| Pages: [1] :: one page |
| Author |
Thread Statistics | Show CCP posts - 0 post(s) |
St Devil
Caldari
Defcon One
Phalanx Alliance
   |
Posted - 2007.06.23 19:15:00 -
[1]
I have the feeling that Invention has reached a good level of maturity.
Shouldn't CCP now publish the exact formula used to compute the chance of getting a T2 BPC?
Of course, you can find some guessed formulas in the forum, but depending on the one I used, I get 26% to 48% chance for a specific module. This leads me to some profits or not.
I think some transparency is recommended for CCP. Otherwise, one of those days, we will once more read that some people are more informed then others. I know this argument is not very subtle but it's not far away from what I hear or read each day.
I don't know if publishing this formula could have negative impacts on CCP business. They will have less freedom to change the rule at their wish, but that's also why I said that I think invention is rather mature and changes are therefore more related to fine-tuning of the invention mechanics than to a complete re-design.
As Invention has proven to work, it's now time to make it publicly accessible.
StD
|
Salvis Tallan
Gallente
The Shadow Order
Hydra Alliance
|
Posted - 2007.06.23 19:24:00 -
[2]
I dunno. I mean yes it would greatly help with predicting profits and whatnot, but I for one have been enjoying this little quest of finding the right formulas. Not knowing exactly does lead to a bit of danger when you go looking for items to invent, since you do not know your profit margin, but like I said before, it is a bit exciting in what many see as a dull section of industry. The only thing I would ask is if they ever change the formulas, or if there is some built in randomness to the equations (which I really hope there isn't). If the formulas are static, then its only a matter of time untill we come upon the proper formula, or one that works really really well.
On the other hand, even if we did have the formula for chance of success, since it is a chance, you can produce for months straight and still have your results differ from what it should be. There is still risk involved, so in that sense it might not hurt to know the formula.
------
|
Ran Tai
Caldari
Highlifter Inc.
|
Posted - 2007.06.24 06:57:00 -
[3]
There actually is a formula for Invention. Unfortunately I am not able to post the location in here. 
"TANSTAAFL" |
Qual
Gallente
Cornexant Research
|
Posted - 2007.06.24 11:15:00 -
[4]
 Originally by: Ran Tai
There actually is a formula for Invention. Unfortunately I am not able to post the location in here.
Just post it here if ya got it. Shouldn't be that hard.
"The short version: Qual is right." -Papa Smurf |
Trak Cranker
Serenity Inc
|
Posted - 2007.06.24 11:41:00 -
[5]
Originally by: St Devil
I think some transparency is recommended for CCP. Otherwise, one of those days, we will once more read that some people are more informed then others. I know this argument is not very subtle but it's not far away from what I hear or read each day.
While I also enjoy the hunt for the formula and generally don't think they should release these things - the above is a very good argument.
Still not sure its best for the game to just release those things. Or best for those people that have actually put some time into researching a little on this.
But it is a good argument, sadly.
|
Trak Cranker
Serenity Inc
|
Posted - 2007.06.24 11:42:00 -
[6]
Edited by: Trak Cranker on 24/06/2007 11:41:31
Originally by: Ran Tai
There actually is a formula for Invention. Unfortunately I am not able to post the location in here.
Ran Tai, you copycatying bastard! 
If you have the formula - spill it. :)
|
St Devil
Caldari
Defcon One
Phalanx Alliance
|
Posted - 2007.06.25 10:14:00 -
[7]
I had nothing against the formula's hunt but something has changed these last weeks and all my invention jobs failed since then.
I was used to invent Large Shield Extender and Invulnerability Shield with a success rate of 40% (4/10). But I changed my research location to a place where I don't have standings and someone told me that Invention has been nesfed two patches ago.
I then searched for a new formula and found one that predicts me a reduced 26% chance of success. But even these 26% should have given me 2 BPC. Therefore before loosing additional expensive datacores, I would like to know if it's just bad luck or if it's also related to the jobs location (and my standings).
The best way to make me feel good about this is to know the Formula.
StD
|
Trak Cranker
Serenity Inc
|
Posted - 2007.06.25 12:32:00 -
[8]
If we are only talking 10 tries, or 10 before and 10 now, that is far from a good sample size to be certain that anything has changed. Or were you speaking in percentages - on a larger sample size?
What was your skill levels?
|
Liro Farr
|
Posted - 2007.06.25 13:19:00 -
[9]
After last patch i got 2 time success chance nerf on vagabond inventions. Run 10 invention jobs every day. Skills 5/5/5 |
Qual
Gallente
Cornexant Research
|
Posted - 2007.06.25 15:02:00 -
[10]
Originally by: Liro Farr
After last patch i got 2 time success chance nerf on vagabond inventions. Run 10 invention jobs every day. Skills 5/5/5
Unless you giv eus your setup and actual succes chance before and after patch, thats not really helpfull. You could just have been lucky before patch or unlucky after...
"The short version: Qual is right." -Papa Smurf |
St Devil
Caldari
Defcon One
Phalanx Alliance
|
Posted - 2007.06.28 12:38:00 -
[11]
Edited by: St Devil on 28/06/2007 12:37:36
Ok. Here am I !!! The king of misfortune.
Here are my skills and decryptor:
Caldari Encryption Method: 3
Hydromagnetic Physics: 4
Quantum Physics: 3
Item used: Basic (meta=1)
Decryptor Used: Interface Alignment Chart (chance factor: 0.4)
Item invented (or more exactly: not invented): Large Shield Extender, Invulnerability Field, Heat Dissipation Field, Ballistic Deflection Field, Photon scattering Field.
I made 5 copies of 300 runs of each of these 5 Items.
One or two weeks ago I get the needed Datacore to launch 8 invention jobs (5 Large shield Extender and 3 Invulnerability Field). I get 2 Large shield Extender II BPC and 2 Invulnerability Field II BPC of 19runs each.
Then came the Patch and I started the remaining Invulnerability Shield BPC, both failed.
And then Revelation II was deployed and since than I have launch my 15 remaining jobs all of them failed.
I found a formula in this forum that gives me a 26% chance of success with my current skills and decryptor.
I try to find with this value what was my chance to get 17 consecutive failing jobs.
Chance = (1-26%)^17 = 0.6% !!!!!!
No way, this is not possible !!! Something is wrong.
Can someone help me?
StD
|
Jifai
|
Posted - 2007.06.28 16:05:00 -
[12]
Originally by: St Devil
Chance = (1-26%)^17 = 0.6% !!!!!!
No way, this is not possible !!! Something is wrong.
Can someone help me?
StD
Highly possible. If you do 20 inventions per day, at your claimed success odds of 26%, failure streaks over 17 long can be expected once or twice a month. I think your success odds are actually lower than 26%.
In practice using the 0.4x decryptor, I see failure streaks of 30+ ship jobs at least once a month.
|
chapter13
|
Posted - 2007.06.29 01:29:00 -
[13]
I'm going to go ahead and help you with a small math lesson.
a .4 +9 decrypter should not be used on any modules (unless the bp copying time is very long). Why? the math...
max run bpc=a 10 run t2 bpc (no decrypter)
use a .4 +9 decrypter you cut your success rate by over half and dont even double the run output (sucess = a 19 run bpc) on your sucesses.
You're only hurting yourself using those mods on anything but mods with crazylong copy times and ships.
|
Kylar Renpurs
|
Posted - 2007.06.29 03:47:00 -
[14]
Edited by: Kylar Renpurs on 29/06/2007 03:51:59
Quote:
As Invention has proven to work, it's now time to make it publicly accessible.
Huh? It's not publicly accessible?
Sorry, guess I'm not an inventor when I buy the interface, some datacores, some decryptors, BPCs and meta items and make T2 goods
As for your chances,, 26% does not imply you'll get anything from 17 attempts. In fact, there's another 58 failures you're supposed to have if you've got a real 26% success rate in an outcome-eliminating set of 100.
And if you tried 1000 times, you've got another 580 failures before you should start getting success.
You have 26% chance over an infinite amount of attempts. So 16/17 attempts failed? so what? Complain about it after 100 failures.
Sorry, but too many people make the false assumption that 26% chance of success means 'Every 4th attempt will succeed'
Improve Market Competition! |
St Devil
Caldari
Defcon One
Phalanx Alliance
|
Posted - 2007.06.29 11:49:00 -
[15]
@ChapterXIII: You're perfectly right, it sometime hurts to find yourself so stupid  .
I get to fast at creating my sheets and forget to unbind invention & manufacturing, this leads me to stupid behavior. Shame on me.
Thanks anyway for your math lesson ;).
StD
|
Mi'ira
|
Posted - 2007.06.29 11:49:00 -
[16]
Edited by: Mi''ira on 29/06/2007 11:52:59
To spill some beans, some people claim the formula is the following:
(this was not derived from tests, available since around when invention hit the test server, maybe some people dont like me anymore after posting this  *waves*, but imho the formula is worthless even if its real, the vital dependancies have all already been measured ingame to an accuracy which is sufficient for making invention decisions):
Success chance = base * ( 1 + (Skill1 + Skill2)/(10*( 5- Metalevel))) * ( 1 + Skill3 / 100) * DecryptorMultiplier
Skill3 is the decryption, everything else should be clear.
Problem is, all depends on the base chance really, the skill part can easily be measured using a few hundred cheap runs of something if one would know the base chance.
With ships the base is supposed to be 20% (lowest), thus one can measure skill influence with meta-level 0 there, and end up having(for one fixed set of skill levels):
Chance = Base * (measured value) * decryptor
Thus the vital thing would be the table of base chances for the modules, or some formula to derive it from other values if its derived(f.e. from type(module,ship,rig) and number/type of datacores).
|
Salvis Tallan
Gallente
The Shadow Order
The Shadow Ascension
|
Posted - 2007.06.29 12:30:00 -
[17]
Originally by: Mi'ira
Problem is, all depends on the base chance really, the skill part can easily be measured using a few hundred cheap runs of something if one would know the base chance.
With ships the base is supposed to be 20% (lowest), thus one can measure skill influence with meta-level 0 there, and end up having(for one fixed set of skill levels):
Chance = Base * (measured value) * decryptor
Thus the vital thing would be the table of base chances for the modules, or some formula to derive it from other values if its derived(f.e. from type(module,ship,rig) and number/type of datacores).
Base chance is given in data dump
------
|
Mi'ira
|
Posted - 2007.06.29 12:37:00 -
[18]
Wasnt in the last one i checked ;P, which was some time ago.
As i only do one type of invention it doesnt concern me anyway, but if its really an independant parameter and can be seen in newer data dumps, thats it, open invention.
Dependancy on skill is not that vital and dependancy on metalevel can and has been measuered.
|
Salvis Tallan
Gallente
The Shadow Order
The Shadow Ascension
|
Posted - 2007.06.29 12:54:00 -
[19]
Originally by: Mi'ira
Wasnt in the last one i checked ;P, which was some time ago.
As i only do one type of invention it doesnt concern me anyway, but if its really an independant parameter and can be seen in newer data dumps, thats it, open invention.
Dependancy on skill is not that vital and dependancy on metalevel can and has been measuered.
Ive been using... dbo_invBlueprintTypes, the chanceOfReverseEngineering. This is because (to me anyways) RMR export had everything at 0, whereas now tech 2 things have a value but they just didn't rename the table. Also, things that come in pairs, like arazu and lachesis have a 0.2 value but other ships with only one tech 2 version, like helios, have a 0.4 value.
------
|
Mi'ira
|
Posted - 2007.06.29 13:17:00 -
[20]
Found it, looks reasonable.
From my info (chruker) helios seems to have 0.2 though.
Wich sounds sensible, cause imho the game first decides which variant you get in case of 2 possibilities and then the chance is calculated.
Thus no need of a factor 1/2 for the pairs, or did i miss something?
|
Salvis Tallan
Gallente
The Shadow Order
The Shadow Ascension
|
Posted - 2007.06.29 13:34:00 -
[21]
Originally by: Mi'ira
Found it, looks reasonable.
From my info (chruker) helios seems to have 0.2 though.
Wich sounds sensible, cause imho the game first decides which variant you get in case of 2 possibilities and then the chance is calculated.
Thus no need of a factor 1/2 for the pairs, or did i miss something?
I just picked helios off the top of my head. Bad example I guess. But even still a base of 0.4 then modified will give you twice the chance of a 0.2 modified, and so either way the chance would be half.
------
|
Trak Cranker
Serenity Inc
|
Posted - 2007.06.29 22:51:00 -
[22]
Originally by: Mi'ira
With ships the base is supposed to be 20% (lowest), thus one can measure skill influence with meta-level 0 there, and end up having(for one fixed set of skill levels):
Chance = Base * (measured value) * decryptor
Where is the skill level taken into account there?
|
Mi'ira
|
Posted - 2007.06.29 23:33:00 -
[23]
Edited by: Mi''ira on 29/06/2007 23:32:51
Originally by: Trak Cranker
Originally by: Mi'ira
With ships the base is supposed to be 20% (lowest), thus one can measure skill influence with meta-level 0 there, and end up having(for one fixed set of skill levels):
Chance = Base * (measured value) * decryptor
Where is the skill level taken into account there?
In the measured thing.
Base is (probably) well known, decryptor the same, and enough people have posted on these forums to at least extract a skill level/meta level to success chance table, where you would only apply base and decryptor afterwards, if not the actual formula.
|
Trak Cranker
Serenity Inc
|
Posted - 2007.06.30 09:23:00 -
[24]
Originally by: Mi'ira
In the measured thing.
Base is (probably) well known, decryptor the same, and enough people have posted on these forums to at least extract a skill level/meta level to success chance table, where you would only apply base and decryptor afterwards, if not the actual formula.
It might work to a degree, yeah. But you would have to trust the accuracy of the reporting and there hasn't been posted _that_ many datasets.
But it might be a good substitute, while we figure the formula put.
Because your "(measured value)" is one of the grayest boxes in Eve.
Our skills does not even say what chance increments they give, unlike almost all other skills.
|
Mi'ira
|
Posted - 2007.06.30 13:15:00 -
[25]
Edited by: Mi''ira on 30/06/2007 13:16:53
I know f.e. that with skills 4/4/4 and meta-level 0(no item) my success multiplier due to skills is 1.20 +-0.02.
Using a base chance of i.e. 20% for ships and a 0.4 decryptor that results as 9.6%. Inventing modules with a 1.3 decrypter would yield 62.4% with meta-level 0.
With skills 2/2/2 one would expect 2-3 multipliers to be half that big, resulting in a skill multiplier of 1.1.
(in the 2 factors case(1 encryption, 1 sum of the two others) the factors could be i.e. 1.16 and 1.04 or 1.1 and 1.1 or whatever for 4/4/4, result if you half them will yield 1.1; in the three factor case i.e. 1.06 to the power of three, giving 1.096 for 2/2/2, nearly the same)
This information is fully sufficient to make a lot of decisions.
The tricky part is how meta-level enters the whole thing. If its a "waste" like modifier(higher levels reducing waste) its also easy enough, and enough data available, if its not, happy times
For illustration if the formula would be:
Success chance = base * ( 1 + (Skill1 + Skill2)/(10*( 5- Metalevel))) * ( 1 + Skill3 / 100) * DecryptorMultiplier
This would admittedly be quite mean and screw with decisions, as high meta levels would only be worth it with high skills in the research fields. Determining such a dependancy from data would be a tough nut, but would be vital for module inventors who often switch fields.
|
Mi'ira
|
Posted - 2007.06.30 13:24:00 -
[26]
Originally by: Qual
Originally by: Ran Tai
There actually is a formula for Invention. Unfortunately I am not able to post the location in here.
Just post it here if ya got it. Shouldn't be that hard.
If Ran Tais claim is genuine, could be the above one 
|
Mighty Baz
HUSARIA
Curatores Veritatis Alliance
|
Posted - 2007.07.02 06:17:00 -
[27]
Edited by: Mighty Baz on 02/07/2007 06:16:45
Originally by: Mi'ira
Edited by: Mi''ira on 30/06/2007 13:16:53
Success chance = base * ( 1 + (Skill1 + Skill2)/(10*( 5- Metalevel))) * ( 1 + Skill3 / 100) * DecryptorMultiplier
how can I find out "base"?
______________________________________________
based on legendary XVII century Polish winged cavalry |
Qual
Gallente
Cornexant Research
|
Posted - 2007.07.02 06:36:00 -
[28]
Originally by: Mi'ira
This would admittedly be quite mean and screw with decisions, as high meta levels would only be worth it with high skills in the research fields. Determining such a dependancy from data would be a tough nut, but would be vital for module inventors who often switch fields.
Unless you happen to have all the skills maxed out... 
"The short version: Qual is right." -Papa Smurf |
Mi'ira
|
Posted - 2007.07.02 16:06:00 -
[29]
Edited by: Mi''ira on 02/07/2007 16:06:26
Originally by: Mighty Baz
Edited by: Mighty Baz on 02/07/2007 06:16:45
Originally by: Mi'ira
Edited by: Mi''ira on 30/06/2007 13:16:53
Success chance = base * ( 1 + (Skill1 + Skill2)/(10*( 5- Metalevel))) * ( 1 + Skill3 / 100) * DecryptorMultiplier
how can I find out "base"?
Supposedly its the chanceOfReverseEngineering of T2 blueprints in the database. Can be found in the datadump, browsable online i.e. http://games.chruker.dk/eve_online/market.php.
Would make sense, the field has been in the database for ages(accounts for the name), since invention its populated with numbers like 0.2 and 0.4.
|
| |
|
| Pages: [1] :: one page |
| First page | Previous page | Next page | Last page |