Originally Posted by Varsam
I'd say that the Striker is overwhelmingly popular as the go-to lock-on for the TR - the TR ES lock-on uniques are a fraction of their VS and NC counterparts.
The closely grouped quartile KPUs is indicative of the Striker's oversimplicity - its design makes it simultaneously easy to use for all skill levels while leaving no ceiling for more skilled players to excel. The slightly lower ABR (which I'm guessing means average BR of uniques) compared to the other two ES launchers corroborates this.
|
edit: sorry, totally broke that quote up there.
Yes, but it's an extremely (even ludicrously) popular weapon with a lower KPU than a less popular counterpart, was the point I was making.
Originally Posted by Varsam
I'm sure that weapon effectiveness does indeed have an effect on KPU, but there are also so many other variables affecting it as well (both known to us and unknown) that attempting to infer meaning in KPU trends in anything but very broad contexts means you also get myriad outliers, exceptions, and contradictions which hurts the very argument you were trying to make. There are also major factors that contribute to weapon effectiveness that we simply don't have metrics for, like area denial.
|
I agree completely. I also think that, like in any large sample set, many of the outliers, exceptions, and contradictions tend to cancel out, especially when looking at things in the same item class over long periods of time.
I get that there are lots of things we don't know here, and I also get that there are a bunch of unmeasureables. You make the best effort on a model with the best data you get, and you try to make it better with every revision. I said it early on - everyone is welcome to take away whatever they want from this project. I haven't seen anyone use that license to fly off the handle and ask for ridiculous things. I think there's a healthy dose caution here. And pardon, because I'm using this post to segue right into a data set -
Original here:
http://bit.ly/16LFGEk