SWTOR formula list

[Fraser]

(01-21-2012 08:54 AM)Kor Wrote:  

Quote:notice significantly less benefit from the next 10 points than we did from the previous 10

This doesn't really ever happen. The key is when we see significantly less benefit from the next 10 points than we did from the first 10. Case in point, for Accuracy at 10%, the last 10 points are 2/3rds as potent as the first 10.

If we approximated it as linear, we'd have to define from what linearization we're using. If we use the rating necessary to reach 1% (27.5), that gives us a 5% value of 137.5 (7.1% error) and a 10% value of 275 rating (16.4% error). If we use the rating necessary to reach 10% (329), that gives us a 5% value of 164.5 (11.1% error) and a 1% value of 32.9 (19.6% error). If we use the rating necessary to reach 5% (148), that gives us a 10% value of 296 (10.0% error) and a 1% value of 29.6 (7.6% error).

Those are pretty ridiculous error margins.

I don't really know how to respond to this without the words souring in my brain and sounding lame and defensive... But I'm going to give it my best shot anyways. I will be honest though, I didn't do the math upfront, so I really have no reason to complain, but those are much larger margins than I had expected over the region.

The region used is somewhat larger/different than what I would have used, but the data presented gives me little reason to doubt that taking the value of 8% for a region of 8% +/- 2% would be significantly more usable.

Just in case you wondered why I chose that range: 8% used as a base value because that is the current theorized Mob defense chance, and +/- 2% because its a large enough range for an approximation to be relevant while still being small enough to stand a chance at being accurate enough to be useful in a comparison against other stats. Granted, I have still done no upfront math on this so it's still all pretty meaningless. Hell, in hindsight it would probably be pretty meaningless to begin with because it wouldn't really simplify things enough to make an overall easier set of equations to work with, even if there were similar and 'equally accurate' assumptions for the other stats there are still just too many variables.

Man I've really let some verbal (typal?) spewage happen here, all based on a half formed and badly expressed idea about why accuracy has a soft cap and how the DR curve never enters the truly punishing portion relative to other stats because of it. I think I just let it happen again, I'll just shut up now and go sulk in the corner properly embarrassed for myself and my lack of ability to properly articulate what I mean.