Doctor Archive

Thread: Official fact finding My buffs are coming out too low thread (Jan 18, 2004)

AnnnThrax
Wed Feb 18, 2004 11:41 am
#79

1.) Pre-problem normal buff average. - Pre-problem range mid 1200's to 3460. averages between 2500 to 2999

2.) Post-problem normal buff average. - Post problem range 1018 to 2680 with vast majority between 1080 to 1900

3.) Date when you noticed that buffs got lower (the closer you can get on this the more helpful it is.) - Date I noticed the buffs started stinking was the day of Publish 6 (Feb. 11, I believe)

4.) Normal Buff pack power level (range) and which adv components do you use. - Normal buff pack level I use are 861 to 923 (I have both sets of power ranges, and have experimented using one from each power pack on the same character multiple times - power rating seems to be basically meaningless now and each falls within the same buff range). All made with adv components and over 900 overall quality resources.

5.) Droid rating (If no droid, please type "med center 100") - Probot with 110 med rating. Also tried 100 hospital, and a combination of hospital AND droid. No difference in buffs.

6.) Wound treatment rating without clothes - Master doctor at 100 rating

7.) Wound treatment rating with clothes (include ratings over +25 just in case). - Wound treatment rating with clothes +200 (+25 headwrap, +25 shirt, +25 skirt, +25 duster)

8.) Server - Eclipse


Kubernetes
Wed Feb 18, 2004 4:29 pm
#80

Oh good, I thought it was just me.



1.) Pre-problem normal buff average. - 2200+


2.) Post-problem normal buff average. - about 1400



3.) Date when you noticed that buffs got lower (the closer you can get on this the more helpful it is.) - Around the time of Publish 6

4.) Normal Buff pack power level (range) and which adv components do you use. - 875, all advanced components

5.) Droid rating (If no droid, please type "med center 100") - R2 or Probot- 110

6.) Wound treatment rating without clothes - 100

7.) Wound treatment rating with clothes (include ratings over +25 just in case). - 127

8.) Server. - Valcyn





Hasdrubal Carthago/ KNR/ Valcyn

Master Doctor/ Novice TKA/ ex-Master Carbineer
Doctors' Motto: Stop the Tumble Monkeys!
MarlboroRed
Thu Feb 19, 2004 3:05 pm
#81


1.) Pre-problem normal buff average.

2.) Post-problem normal buff average.

3.) Date when you noticed that buffs got lower (the closer you can get on this the more helpful it is.)

4.) Normal Buff pack power level (range) and which adv components do you use.

5.) Droid rating (If no droid, please type "med center 100")

6.) Wound treatment rating without clothes

7.) Wound treatment rating with clothes (include ratings over +25 just in case).

8.) Server.

-----------------------------------------


1. Was not a master doc.


2. 1800-3000


3. No point of reference


4. all advanced comps, janta blood.... power: 1007 (action) 1000 (others)


5. 110


6. 100


7. 125


8. Gorath




Ceri
Mastered: Everything but Bounty Hunter, and therefore I QUIT!


zebraone100
Fri Feb 20, 2004 3:21 pm
#82



Cudos to you, Traigus, for what your trying to do, but that is where my compliments end. The data you have collected is quasi-scientific at best and i will be more than happy to explain why.



Problem 1. No clear direction on how to collect statistical data.

Not everyone in this forum knows how to give you the data that you were looking for. After all, not everyone is math inclined. Since the data itself may not have beenacurately collected and therfore may be wrong in and of itself, it can not be assumed that the data supports a legitimate conclusion. Another way of saying this is: If you use bad information to make a decision, you will probably make a poor decision.



This is how the data shoud have been collected. In order to find the average ('mean average' for those of you who are math inclined...), you add up all of the results and divide by the number of results. For example, say you buffed five people and their results were as follows: 2835, 899, 1476, 1340, 3011. You would add these numbers up,



2835+899+1476+1340+3011=9561



then divide it by the number of results in the sample (five... count them if you dont believe me!)...



9561 / 5= 1912 (actually 1912.2 but I'm going to round my results from now on...)



and presto... the 'mean average' health buff is 1912. Note: The only way to compute the average is to physically write down every result that you get, add it up on a calculator or spreadsheet and finally compute the average. By the way, did i remember to tell you to physically write down every result? If not, let me say it again: Physically write down every result that you get. You cannot eyeball this. period. When ever you buff some one for 900 points, you get angry or anoid. When you buff some one for 3400 points you get excited or happy. when you buff someone for somewhere in the middle you think very little of it. The point is that some results are more "memorable" than others, and may cause you to skew your analysis.We must be able to remove the emotion from this soI'll say it once more for good measure... Physically write down every result and use these results to compute the actual mean average.



The second part of collecting data is the sample size. If you just take two numbers and find the average, your result will be less accurate then if you had taken three numbers and found the average. In fact the more numbers you take into your computation, the more representive the mean average will be of your actual future and past results. For example, if you took five results and used them to compute the average, who's to say that the five results weren't just a string of really lucky buffs ( or unlucky for that matter )? However if you used 100 results, the odds of all 100 being abnormaly high or all 100 being unusually low, becomes less and less.




Problem 2. The Statistical data presented in the follow up posts is most likely invalid.

Many of the posts describe the average as being somewhere between this number and that number. If it were computed correctly (ie. see problem 1.) they would have had a specific result. As for most of the remaining posts, most of them are presented as a number that ends with 2 zeros (ie. 1400, 1500, or 2300 not 1467, 1521, or 2301). The odds ofa properly computed meanaverage ending with 2 zeros is one in one hundred. When you consider that most of the remaining posts are presented this way, it is easy to conclude that something is a-miss, since the odds of that happening is astronimical. Not that it couldn't happen, but it is doubtfull.



Problem 3. The improbability that anybody had viable statistical data prior to the "problem".

First of all, see problem 1 and 2. Second, why would anyone have bothered to collect the data in the first place? Did everyone in this group have a crystal ball that gave them insight in to the future that made them say to themselves: "Hmm, I better collect this data now so i can compare it to next weeks results" UH... NO. Perhaps they traveled back in time using their handy dandy time machine to collect the data? Not likely. Most likely is that it was that special brand of emotional data that i talked about in problem 1.


Problem 4. Too many variables between now and pre-patch make comparisons less reliable.

This is probably not as significant as the proceeding 3, but is certainly worthy of making note... Were your buff packs the same? Bio enhanced clothing better or worse (even better clothing might not be better. Remember the medical knowledge bug, that when u experimented on knowledge requirements it actually went up instead of down?) and the list goes on and on...


Summation.

It is not my point to say that there is no problem. There very well may be one.The fact that this looks scientific only perpetuates the myths inour career field. The only way we are going to get past these myths, is for the devs to publish, outright, the formulae used to compute buffs, heals, wounds, etc... That would end half of the endless debates on this board in an instant, and also allow us to more easilly spot bugs.


Other than that, great job.

Traigus
Sun Feb 22, 2004 12:15 pm
#83

The science is not relevant here really. I did take a stats class in college.

The average user doesn't log info, nor has enough of the exact same pack to really get a meaningful sample (using just enhance action from a 1000 pack run for example).

The thread is mostly to tally up the number of people with/without the problem, and to look for big and obvious problems (BE clothes, droids etc.)

We aren't trying to solve the problem... just prove a problem exists, so the devs can find and fix it.

QA guys can buff 6000 times to action and find a real sample, as they get paid for it.

Honestly, my 2 doc characters don't suffer from this. In a game this comnplicated, the ability for a player to find the actual (or all of) the cause(s) of a bug is pretty small.

-T



------------------------------------------------------------------------------------------------------

"We've got a blind date with destiny -- and it looks like she ordered the lobster."

------------------------------------------------------------------------------------------------------
zebraone100
Mon Feb 23, 2004 10:45 am
#84





traigus posted:



We aren't trying to solve the problem... just prove a problem exists, so the devs can find and fix it.





First of all, I should apologize for coming across as arrogant in my first post, sometimes I get cranky.


That was the crux of my argument. The "data" that was collected was so flawed, that it can not be used to prove or disprove anything. I have no doubt that people think there is a problem, but this data doesn't provethat thereis a problemin any way, shape, or form (seemy previous post for the reasons). Furthermore, I do not deny that thier may be a problem. Good luck in proving it though, since we still don't have any hard datafrom the past (hard data = actual numbers not"gut feelings" or as i refered to it in my original post, "emotional" data) to compare with todays hard data.


There is a reason I called it quasi-science. On the surface, it appears to be good data, andIt appears to support an unwaivering conclusion. It is "quasi" because these "facts" are mearly illusions.


Quasi-science is dangerous. It focuses attention on items that may not need that attention, andin the process, it diverts that attention away from actual problems. It also generates or perpetuates myths. Not only that, but quasi-science insights undo anger by portraying these myths as bonifide facts. In short, the illusion tricks us into looking the wrong way, or encourages us to make poor choices.


Like I said before, I like what you're trying to do. I just don't feel this thread came anywhere close to accomplishing it. As an aside, I believe the real problem lies in how random buffs are, and should be fixed by establishing a "bell curve" random function that places 90% of all buffs in a 200pt range or something similar, 5% in the next 100pt range above and below, 2.5 % in the next range, and so on (these numbers aren't presice, but they give you a rough idea of what I'm saying).... That, however, is for another thread when I've had a chance to really think it through and come up with an implementable formula.

EnFERn0
Thu Mar 04, 2004 3:12 am
#85

This occured Yesterday, pre/post-Bivoli Tempari


1.) Pre-problem normal buff average.




2.) Post-problem normal buff average.

3.) Date when you noticed that buffs got lower (the closer you can get on this the more helpful it is.)

4.) Normal Buff pack power level (range) and which adv components do you use.

5.) Droid rating (If no droid, please type "med center 100")

6.) Wound treatment rating without clothes

7.) Wound treatment rating with clothes (include ratings over +25 just in case).

8.) Server.


POST:


1.) Pre-problem normal buff average.


2.) Post-problem normal buff average.

3.) Date when you noticed that buffs got lower (the closer you can get on this the more helpful it is.)

4.) Normal Buff pack power level (range) and which adv components do you use.

5.) Droid rating (If no droid, please type "med center 100")

6.) Wound treatment rating without clothes

7.) Wound treatment rating with clothes (include ratings over +25 just in case).

8.) Server.



-----------------------------------------------------------------------------------------------
'Vek
Master Shipwright
Vendors at Naboo, Krath (5350 3610)

-----------------------------------------------------------------------------------------------

EnFERn0
Thu Mar 04, 2004 3:19 am
#86


This occured Yesterday, pre/post-Bivoli Tempari


BEFORE using Bivoli Tempari


1.) Pre-problem normal buff average.

Average on 2.5k

2.) Post-problem normal buff average.

Average on 1800

3.) Date when you noticed that buffs got lower (the closer you can get on this the more helpful it is.)
Approx 2 weeks ago, buffs started to be very random.


4.) Normal Buff pack power level (range) and which adv components do you use.

D-class 900+ Rating

5.) Droid rating (If no droid, please type "med center 100")
Droid, 110.


6.) Wound treatment rating without clothes

100

7.) Wound treatment rating with clothes (include ratings over +25 just in case).

125

8.) Server.

Sunrunner



AFTER using Bivoli Tempari +24%


1.) Pre-problem normal buff average.

No data

2.) Post-problem normal buff average.

Average on 1400 <- Actually dropped 24%

3.) Date when you noticed that buffs got lower (the closer you can get on this the more helpful it is.)
Approx 2 weeks ago, buffs started to be very random.


4.) Normal Buff pack power level (range) and which adv components do you use.

D-class 900+ Rating

5.) Droid rating (If no droid, please type "med center 100")
Droid, 110.


6.) Wound treatment rating without clothes

100

7.) Wound treatment rating with clothes (include ratings over +25 just in case).

125 + 24% Bivoli Tempari

8.) Server.

Sunrunner



-----------------------------------------------------------------------------------------------
'Vek
Master Shipwright
Vendors at Naboo, Krath (5350 3610)

-----------------------------------------------------------------------------------------------

Traigus
Thu Mar 04, 2004 5:39 am
#87

Unstickied in preparation of new buff system on TC. Any oddities will be easier to track when it goes live. I'll start a new thread if there are problems after Pub 7.

-T



------------------------------------------------------------------------------------------------------

"We've got a blind date with destiny -- and it looks like she ordered the lobster."

------------------------------------------------------------------------------------------------------
Page 7 of 7