Suggestion for Avon FanLit
Posted by CM under Random on Thu 19 Apr 2007
kay. Some of you have heard my theories about how to game Avon FanLit before. Some of you even saw them being tested, when we first posited them. But here’s the truth:
As constituted in the game we played, the rankings in Avon FanLit were susceptible to gaming. I’m not talking about cheats or zero-bandits or skippers or any of that. I’m just talking about regular old gaming that resulted from the game mechanics. A lot of what I’m going to say here is conjecture, but it’s conjecture that’s supported by observation, evidence, and FanLit’s own comments.
Things we know. First, your score is determined by two things: the votes you get, and the number of eyeballs you can grab from the crowd.
Important conjecture number one: The database stores one score for each voter. It either store a number (e.g. 4.5) or a skip. If a person skips a story, but then has it come up in the pool again and reads it, their previous skip is replaced with the score they give. Holly verified that, although it’s possible to skip a story more than once, it would only record one skip per person. I conjecture that it stores only one score–otherwise a person could skip a number of stories and then give them zeroes, which gives them double-ding power. Believe it or not, this hidden mechanic is vastly important.
Important conjecture number two: The database tries to get every entry to have a certain “minimum” number of scores (inclusive of skips). That minimum number is determined, to some extent, by the total number of votes relative to the number of stories. It won’t be even, but let’s say that any story should have at least half the average number of votes available in the pool.
Important conjecture number three: The number of reads you get is inclusive of invites.
Important conjecture number four: Not all positions in rotation are equal. Your position in the rotation is determined by two things: (1) whether you need more scores so as to fit criteria 2; and (2) your last N scores, where N is probably 3 or 4. If your last three or four scores are not very good, even if you are technically “in” rotation, you’ll probably be near the bottom of the pack. If your last N scores are decent, you’ll be near the top of the rotation.
This second bit is partially supported by claims by FanLit, who claimed that if your story isn’t getting skipped, you’ll fall out of rotation faster, but it’s more supported by observation. There’s a reason why getting a 5 from an invite was more likely to be obliterated by a low-score within minutes after you’d been sitting on a 2.0 for three hours: The 5 pushed you towards the top of rotation, and so you were more likely to be pushed in front of eyeballs, and thus get a bad score.
In Rounds 3 and 4, I tracked score timing and data, and while I didn’t have enough information to say definitively, it seems to me that this was a very plausible theory.
Okay. That’s the sum of my guesses about the inner workings of FanLit. Now you add in some observations about the FanLit reader’s habits, also garnered by paying a great deal of attention.
People were less likely to skip stories when the pool was thin (few entries available). People were less likely to skip stories late at night. They were less likely to skip stories that were presented to them when the database was serving up batches of 10 or so near the end of the voting period.
People were more likely to skip stories in that first rush on Monday.
So what do you do about this? Easy: you use your invites to game the system so that you’re only getting read when the pool is favorable. You submit your story on Saturday afternoon. You don’t use any of your invites. It’ll get reads up through Sunday mid-morning, where (chances are) you’ll get a few bad scores in a row. You lay low all through Sunday. Sunday night, you send out three invites to people you know respond quickly; you’re tossed back into the voting pool, and you pick up more high scores. At this point, you’re getting more scores from non-invites, because the database will push you to the top of the rotation to try and even you out. You should have something like 30 votes at this point, and 22 invites in your pocket.
Then you blast your 22 invites all at once. Boom. You’ll have 50-something votes; way more than anyone else, and you’ll be so far out of priority on Monday, compared to the stories that were submitted, that you won’t surface until the database starts batching at the end, when people are unlikely to skip.
That’s how you use your invites to minimize skips. Is this indicative of anything other than an ability to game the system? No. Will this help if your story isn’t any good? No. Must you do this to final? No. But it probably helps. And it shows that the main problem with the Avon FanLit calculus is that skips are weighted far, far too heavily. They’re the easiest thing in the world to game, and so their effect needs to be minimized.









April 19th, 2007 at 7:02 pm
I noticed that the entries I submitted Friday night always did the best. Timing had a lot to do with it.
Alice
April 19th, 2007 at 9:19 pm
I have to say, however, that I have a sneaking suspicion that “invited” votes are weighted lower than ones you pick up in the pool. Just a guess, but it would make sense. Of course people you invite to read your story will, on average, give you higher scores than those who encounter you in the pool. It just makes sense.
So I am not completely certain that saving up your invites as you suggest would make a quantifiable enough difference to raise an entry more than (at most) a couple of positions in the rankings.
I’m still completely bamboozled by the final ranking of my Fractured Fairy Tale (the Shrek cum X-Files parody) entry. It had more hits and a higher average than several entries that wound up in the top 20 in that round, yet it was ranked 284, IIRC. I still haven’t figured out how that computes, but it also makes me think there’s more to it than just math!
April 19th, 2007 at 9:31 pm
This is fascinating, CM, and I can totally see how it would work. Did you actually use this invite strategy for your winning entry? Because it had the highest rating of the entire competition didn’t it? Not that avg rating is necessarily relevent to your point, since it can be influenced by factors other than skips.
As an aside, my computer’s firewall hates your blog. I have to change all sorts of setting to be able to post comments, and since I’m usually too lazy, I just convince myself that my comment isn’t that insightful anyway. But know that I’m reading!;)
April 19th, 2007 at 10:02 pm
I think we asked about invites at some point. The system doesn’t know whether a score is an invite or not.
Did I “use” this strategy for the winning entry? For the most part, yeah. But it shouldn’t have affected the final score because they reset the scores for voting. IIRC, Tiffany’s entry that round had a higher average than mine going in.
I have no idea why that entry was so popular, by the way. It wasn’t the best entry I wrote. It wasn’t even my best entry that round.
April 19th, 2007 at 10:33 pm
I think your powers of observation are keen. Have no idea if that actually works. But I tend to believe it might.
I did not have a system, other than to try and use my invitations . Of course, I should point out, I never won. But I did final twice. In round one I used (I think, don’t remember the exact number) about half my invitations and finaled. I know another finalist in round one who told me she only used a handful her invites.
The last round I used all my invites and tried to be smarter about who I gave them too. I placed in the mid-thirties or something, low compared to most of my other entries.
I agree the skips are weighted too heavily. In fact, I don’t think the skips should be counted and I think outlying low scores should be eliminated. Mostly because those zeros hurt my feelings!
Perhaps zeros already are eliminated. I can tell you T an Mud recieved a truckload of zeros from the “no fair writing a prologue” contingent. So I was shocked when it finaled. I would have been shocked in any event, but I was extra shocked.
I know you’re a big fan of the invitation system. And I’m not necessarily against it. One problem I saw towards the end is that when the invites started going mostly to other fanlitters (our families were sick to death of it by then), people began to read entries only
by invitation or as strategy.
Okay, I just went back and checked myfanlit. In round one, I sent 16 of my 25 invites and 10 people responded. Obviously, not brilliant strategy on my part.
April 19th, 2007 at 10:51 pm
Ha. Round one, I used two invites. Two, total.
Round 2, 12/17 were used.
I didn’t start playing the invite game proper until Round 3, when I got 24/25.
There after, I used all my invites.
Just for fun, I used the invite-game strategy in Round 6 with my paranormal-not-serious entry, “Get thee belime me!” Which ended up at a spooky #13.
Which tells you how scarily successful this thing is.
April 19th, 2007 at 11:01 pm
Out of curiosity, do you think they elminate zeros? I think they do.
April 19th, 2007 at 11:02 pm
Do I think they eliminate zeroes? Nope. Unlikely.
April 20th, 2007 at 8:44 am
Ah yes, I forgot about final score resets. I got so into this discussion that I had to go back and look at the site, but it’s hard to figure out anything conclusive from page views, etc., since there’s real way to compare finalists & non finalists.
I totally buy that you can work the system this way, but as you say, it only takes you so far. Except Round 1 (which was a whole other ballgame anyway), every one of my entries was submitted within three hours of the deadline. Two of those made the finals, the other two didn’t. Though I suppose if no one was trying to work the system, it doesn’t matter that I’m an exception.
And I liked your other Chapter 5 entry better too - though who can explain the tastes of the reading public?
April 20th, 2007 at 10:50 am
CM — what is your “take” on the being able to vote twice on a submission. . . I found that I could vote in the pool and then when I got an invitation, I could “change” my vote.
I suspect that only one vote counted and my suspicion is that it’s the first “pool” vote. . .but that’s the pessimist in me.
April 20th, 2007 at 11:23 am
Only one vote counts, and the latter one counts. Later votes always overwrite earlier ones.