I did this while drunk. I am still drunk.
I autistically went through and analyzed word count v commendations on storygames and found that the correlation is a 2nd degree polynomial
and that the best way to min-max is to write very little words and get the game commended
the two major outliers are both BZs games
and the top 3 outliers from the polynomial curve are BZ games
cows v aliens in particular is over-commended by like 90%
it has more comms than most 100k games yet only has 25k words. You might think the algorithm takes into account scripting but there is an abundance of evidence against that because a lot of other games have fucktons of scripting and do not deviate from their expected value in comms
I took the top 3 commendation holders from each order and recorded the word count and commendations obtained for all their commended games
"Well maybe theres more to the equation than meets the eye. It could be logarithmic or a power series"
wrong.
a linear regression doesnt fit the data
exponential doesnt fit the data
and a power series strays too far
a 2nd degree polynomial fits the data best
"But you graphed on a linear scale!"
shut up its the same shit on log log
linear on log-log
exponential on log-log
power series on log-log
you might think "oh but that power series fits pretty well!"
until you see the polynomial fit
and lets be real, when 3J made the equation he wasnt going to program a power series or any of that other bs. he clearly went for a simple 2 degree polynomial.
maybe 3 degree, but even then its more simple than a power series or dealing with exponential bs
Here I've gotten rid of games over 300k words and divided wordcount by 1000 to give easier numbers
we can now see that BZ's cows v aliens game, all the way on the right, with the most comms and roughly 26k words, is so far off it's hilarious
the least comms was ogre's game which was 80 words and all scripting which gained 28 comms
which just goes to show that shorter length games, as long as they're quality enough to commend, get the biggest gains for the least effort
the difference between 0 comms and 50 comms is about 10k words. the difference between 100 comms and 150 comms is on the order of 100k words.
so you could write an epic thats super long and maybe amazing and get 200 comms
but then someone else can write a tenth of what you did spread out over many small games and get over 200 comms
or you can be BZ and write a paltry amount and somehow still get 130 comms
dungeon stompage is the 2nd furthest from the average line with 74k words and 120 comms
a normal 74k word storygame can expect to get 80 to 90 comms
over the line = over the average amount of work, and less than average comms for it.
under the line = less than average amount of work, and more than average comms for it
the values towards the bottom of the graph may be small in comms but they're also very small in word count, so they're almost all below the line by a little bit
as word count goes up it becomes harder and harder to squeeze out more commendations
a 100k game would need an additional 70k on average to gain 20 more comms
a 10k game would only need an additional 20k for the same
the sweet spot for word count seems to be 20 - 70k, with 30k and 40k being about as ideal as possible
not only are the majority of commended games in this area for word count (ie minimum words to get commended is the name of the game) but they are also mostly below the polynomial line
the best way to min-max would be to write 5k and get 40 comms at a time
however most people cant write a qualitative 5k consistently
to ensure the greatest probability (no longer looking at the frequency graph, but the distribution of comms to word count) the best play is to land a game in the 20k - 60k range
most people who CAN write 30k-40k are the kind of people who can make those 40k qualitatively worthwhile to commend
the best players in this game are those who can write 5 - 10k and get commended consistently. The best an average person can do is churn out 20-50k and get commended consistently thanks to the length making up for some of the quality.
END OF RANT FOR NON-PROLIFIC AUTHORS
START OF RANT FOR PROLIFIC AUTHORS
The real question: word count v rating
it may behoove an autist like me to go through the games and graph rating versus word count
but that would be erroneous
because many of these games are old
and old games are rated higher than new ones by a fair margin. possibly +1 or even +2 to the average rating thanks to years of weak raters
it would be better to look at games made in the last 2 years, and of those, record ratings and see what readers like more.
obviously epics are going to be rated highly but the scale only goes up to 8 so it has to taper off at a particular word count
and that means theres a linear region or anchor point
and anchor point wherein there is a definitive "If you write more than this, your average rating will be 6+"
and less would be unpredictable
games like 16 words are rated very highly despite their word count due to excess quality
and gamey games are rated highly despite having no story, so the outliers will be towards the lower word counts and the average gets very wobbly in the beginning but stablizes later as word count increases
but since only people who actually write would care about this, they probably arent interested in low word count game creation
so our filter then becomes:
- published in the last 2 years
- greater than x words [some significant, but possibly arbitrary value for x]
- sufficient number of ratings [no brand new games because their ratings arent stable]
and put it on a graph and I bet you it would look like the black mesa logo reversed
DC = dont care
DNE = does not exist
the blue area is games with high word count that dont meet minimum site standards or are rated below 3/8 which basically doesnt exist
and anything with a word count less than x we dont care about because this is for authors not noobs
the high variation at the start is games with "low" word count but vary in quality a lot so the ratings are all over the place, but as word count increases that variation stabilizes and plateus
lets focus now on the part authors care about
they want at LEAST a 5/8 on their game
and they have the minimum word count to get a 5/8 just for effort (or close to it)
theres still variation because their word count is low so what they seek is a point where that variation in rating is below a threshold, in this example 10%
if we took the data of all the storygames made in the last 2 years and graphed them word count v rating we could point out where the variation is low enough (less than +-10%) we can confidently say that a certain amount of words is highly likely to get you a 6+ rating
and boom we've just solved motivation for nervous wreck writers
because we then give them a goal based on statistics that as long as they meet a certain word count with their idea, they're basically guaranteed to get a good response
END RANT FOR PROLIFIC WRITERS
I am still drunk. maybe format it better later, maybe not.
