Bayeth Mal
OSG Planetary Operations
Covert Intervention
396
   |
Posted - 2014.05.24 07:40:00 -
[1] - Quote
This is a post from a thread on tweaking rifle numbers.
https://forums.dust514.com/default.aspx?g=posts&m=2175888#post2175888
Most of these suggestion threads start off OK but then degrade into random crap and don't want the post to disappear in the noise.
It's also a divergent subject from the original threads talking points so thought it belonged in its own thread.
I'm happy to let Devs decide if it should stay that way.
Begin Post:
CCP Rattati wrote:This will be interesting. First I would like to get relative parity or usage between the racial variants, within each class (std, adv, pro) and leave the gallente variants and other race full auto variants out of the picture (for a time being) That said, currently these rifles are in relative balance, sorted from most consumed to least.
- Rail Rifles
- Combat Rifles
- Plasma (GA) Rifles
- Scrambler Rifles
Testing this as anecdotally I disagree with this ranking. Currently performing survey.
Edit: Have the data but it's 1am and need to go to bed. Will do basic analysis and post the results in the morning (in about 8 hours)
Ok so I have the data now. Sat and watched the kill-feed for 4 games. A small sample size I know but it's boring to do and the clear winner made itself apparent quick.
While RR may be the most consumed weapon, we don't play with our favourite weapon so we can die holding it.
So I recorded what rifles were getting the most kills.
Also I noticed Laser rifle isn't even being discussed any-more like it doesn't exist. Weren't they supposed to get some love this build? HA!
RR: 29.4%
CR: 61.8%
AR: 8.8%
ScR: 0%
Laz: 0%
RR: 19.1%
CR: 47.6%
AR: 14.3%
ScR: 9.5%
Laz: 9.5%
RR: 16.3%
CR: 55.8%
AR: 9.3%
ScR: 18.6%
Laz: 0%
RR: 28.9%
CR: 51.1%
AR: 13.3%
ScR: 6.7%
Laz: 0
This is just a percentage of kills with rifles, not percentages of all weapons totals. Also didn't count militia ARs as that would over represent the plasma rifles due to being paired with starter fits.
Random things I noticed while collecting the data (anecdotally, I didn't record these numbers):
- The assault CR variant is the most popular, with it being roughly a 70/30 split between assault/standard variants.
- The Advanced and Prototype CR's were very popular with just the standard CR being one of the rarer weapons I saw in the kill feed.
- The reverse was true with the RR. It was the lowest level "Rail Rifle" the vast majority of the time, when above standard it was again a 70/30 split assault/standard. High level RR were so rare I only saw "Kaalakiota Rail Rifle" twice from the same guy before, I assume, he got killed and came back with a lower level weapon.
- The Duvolle Assault Rifle did not make a single appearance in the kill feeds. Gek made up almost all Plasma Rifle kills.
- When Laser and Scrambler Rifles made an appearance it was only at Prototype. Scrambler was around 50/50 split standard/assault
Other random things apart from this data I noticed:
- Aurum gear is a lot more popular than I realised, but again, it tended to be the high level CR getting kills.
- Installations kill a lot more random players than you would think.
- The standard level "Shotgun" is far and away the most popular shotgun variant/level and was used and successfully scored kills regardless of the amount of proto-stomping I could see going on. Adv and Proto shotguns were as rare as laser based weapons.
- The HMG seemed to be pretty well spread over levels, with it being sorta 50/30/20 split between std, adv and proto. That's what we want isn't it?
Ultimately I'm concerned that you're looking at your data the wrong way and may have an error going into it.
The number of a given weapon consumed is irrelevant.
It's the number of people getting killed by a specific weapon.
In a game like Battlefield or CoD where you can easily switch to the OP weapon looking at the number lost would be a good indicator as everyone would be running whatever was OP.
But we all sunk our SP back in 1.7. And in 1.7 RR was easily the most popular because it was the most OP.
But now the tides have turned, and all those players who specced RR are stuck with them.
Let's say tomorrow you changed the prototype level massdriver to fire Core Locus Nades and tripled the size of the clip.
Mass driver usage would go up, but not drastically because not everyone has proto massdrivers.
And if you're looking only at items lost they would still be under represented because they are the ones doing the killing.
Addendum: I'll note these possible biases to ward off complaints.
Time of day - It may have been the proto stomping hours on the server I was on, but still proto levels of other weapons were not so apparent. So the CR is the preferred weapon of proto stompers.
Server - See above, also maybe a cultural bias, maybe each server is different (do French players RP Gallente? Japanese RP Caldari?)
Players - when going into a match I checked the leader board to make sure I wasn't seeing the same players/squads/corps if I recognised more than 3 names on either side I left and tried a different match.
Game mode - 2 skirmish, 2 dom, maybe numbers are different in ambush (doubt it)
Small sample size - Yeah, I know. If you want a full study I could do a huge survey but you'll have to start paying me in ISK 
In the match where both ScR and Laser were present it was the same player who started Laser and switch to ScR so that shows the issue with my small sample, one player was able to be over represented.
Heading over to Destiny Beta and a few others
Hit me up for Skype and PSN
|
Bayeth Mal
OSG Planetary Operations
Covert Intervention
411
|
Posted - 2014.05.24 14:55:00 -
[2] - Quote
CCP Rattati wrote:There is such a direct correlation between sales, consumption and efficiency that it is almost irrelevant which one uses on a macro level to study behavior. Thanks for your concern though. 
I'm not so concerned what the data actually says, I use a specific type of rifle and play the game so of course this taints my bias (man that sounds like a euphemism) but I would actually like to see this data in more detail.
Specifically just straight stats for 1.7 and 1.8. With each split in half, the first half of the build and the latter half.
First half being people figuring out best builds and speccing into them, the latter half being that they are now specced and are using them.
This direct correlation you speak of, is this from something actually analysed within the game? what time period does this cover?
If not from the game is this in relation to more general theories of economy?
Mostly because, like I said, in other subject areas you don't so much get the lock in to a product that you get in Dust.
Just trying to make sure the model that's being used here is comparing apples to apples, and that the data that is being used takes into account dust's current state, not how it was a few months ago.
Heading over to Destiny Beta and a few others
Hit me up for Skype and PSN
|