#Ingame difficulty estimation

27 messages · Page 1 of 1 (latest)

latent ruin
#

I am under the assumption that every playthrough of a map has the position of the player recorded as it moves (#announcements message) and this feature idea relies on that heavily

Is your feature request related to a problem? Please describe.
the difficulty of maps is completely opinion based, resulting in difficulty boundaries changing over time, which is confusing. community ratings are susceptible to trolling

Describe the solution you'd like
My idea is to use the data the game has on the paths players take while they play to find an estimate of how strict the pacing and precision is in a map.

Estimation of pacing and precision would both work slightly differently:

Pacing Estimation
I am assuming that the data points of the player's position are taken at certain time intervals (?). The game could take a sample of four different playthroughs of a map and for every second into a map (or whatever the time interval is) and find the relative distance of each point away from each other, and then calculate an average. This can then be done for all position points in all four playthroughs, and then the distances can be averaged out to one value for the entire map. My idea is that with a larger the difference between different playthroughs, the pacing must therefore be more relaxed as there is a larger number of more diverse paths to take while still surviving the map.

Alternatively if the position samples taken from the player are at irregular time intervals, samples can be made by averaging out two other points. Not the most accurate but this is an estimation after all.

tria_difficulty_chart.png
#

Precision Estimation
This focuses on jumping specifically, or whenever you are in the freefall state i guess.

Whenever the player jumps in a run, the points representing the start and end of the jump should be added to a table. Since it is incredibly likely that mutiple playthroughs of the same map have a different number of total times jumped, jumps across different playthroughs can be sorted and classified as "the same" by finding the closest matching start and end points of a jump from both runs (anomalies are ignored). Then, like the pacing estimation, the game would then find the distance between the start, end and middle points of every jump of both playthroughs and find an average difference in position. With a larger difference, the map is 'less precise' and therefore probably easier.

Fixed boundaries between difficulties can be set, and these two estimated precision and pacing values can be used to define a map as within a certain difficulty. This is not an objective way of measuring difficulty but it is impartial and hopefully more accurate

Possible issues with this idea

  • Maps with split paths would not be estimated accurately. This could be solved with more logic and a larger sample size probably, but idk how that solution would work (also maps with variants are similarly affected)
  • Maps with procedurally generated gameplay (in the Future™) would be estimated as being far easier than they actually would be. Not a problem right now because no maps like that have been finished.
  • Exploiters can mess up the difficulty estimation of maps with low completion counts. Runs that stray incredibly far from the average could probably be ignored to fix this

Additional context
this is probably my most unhinged feature suggestion sooo im sorry if i wrote something that doesnt make any sense or if theres a flaw in my logic. Also the chart i attached is not meant to be taken seriously

silver moon
#

ah a fellow "person with ambitious idea makes a post"

#

funny how you mention the algorithm because I brought that up in my post too

#

and yeah, I don't REALLY know how the algorithm works but you seem to know a little more than me so if it is actually doable this could be actually be pretty nice

#

I'd still have a separate community rankings bar though

#

but if this works well this could be a good default

latent ruin
#

yes i think community ratings should stay

silver moon
#

and like, would this be the first kind of "definitive difficulty" in any kind of game like this?

#

like gd or etoh?

#

have they ever done anything similar or even tried to?

latent ruin
#

idk about etoh but in gd no

#

gd only has one input so i’m not sure why no one has tried

silver moon
#

this could really redefine the phrase in the game's description "Revolutionary systems unseen anywhere else on roblox"

#

or however it goes

latent ruin
silver moon
latent ruin
#

there’s ‘frame perfect’ counters which is the closest thing i’d imagine they already know about

#

but they ignore all other inputs so it’s not an objective measure

silver moon
#

especially taking into consideration nerves and all that

amber nacelle
#

uh i think it would be far simpler if map makers could put an estimated difficulty rating next to the community rating

silver moon
#

also like this would be a really cool system to see

latent ruin
#

map makers tend to find to find their maps easier than they are for other people

silver moon
latent ruin
#

you can do things to improve community ratings but my angle is that everyone has biases and giving an impartial opinion is very hard