjoyable.

I hadn't read any stories from it in many years. So because of your post, I made myself read some. I looked up some recommendations, and found this list, and looked at the free stories.

I couldn't bear to read more than a few sections of The Largesse of the Sea Maiden or The Embassy of Cambodia; they were that dreary.

I had to force myself to read Town of Cats, which centers on the conflict between a man and his father, and ultimately their reconciliation. I took no joy in reading it, and the only reason I went all the way through it was because I forced myself to do so. It didn't emotionally engage me, it just inspired a sort of dreariness in my soul and made me wonder - why? Why am I doing this? Why am I subjecting myself to this? There was something there, in the very end, but it took far too long to get there, and the writing was not beautiful.

The Bear Came Over the Mountain was almost readable, even if it drug in the middle. But it had conflict AND a resolution! Crazy. If they published too much stuff like that they might lose their rep. Alas, I would not say that I enjoyed it. In fact, I would say that it was very long for what it was.

Symbols and Signs was short (and contained a typo/copy error which stuck out at me), but was the sort of nasty, despairing story you were talking about. It sort of had conflict, but then didn't actually have them push back at all, and in fact just outright stated:

All this, and much more, she had accepted, for, after all, living does mean accepting the loss of one joy after another, not even joys in her case, mere possibilities of improvement. She thought of the recurrent waves of pain that for some reason or other she and her husband had had to endure; of the in visible giants hurting her boy in some unimaginable fashion; of the incalculable amount of tenderness contained in the world; of the fate of this tenderness, which is either crushed or wasted, or transformed into madness; of neglected children humming to themselves in unswept corners; of beautiful weeds that cannot hide from the farmer.

Alas, it lacks self-awareness of its own ridiculousness.

And possibly ends with their mentally ill son committing suicide.

A lot of these stories seemed to involve mental illness. It's like Rainbow Dash breaking her wing, except instead of an overused setup for a shipfic, they just wanted to make things horrible.

I dunno. I can't imagine sitting down and reading these and enjoying them. The writing in them was unenticing in many cases, and outright ugly in some of them. I suppose the ugliness of the writing reflected the ugliness of the stories, but... why would I want to read ugly writing about an ugly story?

Maybe it is just the sort of thing people read so that they can convince themselves of their personal depth? They read a story about someone losing their mind in a mental health care facility and someone's mentally ill son trying to kill themselves; they must be deep!

3572769 3572779 I'm all for longer contests. While 24 hours is more than enough time in principal, it becomes very difficult if you have other commitments the day of the contest. I was rushed to complete mine in a few hours of late night work, and the result is something that I'm not overly proud of. As Bad Horse said, it makes little sense to have authors putting up entries that even they don't consider finished or ready.

While there's a place for "sport" writing (EFNW's 2-hour Iron Author for example) 24 hours is a weird middle ground. Long enough that some entrants, who are lucky enough to have a free day that coincides with the contest, can write and edit something to a pretty good polish, but many with other commitments are forced into speed-writing mode. I'd really love to see things go one way or the other. A dedicated, short window if you want speed-writing, or something more like a week-long contest to maximize the potential for quality entries. I would note that it may be necessary to restrict the number of entries per author in that case, or we'll all spend a month reading nothing but Titanium Dragon stories.

3572769

because they were trying to cram 1800-word stories into 750 words.

A valid concern all on its own. 24 of the 51 stories were within 10 or fewer words from the max cap. One could argue that a 1800-word idea is not an appropriate choice for a minific competition, but perhaps the cons of the tiny 750 limit outweigh the pros. For instance, you can still say a lot in 2000 words without feeling claustrophobic, yet still being required to write a small, contained idea. Of course, 24hr for 2000 words would be far too short a time period, so you'd want at least 48hr, maybe 72hr. Or perhaps, there's little merit in restricting to shortfics only, and we should go for the more usual (for this group) 20,000 max cap.

Maybe, as a gimmick, but it would be sport, not art.

Wwwwell... yes? This competition group, and the recent EqD Outside Insight contest, and pretty much any writing contest enforces the following: a time limit, a minimum and/or maximum word-count limit, a scoring and/or judging system, and awards for overall best entries (and possibly for categories, like the judges' choice awards this round). I think these competitions are unquestionably a sport-ification of writing. I wouldn't go so far as to suggest that art cannot result from it, given the number of successful stories that have been spawned onto Fimfic proper as a result of these contests, but sure I agree with you, the sport nature does interfere with artistic integrity.

But that's just it, though. A contest has rules, which are restrictions, which de-facto interfere with artistic integrity. What if the author didn't want to use our prompt? What if the author wanted to use more words than the max cap? The snug time limit is absolutely intended to apply pressure to the writer and limit the amount of time they can devote to brainstorming/outlining/drafting/polishing. Like, maybe the author really needs over two years to write that second chapter... Nah, nevermind, that just sounds like laziness. The only way to allow "true art" is to minimize the restrictions as much as possible. That said, it seems that with the number of participants we get, the writers are tacitly agreeing to some level of sport-ification. It's just a matter of tweaking the variables.

Think of it this way: Would anybody hold a speed-art exhibit? A gallery full of paintings that all had to have been completed in one day? Sculptures that hadn't cured yet, whose heads and legs were slipping down towards the floor?

Serious answer, if mildly smarmy: The Navarre Beach Sand Sculpting Festival offers its master sculptors only nine and a half hours to work on their sand sculptures. Its amateur division only has a scant four hours to complete their masterpieces.

3573412

2 hour contest

Time zones. In this Fimfic group alone, we have a bunch of people in the US, a number of people in Europe, a few in SE Asia... The principle is that you'll have some time in that 24 hour window to write, no matter where you live. But you're still right, RL demands (or the lack thereof) will give an extreme penalty (or bonus) to some authors, which is hard to avoid for a short time limit like this. If we were doing a strict 2hr limit, it would need to be honor system.

3572769 3572779

From my experience, it was more the 750 word limit that severely hampered my stories, rather than time to complete them. I wrote both of those within a 2 hour period each, and went back and looked them over about an hour before the deadline to see them with fresh eyes. I just didn't make use of my 750 as efficiently as I should have.

3573497

of the in visible giants hurting her boy

and the giants that are in plain sight?

3571839

I had to stop and chant a modified MST3K mantra to refocus before continuing.

The first thing that popped into my head was "I'm a q-tip, what are you?" and that has made me incredibly happy.

3572457
...Oh crap, I didn't get that reference at all. XD Have you considered my suggestion to change the title?

3572779
I really hated the week-long writing period. Maybe it's just me, but having that long to write something ensures that I will spend at least four days doing jack-all and scramble to throw something together at the last minute. Plus, it encourages longer works, and having a few hundred thousand words to read at the end of the contest is never fun.

3573545

Plus, it encourages longer works, and having a few hundred thousand words to read at the end of the contest is never fun.

That would be tough to get through. We had 35k words this time and, although I read them all within the first few days, I started feel less enjoyment towards the end and more like work. Though I think with longer stories we could do the preliminary round so you only have 5-10 stories to read.

3573525 I was using the two hour "Iron Author" contest as an example of sport-writing, not actually saying we should try two hour limits here. I've spent years in IT work for global companies, supporting people all over the world, so I'm probably much more aware of the issue of "it's not the same time everywhere" than most.

On that note, there are (theoretically) ways to enforce short time limits technologically, without requiring a universal "start time" but most of them aren't worth the hassle for something like this. I'd just rather see a 1,500 word, week long contest, limited to maybe two entries per author. Then have two judging rounds (with judges randomly assigned stories) so no one has to read more than maybe 10 stories.

3573506

A valid concern all on its own. 24 of the 51 stories were within 10 or fewer words from the max cap. One could argue that a 1800-word idea is not an appropriate choice for a minific competition, but perhaps the cons of the tiny 750 limit outweigh the pros. For instance, you can still say a lot in 2000 words without feeling claustrophobic, yet still being required to write a small, contained idea. Of course, 24hr for 2000 words would be far too short a time period, so you'd want at least 48hr, maybe 72hr. Or perhaps, there's little merit in restricting to shortfics only, and we should go for the more usual (for this group) 20,000 max cap.

Yeah, but that doesn't necessarily mean anything. Three of my stories were at 750 words, but Final Witness's first draft was exactly 750 words long, and The Dying Words of Starswirl the Bearded was like 742 words on its first draft. That just means that I was very good at estimating the legnth of my stories and wrote things which exactly fit into the given space.

On the other hand, Moving Heaven and Earth was cut down by a couple hundred words and I was conscious of the word cap as I was writing it. That being said, the second scene could have been written as a 750 word story, and really should have been, but I missed the correct way of doing it.

I think these competitions are unquestionably a sport-ification of writing. I wouldn't go so far as to suggest that art cannot result from it, given the number of successful stories that have been spawned onto Fimfic proper as a result of these contests, but sure I agree with you, the sport nature does interfere with artistic integrity.

I think that people also forget that restrictions breed creativity; I wouldn't have written several of these stories if it were not for the contest. Mark Rosewater often talks about this as far as game design goes, but it is true that if you create tight restrictions, you can really make for creative writing. Look at Magello's Most Dangerous Game contest entry, for instance, or The Cat In The Hat.

It is true that you CAN end up with really dismal results, but on the other hand, restrictions like this can make people write very clever things.

The idea that restrictions interfere with artistic integrity is very questionable; heck, the show we're all fans of is a result of Hasbro telling Faust that she couldn't write an adventure show, and MLP:FIM is vastly the better for NOT being an adventure show, but a slice of life show which dips into adventure.

3573777
Oh, I can definitely relate. I'd have never written a donkle donkey fic without getting strange inspiration from a contest. And despite the cons, having a tight deadline does huge for my personal motivation and word-per-day output. So I'm definitely a huge fan. There's a reason I offered to do reviews for this group without having actually submitted anything this time 'round.

3572779
Because deadlines get people to work.

I don't notice any increase in revision quality on the whole in the events where people were given more time. There's little correlation between the average wordcount of entries and writing time allocated. Look at the data:

https://gist.github.com/RogerDodger/b1d8d4c85f7badca5627

An analogy that I think was used before: speed writing has about as much to do with regular writing as speed chess does to classical chess. If you look at the top players' ratings, you'll see some of them don't do as well as others with less time, but they are still very strong players in all situations. (And Carlsen is simply the best under all time controls.)

This is a speed writing competition. That said, as far as speed writing goes, days of allocated time is actually quite a lot. As mentioned, going anything not in the magnitude of days would make time zones an issue.

Right now, I don't see any compelling argument to change the allocated writing times. The data says that changing it has little effect, so I favour the status quo.

3570829
3570901
3570911
The current scoring system is a bit of a mish mash of intentions with varying degrees of success. I'm not entirely happy with it, so it's going to be changed soon.

The scoreboard is supposed to say "people nearer the top are doing better". That's a *really* handwavy answer, though, so it's not very useful.

The intentions of the scoreboard is to incentivise strong participation. I didn't want people just spamming entries for points, so I thought I should add a cost to a poor performance. In addition, I thought this would serve to circumvent repeat participants from simply accumulating a huge score.

Having things be entirely zero-sum, though, ends up being too punishing. Obviously letting someone sit on a negative score would be a bad idea, because that would simply incentivise people creating new accounts to zero out their scores. The end result is immediately obvious if you look at the scoreboard: a huge number of people are sitting on a score of 0.

It was also apparent that some events were more difficult to place in than others, so scores should be weighted accordingly. Using just the number of entries worked as a useful heuristic initially (more entries implies placing higher is harder), but then minific contests got introduced and started skewing the results.

Anyway, to sum up the goals of the scoreboard:

(1) Rank participants in terms of performance across all events
(2) Encourage participation
(3) ...without encouraging spammed entries
(4) Allow new participants to catch up to old ones

The zero-sum scores achieves (3) but fails (2) and (4). The "score across recent events" attempts to patch (4) in, but keeping the "all time" score as a separate column somewhat demeans its value. The scores being proportional to entry count attempts (1) but disproportionately favours minific contests.

So, the changes I'm thinking of that better achieve these goals:

Events are weighted not just proportional to the number of entries, but also to the word count of the entries. However, this shouldn't just be a linear proportionality, because I don't think it's accurate to say a contest with an average word count of 4,500 is 6 times stronger than one with an average of 750. I've experimented with a few different calculations:

https://gist.github.com/RogerDodger/a87763e554c967757e6f

The last three columns are the different weights.

The first is the sum of the logs of the wordcounts. log on the whole didn't produce very useful results because it's too concerned with orders of magnitude. The end result is almost no difference in weight based on word count, and because it's a sum, direct proportionality to n.

The second is the sum of the square roots of the wordcounts. The numbers here I like, but it's still a simple sum, so it's directly proportional to n. I'd rather the weight be directly proportional to the square roots of both n and average wordcount, so dividing this sum by square root n does that. (n divided by sqrt n is sqrt n for those who forgot basic algebra.) The third column shows this figure, which I like.

With the calculation of the difficulty of an event considered, there's then the distribution of this weight amongst entrants that has to be considered.

The first and simplest distribution I considered is simply linear: 1 - i/n. First place gets 100% of the points, middle place gets 50%, and last place gets 0%.

I then considered the probability that one would place at a certain position, and thought to based the distribution on that. Assume some difficulty d for coming 1st in competition A. Competition B has twice as many participants, so it stands to reason that coming 1st will be twice as hard: there's twice as many people who can beat you. Coming 2nd in competition B is as easy as 1st in competition A, because you can let 1 person beat you. The same applies for 3rd and a trebled number of entrants. In other words, placement relative to first is 1/i easier, so the distribution is is 1/i. A noticeable property of this distribution is it doesn't involve n, and also never reaches 0, so even last place gets some amount of points, and for example in an event with only 2 entries, last place gets half the points. This might seem strange, but it also is accurate with the assertions: there's a 50% chance that a random entry will get at least that placing. This curve would achieve goal (1) but hurt goal (2), because most entries will get very little score.

The more I think about the above, though, it seems a bit strange, and involving n with the calculations seems to be double dipping when it's already considered in the event weight calculation. It's also incredibly brutal, giving almost all the points to 1st and none to the rest. Still, it's something to consider.

The last thought was to curve out the linear distribution some, i.e., using trig functions, making it look something like this. This has the same property of linear distribution, where first, median, and last place get 100%, 50%, and 0% respectively, but curves out at either ends.

I'm leaning more towards the latter distribution, since it rewards near-first entries and punishes near-last entries a little more, but otherwise maintains the property that position is proportional to score. It achieves both (2) and (1) to some extent.

Goals (3) and (4) haven't really been considered yet, and that's intentional. I think the scoring algorithm doesn't need to concern itself with those goals, because they can be achieved more simply elsewhere. (3) can be achieved by limiting an artist to receiving a score only for their best entry in a particular event, and (4) can be achieved by applying a diminishing score to entries the older the event is. Specifically, every time an event finishes, all existing scores are multiplied by some factor. As they become older, scores will become worth less and less until their impact on, for example, the leader's position becomes almost negligible. (What factor is used depends on how quickly we want old scores to become negligible. For example, to make scores from 10 events ago be worth only 10%, the factor f needs to be such that f^10 = .1, i.e., f = .1^(1/10) ≈ .8.)

I've made quite a few assumptions and assertions here, so I'd like to hear whether or not any of them are dubious before going ahead with things.

3573843
I can't math, so I'll just ask, what's the cutoff for stories that receive negative scores? 50%? You could always drop that down to 40 or even 30% to increase the stories giving positive scores.

(3) can be achieved by limiting an artist to receiving a score only for their best entry in a particular event

This might be a fallacious assumption. It could encourage someone like, say, me, to write one good story followed by three Hot Shot and Hugh Jelly sequels. The one good story will likely give me more points while the others won't be a detriment, while simultaneously sending the overly-uptight fic reviewer participants into conniptions and generally wasting their time. The changes of the good story providing me with a positive score is also higher because I've increased the entry pool and ensured that at least three stories will be in the negative ranks.

3573896
Can be countered by taking out the other stories completely from the list when calculating point awards; in other words, the points are calculated as if every author only wrote his or her best story.

(Well, apart from the author wanting to just troll the community, but I don't think this would be a problem. Besides, it can already be done right now by creating fake accounts.)

3573896
Right now, the cutoff is at the median. I could add an offset to any of the distributions to punish for example the bottom 20% of entries with negative scores, specifically to punish Hot Shot and Hugh Jelly­­­-esque entries. I don't know if this by itself really addresses (3), though, because strong authors can reliably keep all their entries above that threshold, hence the suggestion to limit artists to one score per event should the risk of an entry be reduced.

3573996
Mm, the fake accounts thing is a good counterargument. Nothing can really be done about those unless Roger wants to do email verification or something (but I, for instance, have four email accounts I use regularly). I think it's such an impossible thing to prevent that it shouldn't be taken into consideration.

3573843

(3) can be achieved by limiting an artist to receiving a score only for their best entry in a particular event

Why not do something like this: assign score for an author's top story on the normal scale, then assign bonus score for all the other stories by the same author based on a scale that pretends the median rank is last place. That way a cleanup performance like TD put on this time would be rewarded, but not quite as much as it actually was in the current system, and only above-average second fics would benefit their authors in the author scoring.

3573896

The changes of the good story providing me with a positive score is also higher because I've increased the entry pool and ensured that at least three stories will be in the negative ranks.

Wouldn't this affect everyone else in the contest, as well, and thus end up evening out?

3574048
This could work. It probably keeps things simpler to just apply a diminishing return on consecutive scores. For example, the second best entry is worth only half its base score, third best a quarter, etc. Your system would have different (and not very predictable) effects depending on what distribution is used.

3574081
You've still got an advantage against those who didn't participate in the particular event that you pulled the stunt in.

Also, this made me realise that simply limiting authors to one score per event is incompatible with negative scoring, because if you can guarantee one entry is positive, the extra bad entries won't give any penalty. It would be possible to limit an author to one positive entry, though, and keep the negative scores as they are.

3574081
Obviously, my hypothetical story is superior to theirs in this example. :V Though given your amount of sensemake in 3574048, I have likely overlooked something.

3573843
Which results do you feel give the results which most reflect what you want to give more points to?

Basically, I'd look at that, and try to figure out some rule for them. Clearly some competitions were much more competitive than others, and one would think that those should be the most heavily rewarded.

Honestly, given that we've noted that the minific competitions result in people putting in multiple entries, couldn't you just do something simple like divide by that to get a more reasonable point total?

So for instance, if you have a minific competition, you get about 50% more entries than you do participants, so you could simply divide the result by 1.5 to get something closer to the actual "competitiveness" of it.

I dunno, a lot of this seems like it is very complicated when it could be made much more simple. You could just have a static multiplier for longfic competitions, or a static divisor for shortfic ones, to bring them more into balance.

Why do something very complicated which potentially gives weird, undesirable results at the edges when you could just be lazy and do something which is very simple to understand? The competitions seem to fall into two categories:

Minific competitions, 400-750 words.

Short story competitions, 2000-25,000 words

The only things which are outside of those bounds are really old competitions, so it seems like you could just add a static multiplier to the short story competitions and be done with it rather than trying to do some sort of elaborate formula which is dependent on everything else. You could increase the rewards on the short story competitions by, say, 33% or 50% and that would probably do what you're trying to do, which seems to be balancing out minific and short story competitions' point totals.

As for the rate of decay of scores... really, you could ask it that way, but the other way you could ask it is "how much less valuable should your score be per event?" .8 would be 20% decay; .9 would be 10% decay.

3573843
The main issue I see with the proposed system is that it's hard to follow for anyone without a Math-related degree. I believe people have a tendency to trust more systems that they can understand.

As for specific comments:

It was also apparent that some events were more difficult to place in than others, so scores should be weighted accordingly.

I don't see the need. Not when you use a curve that goes from first to last when distributing points to the participants; the number of points distributed in the event, thus, already scale with participation.

Or, in other words, while the difficulty in getting into a specific place increases with the number of participants, the number of places that award points increase in the same proportion, the end result being that the chance to get a given point award or higher is already close to a constant when the points awarded the winner don't change. After all, if I understood how your curves work, if the points for the winner were constant, second place when you have five participants would award as much as fourth place with thirteen, or 11th place with 41 participants.

You might need to increase the points to compensate for participation if you only awarded points to a fixed number of participants — say, if only those awarded medals got points — but that is not the case here.

Events are weighted not just proportional to the number of entries, but also to the word count of the entries.

I sincerely think you are overthinking this. I would simply go with fixed, but different, scores for normal fic events and for minific events; say, 50 points for the winner of a normal fic event, 25 for the winner of a minific event (or whichever you feel appropriate). The gain in making the scoring system easier to understand would, IMHO, bring more benefit than any gain in precision from going for a more complex system.

Specially if you intend the scoring system to be a participation incentive. Like I said above, easier to be motivated if you understand how the rewards are distributed, and not everyone has the grounding in math that we have.

and (4) can be achieved by applying a diminishing score to entries the older the event is.

I like this a lot. I personally would simply apply a fixed multiplier according to how old the event is, instead of multiplying the whole score by a fraction — say, from latest to earlier event, they would contribute 100%, 90%, 80%, 70%, 60%, 50%, 40%, 30%, 20%, 15%, 10%, and 5% of their points to the total score, with only the last year being counted — in order to make it easier to understand, but I love the concept.

3574021
Then in rounds without trolling going on it would punish writers that made a legitimate effort, which I think is a flaw of the system.

If you want to punish fics made to troll the competition my suggestion would be to allow voters to nominate fics for a punishment; a "troll" score, so to speak, that when calculating the hugbox (or the score for fics that don't reach a certain threshold of "troll" votes) would be counted as zero, but if a fic received enough of those it would cause the author to lose points and remove the fic from the scoring process.

3571372
I'd take issue with someone plotting a story out, having it on the shelf, then dusting it off for a write-off. But what I do, and what I have to think is pretty natural, is that I have all these one- or two-sentence premises floating around in my head, and a prompt may resonate with one of them. But the story still comes about as a part of the write-off. They're just a starting point and a resolution, not the details of how to get from one to the other.

As to all this talk about time limits and the purpose of these things...

Any time limit will affect someone. You have to draw the line somewhere, and one day is plenty of time to write something under 750 words. Yes, some of these stories probably would have done better with more time and word count. Save them up for the longer events. 3 days, and 2k-25k words. Another issue with time is that the longer these things drag on, the more people lose interest. A week writing time? People lose enthusiasm by the end, and discussion drops off. We used to have 2 weeks of judging, and most people just didn't care anymore by the end. We had to add an additional week to the voting period once because it was taking people too long to get through the entries, particularly SS&E's 24k behemoth. Longer time = more and wordier entries = fewer voters who could get through enough of them to count or a longer voting period. As with many established systems, it's the way it is for a reason.

And I, for one, hate to see anything the pushes the write-off further toward being a competition and away from being a writing workshop, obsession over scoring included.

3573843
Well, depending on how much you want to rejigger things, you could just do something like this:

First place: 3 points or total entries/4 round up, whichever is higher
Second place: 1 less than first place
Third place: 2 less than first place
Repeat until the score = 0, then you don't give any points for that place and any below.

No negative scores.

You could then (after assigning these scores) multiply these scores by 2 for short story competitions (the 2k+ word ones) so as to give more points for them as they have a smaller number of entries, without inflating the number of stories which are granted points. (So in the last minific competition, for instance, you'd give 6 points to 1st place, 4 points to 2nd place, and 3 points to third place as there were only 9 entries).

In this way the top quartile of stories, or the top 3 (whichever is greater) would get points and everyone else wouldn't.

Then you could just subtract 1 from everyone with a positive number's score after every competition to cause the scores to decay.

That'd be quite simple and easy to do and is pretty easy to understand, I think.

What about changing it from awarded points to percentage points. Doing this keeps the relation to inside the competition, so high numbers of participants doesn't negate previous challenges. It also discourages high volume of stories from a single author, but not so much that it will prevent those who do have multiple ideas they want to write about from submitting them. Plus, the final number shown, an overall average, would show where an author usually places. For instance:

A prompt had a total 4 users stories submitted by three authors. First place gets a 100, second a 75, third a 50, and fourth a 25. If author A wrote both the first and fourth place stories, his score would be 75, while author B is 75, and C is 50.

The next competition has 50 participants, each with only one story submitted. First place gets 100, second 98, third 96 and so on. This keeps the scores overall equalized, as first place here can not score higher than first place in a previous competition just with one story. But it also allows for those who placed higher in competitions with more participants, receive more points.

This eliminates the idea of submitting fics to boost your highest ones score, since if those fics place low, they will bring down your average. However, it isn't so devastating that you never want to submit multiple stories (like the negative number system currently is), it merely deters submissions that authors themselves believe are sub par.

3574530
Took a shower, realized a problem with this. Better still:

First place: total entries/4 rounded up, *10
Second place: -10 less than first
Third place: -20 less than first
Repeat until 0 points, then stop; any entry below that is worth nothing. No negative scores.

Your score from each competition drops by a static 10 percentage points every time, so a first place in a 9 entry competition would be worth 30, then 27, then 24, then 21, then 18, then 15, then 12, then 9, then 6, then 3, then 0.

This would:

A) Be relatively simple and easy to understand.
B) Keep all scores as whole numbers.
C) Cap scores to the last 10 competitions with easy-to-understand decay.

3574578
This would penalize people who submitted multiple stories, as they'd score less than if they only submitted what they believed to be their single best story, which I don't think is what we want at all.

3574602

Only if you truly believe your other story isn't very good. And for those who are not that interested in scores it will not affect them at all. The reason I like the percentage route is that you can glance at a number and determine how you compared to everyone else. For instance, in this comp my stories received 48th and 37th respectively. With the percentage rating, I know that my average story rated in the bottom 17.6 percentile. Now next month, I can compare how my story/stories rate and determine if I am improving. Can have each individual stories percentage rank listed when you click on overall, like the current system does for points. This way, you can compare your highest results.

Since everyone has been doing that, I guess I should talk a little about my entry.

The Last Words of Star Swirl the Bearded was my entry. I mentioned the prompt to my brother, and he told me about Einstein's last words having been told in German to a nurse that only spoke english. When trying to sleep a few hours later, the idea to write something related to that, but applied to Star Swirl came up. I knew that the whole thing was terribly predictable, so I tried to go for the driest tone I could, and then work a tonal shift right at the end.

However, I couldn't think of anything to make it credible that the words would ever be recorded, so Twilight became my TARDIS. I justified the perspective shift by thinking it would be like one of those documentaries that start talking about something in a pop-sci language, then add some personal story about how scientist X discovered that thing (think the new Cosmos). I didn't go the full route with this either, simply because I doubt anyone would ever talk about those things, and I was too sleepy to rework it. One editing pass later, I just submitted and forgot about it.

Quite frankly, 16th place for something I did half-asleep in one hour was more than I expected – and looks like I even managed to make some folks laugh! Hopefully I will have something better next time around.

3548444 mentioned that it could be imagined as an episode narrated in Hitchhikers or in a Pratchett footnote, and I really wish I went with this the whole way through. Would make the story a bit shorter, and would tie both halves together a little better.

3553837 In my defense, I was sleep-depraved.

3548624 is a cheeky bastard to say that a better story with the same premise was made. He is also correct.

3567803 In my mind, I was trying to imply that he didn't care for the attention given to his person rather than his work – hence the room full of ponies in his death bed. Certainly needed more work here.

3574638
The problem is that it would always hurt your score every single time versus if you submitted only your best. If your goal is to get people to submit stories, then you shouldn't penalize them for doing so.

Yours does. Always. An average will always be lower than the highest number in it, assuming that there are at least two numbers in the set and they aren't the same number.

3574652

3574674

Thats only if you know which story you wrote is the best one. I think everyone here can agree their stories did not get the reviews or perform exactly how they expected. Submitting multiple stories, which usually you put effort into, hedges your bet so that the higher performing one will bring up the lower performing one. But the reverse is also true as lower performing ones will indeed bring down your overall score. This does make you truly consider a story before submitting it. But that is only in a single competition, multiple stories can be very beneficial even though one may hurt another in the same prompt.

For instance, if I submit a story and get an 80 in the next prompt, my overall score is 38.4 over two prompts. If I had submitted 3 stories, one gets 80, one gets 55, and another gets 40, my overall score for that prompt drops from 80 to 58.3, but my overall score for all prompts goes up to 42.0. It could even be adjusted to only keep scores for your most recent 3-4 prompts, so more recent scores stay relevant.

For those who don't mind overall scores so much, they will be unhampered and submit all the 25 Famous Last Word finds and YOLO jokes they want.

3574703
I correctly guessed which two of my stories were better than than the other two stories.

I don't think people are going to deliberately submit awful stories for no reason, so I don't think penalizing people for submitting weaker stories is really beneficial. If someone DID do that, then we could deal with them socially; you don't need to use a scoring system for something like that, that's an extrinsic issue, like poor sportsmanship.

3574304
3574327

The main issue I see with the proposed system is that it's hard to follow for anyone without a Math-related degree. I believe people have a tendency to trust more systems that they can understand.

None of this goes beyond high-school level maths.

Besides, the exact intricacies of a system don't really need to be easily understood (and for anything sufficiently complex simply will not be understood) by people who only take a cursory analysis to it. Something that is fairly analogous to what I'm trying to do here is osu!'s pp system. The system itself is highly complex involving a lot of this kinds of maths, but users only need to understand that on the whole better performance means more points. Harder songs give more points, and the scores are weighted (in an identical f^n decay as I proposed) so that you can't just get okay scores on lots of songs for easy points.

I'd also like to point out that "complexity" is pretty hard to define. What I'm seeing from TitaniumDragon's suggestions is that simplicity means making sure everything is an integer. That's a pretty arbitrary line to draw and doesn't actually make things simpler, and in attempting to satisfy that constraint you forget what you were trying to do in the first place. For example, consider the curve described in 3574595. Putting it into a rigorous form (which is necessary if you want people to get a computer to do anything), this curve is:

max((ceiling(n/4) - x) * 10, 0)

The most obvious flaw with this curve is that everyone except the top quarter of entries get zero score, so it completely fails goal (2). In addition, saying that coming just below the top quarter is no better than coming last is simply wrong. (Aside from that, multiplying everything by 10 is entirely pointless.)

This is just the linear curve with a much harsher offset and a steep cut-off resulting in very poor granularity.

What I'm trying to get at is that the difficulty of some arbitrary person to understand something is not the same as complexity, and simple systems are not necessarily any better just because they're simple.

As far as the suggestion of using an arithmetic decay rather than a geometric one, what difference does that make? I'd need to store three columns of data rather than one: the base score, the actual score, and the current value of n. If I consider SilentStrider's suggested decay...

100%, 90%, 80%, 70%, 60%, 50%, 40%, 30%, 20%, 15%, 10%, and 5%

That's now four variables, so it's even more complex...

Basically, max(s - f^n, 0) is more complex than s * f^n. The impact of the arbitrary decay rate is also much more apparent, because it determines when something will have decayed entirely. Aside from that, I'd rather avoid zeroes on the whole, because if most things are zero then it's all equal, and that's just not very interesting.

3574578
This, again, is the linear curve. You're just spending a lot of English words describing something that in maths speak is one function.

What you've suggested on top of that is that the participant should get points based on the average performance on their entries. I don't think this is very appropriate because it pretty actively punishes submitting more than one entry, which I don't want to do.

(This is more or less my understanding of the numbers you gave. I'm not entirely sure, though, because you haven't explained it very rigorously.)

3574304

Honestly, given that we've noted that the minific competitions result in people putting in multiple entries, couldn't you just do something simple like divide by that to get a more reasonable point total?

The reason why minific contests should be worth less is not because people submit multiple entries. That's irrelevant. It's important to weigh the short story contests more heavily because, in line with goal (2), they take considerably more time investment. Writing a short story takes much more time than writing a minific. For example, I spent 4 hours writing my entry for this contest. For the short story contests, I usually spend somewhere between 20-30 hours on my entry.

I dunno, a lot of this seems like it is very complicated when it could be made much more simple. You could just have a static multiplier for longfic competitions, or a static divisor for shortfic ones, to bring them more into balance.

Remember what I said about relative complexity? This is another example. You're suggesting I hard code only two categories of event: minific or short story, and determine an arbitrary weight for them. It's actually a lot simpler to say the weight is directly proportional the sum of the square roots of the wordcounts, not just because that data already exists, but because it scales. If, for example, I put up some kind of novella competition with a minimum word count of 20,000, or a novel competition with a minimum word count of 60,000, then their weights will already be defined.

Please don't bother mentioning complexity (especially when defined as "how hard I think this is for some arbitrary person to understand") when considering options. (Occam's razor only applies when all other things are equal.)

I have more thoughts to address, but I needed to say this before you all start following tangents that lead nowhere.

3574781

In order for it to be a penalty, there must first be a reward for avoiding it. The current systm, and the one you propose, have scores for the sake of scores. They don't give relevant information from prompt to prompt and are solely there to have a leader board (of scores, not actual relatable skill). Having a percentage based system allows the author to determine at a glance his relatable skill in an individual prompt, and a general idea if he is improving over multiple prompts. This gives the system a meaning beyond arbitrary relative scoring.

3574787
Its not so much as "get points" as it is what percentile of stories was I rated greater than. It turns the point system into basically a report card.

The individual prompt equation is just the average of your relative percentage position. So (X1/y*100 +x2/y*100...+xn/Y*100)/n where x=position from last +1, y= total number of entries, and n= entries submitted by single author. This can be written as a single summation function, but I doubt I could find the symbol on my computer, much less a kindle fire.

The total average would include the individual stories from all prompts, not just an average of the prompt averages. This allows authors who submitted multiple stories that achieve more points than their averages, increase their overall score more.

So how does it hold up to your 4 goals?

(1) Rank participants in terms of performance across all events

This works very well, since it basically tells you that, "Across all events, I have performed, on average, better than x% of participants."

(2) Encourage participation

Not sure what you mean with this, so I'll give two answers. The first deals with bringing new people in.

Doesn't help much in this regard, other than those wishing to try and increase their scores in order to show off. But I don't think the scoring system should be worried about this point. Thats what the medals, prizes, and enjoyment of competition and discussion is there to do. This system is geared more towards giving authors information on how well they did and how much they have improved.

If this means does this encourage people to submit multiple entries then this does that fairly well, as I noted with Titanium Dragon. It does cause your score in an individual competition to decline, but overall score can just as readily increase. And a single bad entry is less likely to affect your overall score. This adds some risk but also rewards those who choose to submit stories they truly believe are worthy of competing. And many won't care much for scores so will include silly entries anyways.

(3) ...without encouraging spammed entries

As noted above, submitting multiple stories does not guarantee an increase in overall score, so it creates incentive to weigh the risk and rewards.

(4) Allow new participants to catch up to old ones

Since the highest achievable score is 100, and this can be achieved in a single prompt, I think this one is easily achieved.

3574127

The most obvious flaw with this curve is that everyone except the top quarter of entries get zero score, so it completely fails goal (2). In addition, saying that coming just below the top quarter is no better than coming last is simply wrong. (Aside from that, multiplying everything by 10 is entirely pointless.)

The reason I suggested using nothing but integers is because everything appears as integers on the website; multiplying by 10 isn't pointless if your goal is to keep everything as an integer, and integers are really easy to work with for people (and also lets you possibly save on memory, though I understand these days that people aren't as concerned with that).

This is just the linear curve with a much harsher offset and a steep cut-off resulting in very poor granularity.

All other things being equal, the simpler a system is, the better it is.

You're right about it being a higher cutoff value, though right now you cut things off at 50%; you could cut them off never, or use any other number of ways of going about assigning scores.

Basically, max(s - f^n, 0) is more complex than s * f^n. The impact of the arbitrary decay rate is also much more apparent, because it determines when something will have decayed entirely. Aside from that, I'd rather avoid zeroes on the whole, because if most things are zero then it's all equal, and that's just not very interesting.

The latter is much more difficult to grok than the former, though. I can easily do the former calculation in my head, whereas the latter would require me to use a calculator. It is much easier to know that last competition was worth only 90%, the one before 80%, ect. than 90%, 81%, 73%, 66%, 59%, ect.

The easier it is to understand a scoring system, the more useful it is to the user because it is easier for them to know what they need to do to "win".

The reason why minific contests should be worth less is not because people submit multiple entries. That's irrelevant. It's important to weigh the short story contests more heavily because, in line with goal (2), they take considerably more time investment. Writing a short story takes much more time than writing a minific. For example, I spent 4 hours writing my entry for this contest. For the short story contests, I usually spend somewhere between 20-30 hours on my entry.

That's a long time. Admittedly my own experience with this is pretty variable; some stories take me longer to write than others.

Remember what I said about relative complexity? This is another example. You're suggesting I hard code only two categories of event: minific or short story, and determine an arbitrary weight for them. It's actually a lot simpler to say the weight is directly proportional the sum of the square roots of the wordcounts, not just because that data already exists, but because it scales. If, for example, I put up some kind of novella competition with a minimum word count of 20,000, or a novel competition with a minimum word count of 60,000, then their weights will already be defined.

Well, first off, are we likely to do this?

Secondly, I'm not sure that any system is going to deal reasonably with scaling between 750 word minifics and 40,000 word novels.

Please don't bother mentioning complexity (especially when defined as "how hard I think this is for some arbitrary person to understand") when considering options. (Occam's razor only applies when all other things are equal.)

If you want people to care about their scores, then the system should be easily understood and it shouldn't be hard for someone to figure out how hard it would be to be #1.

If people don't really understand the scoring system, they're much less likely to care about it.

3574806
Competing against 37 people and competing against 8 people is very different, if the 37 people are just as good as the 8 people were on average. That's why they award more for doing well in bigger competitions - the purpose is to be a "skill meter" of sorts, I'm pretty sure.

Anyway, to sum up the goals of the scoreboard:
(1) Rank participants in terms of performance across all events
(2) Encourage participation
(3) ...without encouraging spammed entries
(4) Allow new participants to catch up to old ones

Honestly I think the real issue here is that no system can actually really do all four of these well; any system is going to have to trade off between #1, #2, and #4, because their goals are somewhat at odds. ELO is relatively good at #1 and gets more and more accurate with #3, but is bad at #2 because your score can go down (and will, when you lose), and a non-zero sum ELO fails #4. You're talking about awarding short stories more than minifics but that assumes that short story competitions are better at determining relative skill than minifics, which may or may not be the case (I have no idea).

It sounds like from what you're talking about that you consider #2 and #4 to be more important than #1, which I think is perfectly reasonable, and I think any sort of non-zero sum scoring system which has decay over time, as you noted, will work reasonably well for such. It will peripherally measure #1 but isn't likely to be especially accurate, but I'm not sure if that's really a big deal.

3573843

(1) Rank participants in terms of performance across all events
(2) Encourage participation
(3) ...without encouraging spammed entries
(4) Allow new participants to catch up to old ones

I'd like to suggest that this conversation focus (as Roger was working toward) on consensus over the goals of the scoring system. Once the goals are well defined, turning the goals into an equation is the easy part, and I would trust Roger to do it.

I do want the score to measure something useful, and provide both instant egoboo and a long-term incentive for people to do their best (see: Goal 2), but like 3574414 says, the more time that goes into it, the more it distracts from the cool things here.

For instance, if Goal 2 (encourage participation) is agreed upon by everyone, then the scoring should be changed so that participating in a given competition always leaves you strictly better off than not participating in that competition. Easy; done. If the final formula meets that criterion (and everything else we ask of it), I don't think 90 percent of us will care whether it's linear, quadratic, n^2*(O_log_n)/(phase_of_moon), or whatever.

The main debate on this page seems to be over how to handle multiple entries. Should the goal be to:
3-x. Encourage them unconditionally: every entry adds points. (this violates Goal 3 as stated, but maybe it's on the table);
3-a. Strongly encourage them: have each entry past the first boost an author's score, but they provide diminishing returns (or a small flat boost) relative to a contestant's first entry;
3-b. Weakly encourage them: You score based on the best of your entries, and all others are zeroed; multiple strong entries merely increase the chance of one of them rising to the top;
3-c. Encourage them conditionally: "good" entries past the first boost an author's total score, and "poor" entries past the first can make you perform worse than if you had not submitted them;
3-d. Discourage them: Multiple entries either subtract from, or are averaged with, your top score, so that you will always be worse off submitting multiple entries than if you had submitted only your best work;
3-e. Disallow them.

I feel multiple entries (within limits of sanity, and audience judging capability) should be encouraged, which would be A. I thought it was really cool that Pasco wrote two amazing entries for "Behind the Mask", and have to admire TD for landing three in this month's top 10; I don't always score PP's more experimental fics highly, but I appreciate the breadth which they bring to the competition, and I've learned more about writing (and about obscure crossovers) from them than I have from more traditional stories; and I've entered multiple fics of my own in the last two minific comps, because at 750 words I feel like I can push myself to try something crazy, and that's what will make me grow.

I feel that "one well-received fic and one poorly-received fic" is evidence of an author testing their boundaries, and that should be encouraged at least as much as someone sticking to a single "safe" story which they think people will like.

The other debate seems to be over relative scoring of minific and short story comps. What is the goal? Do we want to reward people proportionally to the competition overcome; and/or proportionally to time spent? Do we want to make certain that all competitions are valued equally, or that they are valued fairly, or what? I don't have a simple pitch there, but let's set our foundations before building the equations.

3575096
Maybe we should make a new thread about this, or use the Scoring thread?

I do agree that if #2 is our goal, really how it works specifically isn't a big deal so long as it, you know, encourages people to participate.

GOLD STARS FOR EVERYONE

I personally lean towards 3 or 3A as regards multiple entries; I don't think anyone is putting in multiple entries just to spite people, and I agree with Horizon's rationale re: experimentation. I also doubt that anyone is going to go TOO crazy with it.

As far as minifics vs short stories goes, I think you probably don't want to skew the ratio out too far in relative valuation one way or the other; anything more than a 2:1 ratio would probably be too far, and I think a 1:1 ratio (that is to say, a short story contest is worth as much on average as a minific contest) would be preferable.

3575019

Competing against 37 people and competing against 8 people is very different, if the 37 people are just as good as the 8 people were on average. That's why they award more for doing well in bigger competitions - the purpose is to be a "skill meter" of sorts, I'm pretty sure.

Then lest see how the results mirror this. For the competition with 8 people, the top four places would receive:

1st: 100 points
2nd: 87.5
3rd: 75
4th: 62.5

For the competition with 37 people, the top four places receive:

1st: 100 points
2nd: 97.3
3rd: 94.6
4th: 91.9

So although first place is capped, as you can only be better than 100% of the people who participated, the rest is automatically weighted by how many competed. Therefor placing higher in a competition with more people gives you a higher score.