Tailor Archive
Thread: FS tailor, skill tapes, and socketing: information needed.
RandDarkstar
Thu Oct 07, 2004 11:42 am
#14
???? What's a gausian? Where do they build these curves of theirs? I'm getting a bunch of people together and we are going to group tef the heck out of them.
Akaara
Thu Oct 07, 2004 3:32 pm
#15
Gyopi wrote:
Anyway, I used the wrong term. Its really a multi-nomial distribution, which is almost the same thing. Here is some more information. Don't let any children look at it because it is kind of scary.
Gaussian
Binomial
Who would have known that all that math I had to take could be used for video games?
Message Edited by Gyopi on 10-07-2004 04:06 PM
The sad part is that my child can actually understand it better than myself 
RandDarkstar
Thu Oct 07, 2004 3:48 pm
#16
I looked at both pages and thought to myself - "Did I go to school??"
ASrai
Fri Oct 08, 2004 8:21 am
#17
Gyopi wrote:
Ok. I am was looking at the data of the sockets from the clothes I made last night and it looks to me like there is a problem in how they calculate the number of sockets. If you put the results into a histogram you can see what is going on. In the case of not wearing my outfit with the +20 bonus you can see that it makes a nice Gaussian (bell-shaped) curve with an mean (average) of about 2.5 sockets. The high end of this curve is slightly cut off, but it is basically OK. Now we look at the results when I am wearing the suit with +20 assembly. If we put this into a histogram, we no longer have that nice bell curve. If we look more closely, we can see that this is still really a Gaussian curve, but half of it has been chopped off. If we suppose that there could be up to 8 sockets and fill in the other half of the curve symmetrically we suddenly find that we have an average of slightly less than 4 sockets, which is what we would all expect. The way that the game seems to be calculating sockets is that it finds a number of sockets randomly on the curve. If that number is greater than 4, it re-rolls until it gets a number of 4 or less. As a result, we get a lot more pieces of clothing with 3 or 4 sockets, but the average number of sockets hardly changes with the assembly bonus. I think that the way it *should* work (which is how I think it used to work) is that if the number of sockets is greater than 4, it should simply round down to four and no re-roll. I hope this makes sense.
Now my brain hurts....Message Edited by Gyopi on 10-07-2004 01:49 PM
/glazed eyes
/equip dunce cap +25
But seriously I think your analysis does make sense. Good Work!
Akaara
Sat Oct 09, 2004 5:57 am
#18
NJ
This seems really messed up to me. I was not getting any consistancy in the amount of sockets.
I was restocking my vendors... I made several muscle shirts and flex form shirts. I started out getting 2 - 4socketed shirts, then I got ZERO sockets, then 1 socket, then 4. I have +111 to clothing assembly (including fs crafting assembly). I had more crit fails than ever before. I was like... wth?
I made shirts, pants, flightsuits and I did get more 4 socketed items... but I also was getting more zero to 1 socketed items. Not many in the 2-3 socket range.. like it was either 4 or 1. I couldn't find any pattern but it was the first time (in a long time) I ever crafted an item with NO sockets whatsoever.
I did better when I only had +106 to clothing assembly. However in my experience with being a tailor... I do notice that some days when you craft you have more crit fails than on other days... I used to stop crafting when I would crit fail alot. I do think something is broken.
NJ62
Sat Oct 09, 2004 6:28 am
#19
I notice that I only get 4 sockets when I am making displays, not for sale. We'll keep looking into this, as it's a huge issue.
If it's so big, though, why didn't it make the top 5, and where did factional uniforms come from?!
We'll keep looking into this, as it's a huge issue.If it's so big, though, why didn't it make the top 5, and where did factional uniforms come from?!
Akaara
Sat Oct 09, 2004 9:36 am
#20
Even with Gyopi's tests you can see he clearly got more zero sockets than when he had no assembly bonus at all.
He did get more 4 socketed items but with clothing assembly this high, I wouldn't think you'd get zero sockets at all.
nurTsunami
Sun Oct 10, 2004 1:56 am
#21
Gyopi wrote:
RandDarkstar wrote:
???? What's a gausian? Where do they build these curves of theirs? I'm getting a bunch of people together and we are going to group tef the heck out of them.
As well you should! Those Gaussians are nothing but trouble and should be nerfed! I just hope they aren't the next player race, thats all.
Anyway, I used the wrong term. Its really a multi-nomial distribution, which is almost the same thing. Here is some more information. Don't let any children look at it because it is kind of scary.
Gaussian
Binomial
Who would have known that all that math I had to take could be used for video games?
Message Edited by Gyopi on 10-07-2004 04:06 PM
Wow! High level math! **drools** I wish I could remember some of that stuff again, its been awhile. Cool stuff, using statistics for video game components. And they said you'd never use it again....
Gyopi
Sun Oct 24, 2004 2:19 pm
#22
Ok. This is kind of strange. I think now that assembly skilltapes aren't broken, but that now item complexity is affecting the number of sockets. I am getting much worse results than I used to for complex items like tactical skinsuits, but much better for simple things like ribbed shirts. Here is what I get for ribbed shirts.
Ribbed Shirts (no bonus) -- 3 critical failures
-------------------------------
1331313323
1241342432
3024313304
1231323243
3342122132
Sockets:
0 -- 2
1 -- 10
2 -- 12
3 -- 19
4 -- 7
Ribbed Shirts (+20 Assembly) -- 2 critical failures
--------------------------------------------------------------
3332233444
2443433414
4334404314
3144224234
4322124444
Sockets:
0 -- 2
1 -- 3
2 -- 9
3 -- 14
4 -- 22
TechnoCan
Sun Oct 24, 2004 2:40 pm
#23
Sounds to me, that the complexity-fix in publish 11 got some things right as they should have been all the time.
Gyopi wrote:
Ok. This is kind of strange. I think now that assembly skilltapes aren't broken, but that now item complexity is affecting the number of sockets. I am getting much worse results than I used to for complex items like tactical skinsuits, but much better for simple things like ribbed shirts. Here is what I get for ribbed shirts.
Gyopi
Sun Oct 24, 2004 3:28 pm
#24
TechnoCan wrote:
Sounds to me, that the complexity-fix in publish 11 got some things right as they should have been all the time.
I can't complain about it, I just didn't expect it and kind of panicked. I wish they had better patch notes!
TechnoCan
Sun Oct 24, 2004 4:07 pm
#25
Gyopi wrote:
TechnoCan wrote:
Sounds to me, that the complexity-fix in publish 11 got some things right as they should have been all the time.
I can't complain about it, I just didn't expect it and kind of panicked. I wish they had better patch notes!
Hehe, yeah the patchnotes are kind of confusing at times like this... with all that caffein in their veins and the lack of sleep and the high hopes, that the light they see at the end of the tunnel isn't the "Austin Cephyr Express" that is 10 minutes early on schedule our devs still work miracles... even such small and hidden miracles like this one 
I wonder if the socket-ratio would drop again if we'd use CA's that lower the complexity for clothing crafting...
NJ62
Sun Oct 24, 2004 4:21 pm
#26
I'll start poking around about this soon. Until JTL ships, I'm sure nobody will have time to deal with this. I hate delaying for an "opportune moment" but I think I have to.