that would probably never be accurate unless you've simulated hundred of thousands of sequences.
Recently I've been digging into the RS combat formulas. The best (public) attempt at reverse engineering the formulas was until recently available on the RS forums (apparently it is now removed, see: [Only registered and activated users can see links. ]). There is one archived copy on rune-server:
[Only registered and activated users can see links. ]
I haven't played RS in ages, so I am not sure how accurate these are. Let us focus on the accuracy formula.
In order to reverse engineer the accuracy formula more accurately, we need data.
There are some spreadsheets online of people recording for each attack whether they landed a hit or not.
I was thinking this could be done programmatically, perhaps by writing a RuneLite plugin.
The plugin would record attack sequences and upload the results to a database.
With a sufficient amount of data, we could compare the actual results to the results computed by our own accuracy formula.
And from there tweak the formula accordingly.
The process will take up quite some time and I am not sure it will be worth it but it does seem like a fun project.
I wonder what the people here think about it?
Accepting the premise people will use the plugin, it doesn't seem far fetched to me the collected data can be useful in improving the current formula's.
I don't see this taking thousands of hours. As stan said, if you make the plugin and have people use it while playing, it could accumulate data relatively quickly. Once you have the data collected its just some time, math, and effort.
Stan - I would like to help if you want help on this
Very excellent idea! The only way to truly reverse engineer it, is to record enough data to than find the appropriate formula, that returns the same outcomes as the data collected. Doing the combat wouldn't be hard, but you would want AT LEAST a thousand of as many creatures as you can muster. For me, the longest part would be writing the program/plugin to record such data. It would be handy if you could find a few different people with the same equipment, stats, etc.. etc.… Keep the variables minimal at first, than switch it up. Do like a case study type of deal, a controlled experiment. Once you have sufficient data with minimal variables, collect data with many variables to help fine tune the formula.
Just my opinion. Hope it might help shed some light somewhere. I also want to say, I do believe that sooner or later, we will get it right (a hundo percent), but it wont be for a long while. Thanks for letting me rant!
|« A Positive RSPS Future | RSPS Development Series »|
Users Browsing this Thread
There are currently 1 users browsing this thread. (0 members and 1 guests)