Pages: [1] 2 3 4 5 6 7 8 9 10 11 12 13 14 .. 14 :: one page |
|
Author |
Thread Statistics | Show CCP posts - 0 post(s) |
Vladimir Tinakin
Caldari Hadean Drive Yards Archaean Cooperative
|
Posted - 2007.03.15 16:58:00 -
[1]
All right, we know that invention is chance based. However, what isthe chance of success or failure, exactly?
The Devs are playing the "stay silent, don't document, and let the players figure it out" card. While this has worked in the past, I have to go on record as saying it's a pretty crappy stance.
Anyway, this topic is to try to consolidate everyone's experience in invention, and try to hammer out the approximate variables and benefits that impact it.
I'm going to start with a few assumptions. They may be incorrect, but I certainly hope not!
I)Any invention job that is submitted has a chance--though possibly a miniscule one--of success. - This means that the "base chance of success" for an item, if there is such a thing, cannot be negative. It does not rule out a 0% base chance, however.
II)You cannot have a 100% chance of a successful invention. - This means that even with the best decryptor, meta item, and skills you can't get 100%. Since the best decryptor is known to give you a 1.3x multiplier on your success rate, this means that the skills and meta item and base chance combined must be less than 75% chance of success.
III)Every decryptor is intended to be used/useful. - Basically, this means that no decryptor is intended to be a waste of space. Some decryptors have "Average" benefits or a negligable impact on success rates, but more importantly for this discussion some have a significant negative impact on success rate. What this means, is that the skills+meta item+base chance combination has to be a decent amount on its lonesome, otherwise slashing it in half would basically make using that decryptor an exercise in futility.
These assumptions are in no way verified or proven to be true; but one would hope that CCP would have taken them into account when designing the system. Additionally, a fair amount of the assumptions are derived from the decryptor modifier information from Coldfront. If these have been changed or tweaked at all, then obviously there is a good margin for error here!
Skills There are a few schools of thought we can take here. Obviously skills are an important part of invention. The first school is that the Encryption Methods skill is of primary importance, and that the other two field skills are of secondary benefit. The other school is that all three skills are of equal benefit. For the purposes of creating a model, I'll assume the skills do simple addition to your chance of success, as opposed to the decryptors' percent multiplier.
If we take the first school of thought to heart, we have the following possibility:
Encryption methods: +3-5% chance per level Supporting field skills: +2% chance per level
This leaves us with a base chance from pure skills of anywhere from 35% to 45% when maxed out, depending on how weighted the encryption methods skill is.
The second school would probably play out like the following:
All 3 skills: 2-4% per level
This gives us 30%, 45%, or 60% base value. While it is tempting to throw out the high value as being too high, remember that this is with maxed skills and even with the best chance decryptor allows for a 1-in-5 chance of failure. Still, one of the lower values is perhaps more likely...so for the purposes of argument I'm going to peg the skills part of invention as giving a 45% chance of success when maxed -- slightly worse than even odds.
Meta Item It is hard to say how this impacts. It could be a simple multiplier of the meta value added on top of skills, or it could be a multiplier--though for the last to be of significant impact it would likely have to be the meta value, multiplied, then applied as a multipler (eg, meta value of 5, multiplied by 5, then applied to the skill chance as a 1.25% modifier. Whereas a 1.05 multiplier would be almost insignificant to the best-case skills success chance of 45%.)
----------------------------------------------- Adm Vladimir Tinakin CFO Hadean Drive Yards |
Vladimir Tinakin
Caldari Hadean Drive Yards Archaean Cooperative
|
Posted - 2007.03.15 17:01:00 -
[2]
Decryptors are more or less the only "known" factors in invention chance, from the data dumps given by CCP. However, as mentioned before, these values are subject to modification and "Tweaking" by the devs and are therefore not 100% concrete until Invention gets fully shaken-down and balanced as a game mechanic.
Please post here with any thoughts or impressions you get from your invention experiences; please keep flames and rhetoric to a minimum...I'd like to try to be able to reason this out (unless a Dev would be so kind as to share with us the actual formulas....you'd have a HUGE chunk of the playerbase contented! ) ----------------------------------------------- Adm Vladimir Tinakin CFO Hadean Drive Yards |
Vladimir Tinakin
Caldari Hadean Drive Yards Archaean Cooperative
|
Posted - 2007.03.15 17:29:00 -
[3]
My own invention experience:
Covert Ops Cloak attempts: 7 Successes: 4 Failures: 3
Relevent skills are all at level 4. Best success decryptor used every time, with a Proto Cloak I used as the base item.
So, it looks like slightly better than even odds for the cloak invention at my present skill level. ----------------------------------------------- Adm Vladimir Tinakin CFO Hadean Drive Yards |
Salvis Tallan
Gallente The Shadow Order Hydra Alliance
|
Posted - 2007.03.15 17:38:00 -
[4]
From the data dump, we can get the following: Chance of Duplicating - Item Chance Of Reverse Engineering - Blueprint
When we look up the item, say Sensor Booster I, it has a duplicating chance of 0.07, which we will assume is 7%
When we look up its blueprint, its chance of reverse engineering is 0. When we look up the Sensor Booster II blueprint, we get 0.4 (40%).
If we look up a ship, such as the rifter, which has the Wolf and Jaguar as tech 2 versions, their reverse enginering numbers are both 0.2 (half of 0.4). I believe that this Chance of Reverse Engineering on the Tech 2 blueprint is the base % to successfully get the blueprint. I think when you add in the tech 1 version, it adds the Chance of Duplicating, or 7% for the booster.
We can also look up the meta number for an item, such as the Shadow Serpentis Sensor Booster is part of metagroup 4. Gallante navy sensor booster is also group 4. The Alumel-Wired I Sensor Augmentation (named version) has metagroup number 1. Could this be a straight multiplication to the Chance of Duplicating number? Problem here is that ships have high metagroup numbers, so this theory could be bunk, or more complicated.
Also so far by reading the dev blogs and the results of invention, I do not believe there is much randomness involved in the final product's numbers. Most items with the same decryptor all have the same ME, PE and the runs seem to change a bit, but not alot. The Devs have also commented that they want to add More randomness in for this, leading me to believe there isnt much already. Decryptor information can be found on various site or in the data dump, however these numbers have been tweaked and I doubt we will get an updated data dump any time soon, so the best way to get their numbers now is to try and work backwards from results. ------
|
Vladimir Tinakin
Caldari Hadean Drive Yards Archaean Cooperative
|
Posted - 2007.03.15 17:49:00 -
[5]
At this point, I'm more concerned with the chance of success; ME/PE and number of runs are another ball of wax entirely that may involve another computation involving skills and the T1 BPC quality and runs...
That said, I'm using an ME20/PE0 50-run Proto Cloak BPC for each attempt and getting ME -7, PE -4 (I think) and 9 runs consistently. So the quality is definitely an input-x,y,z; get-a,b,c function. ----------------------------------------------- Adm Vladimir Tinakin CFO Hadean Drive Yards |
Salvis Tallan
Gallente The Shadow Order Hydra Alliance
|
Posted - 2007.03.15 18:08:00 -
[6]
Originally by: Vladimir Tinakin At this point, I'm more concerned with the chance of success; ME/PE and number of runs are another ball of wax entirely that may involve another computation involving skills and the T1 BPC quality and runs...
That said, I'm using an ME20/PE0 50-run Proto Cloak BPC for each attempt and getting ME -7, PE -4 (I think) and 9 runs consistently. So the quality is definitely an input-x,y,z; get-a,b,c function.
Yea, results seem to only depend on the decryptor. But I am fairly confident about the success chance numbers. In which case our equation so far is: (BP trying to Invent chance + (Item meta number*?meta multiplier))*Skills?*Decryptor chance modifier ------
|
Fergus Runkle
Minmatar
|
Posted - 2007.03.15 18:34:00 -
[7]
So far I have tried the following jobs,
1 x cloak 4 x Heavy launcher 4 x Warp disruptor 2 x Shield power relay
All skills have been at 4, I have always used the best meta item (not available for the cloak so used a proto 1) and the "best chance" decryptor.
The cloak job failed, all the others succeeded (with various messages from "just got there" to "could do it in your sleep"). This leads me to believe that the meta item is quite important and maybe carries a higher weighting than you suspect.
Ever single bpc has come out at 14 runs, -7ME, -2PE
The various input bpc's have had varying ME and PE levels but always been max runs.
I wonder if the ME / PE of the input matters at all ? .. maybe past a certain point its worthless (as in production).
Have a look at this site in game
http://eve.databased.at/decryptors.asp
|
Judician Nar
|
Posted - 2007.03.17 21:42:00 -
[8]
Originally by: Vladimir Tinakin
Encryption methods: +3-5% chance per level Supporting field skills: +2% chance per level
I'm just starting to look at invention & have been doing some background reading. What are these supporting field skills you mention??
|
Bentus Kushani
|
Posted - 2007.03.17 23:32:00 -
[9]
Edited by: Bentus Kushani on 17/03/2007 23:33:23 My results: Encryption 3, Supports at 3
Normal Item: Best Chance Decryptor(BCD)/ No item: 0/1 successful BCD/2nd top named items: 18/25
Shield Hardeners: BCD/1st named(only one available): 1/1
E: Also interesting to note is that until yesterday sometime I was having much better results. In the last 24 hours, my results have gone very sour on me. I've failed 5/6 attempts. I changed nothing, perhaps CCP tweaked something in the last day? I know that stastically I fit in, but I went from about 90% success to about 20% success in 24 hours. 6 may not be a huge sample, but it's not statistically insignificant either as these results go. Could just be a run of bad luck, it happens
|
Dyeadmheet
Caldari Khanid Aerospace Group
|
Posted - 2007.03.18 01:24:00 -
[10]
I've started trying to work out what the base is on the test server using level 1 skills only and no base item or decryptor. So far I've gotten:
Shield Power Relay II - 3 of 7 succeeded - 30% Anciliary Current Router II - 2 of 11 succeeded - 18%
Have some ships cooking now but they take a day in NPC station labs.
Of course this isn't really enough samples, and the process of gathering samples is kind of slow. I could speed it up a little by building a POS and using POS labs but I'm too lazy :)
|
|
Vladimir Tinakin
Caldari Hadean Drive Yards Archaean Cooperative
|
Posted - 2007.03.18 02:01:00 -
[11]
Excellent idea. I think I'll do something like that on monday when I get to work....its good to be the net admin. ----------------------------------------------- Adm Vladimir Tinakin CFO Hadean Drive Yards |
AvatarADV
|
Posted - 2007.03.18 02:43:00 -
[12]
Actually, 6 is -by definition- statistically insignificant. You need a larger sample size to rule out clustering effects (essentially, luck of the draw).
This is gonna be a tough one to back-extrapolate because of the various skill effects and the cost of doing it over and over again. Realistically, you need 50-100 identical runs before you can say "this is my expected success rate, with these initial conditions", assuming you do it all in a period of time where the devs haven't poked anything (so throw the data out Tuesday and start over!) This'll get easier. ;p
|
Vladimir Tinakin
Caldari Hadean Drive Yards Archaean Cooperative
|
Posted - 2007.03.18 07:02:00 -
[13]
Beauty of the test server...everything for it costs 100 isk and is market seeded :D
Granted, they may have tweaked something on SiSi, but its probably the best way to figure it out. ----------------------------------------------- Adm Vladimir Tinakin CFO Hadean Drive Yards |
Nedolzna Ovcica
Mindstar Brigade
|
Posted - 2007.03.18 21:59:00 -
[14]
Edited by: Nedolzna Ovcica on 18/03/2007 21:57:19 i did 9 invetion jobs so far, 7 succeded. all skills are lvl 4 and i used best named item and best decryptor everytime. every bpc i got was 14 runs, -7ME, -2PE
didn't try ships yet tough.
|
Dyeadmheet
Caldari Khanid Aerospace Group
|
Posted - 2007.03.19 23:34:00 -
[15]
On the test server I'm up to 33 total, 12 succeeded, 21 failed (36% succeeded). Can anyone with knowledge of statistics tell me what the confidence interval is on this?
All of these jobs were run with: Shield Power Relay BPC(note that all modules are probably the same) 1 run ME0 PE0 All skills at level 1 no base item no decryptor
|
Vladimir Tinakin
Caldari Hadean Drive Yards Archaean Cooperative
|
Posted - 2007.03.20 06:52:00 -
[16]
I'm no statistics major, but I'd say that with a sampling of that size, you're within a few percentage points error of the true "base value"...
Especially with it being around the 33% mark--not objective, but subjective numbers like that "feel" right.
Now the real boring--er, scientific--part is to increment the skills while trying a bunch of jobs at each skill level.
But still, just on the face of it, 30% or so base chance of success...toss in 24-30% from skills all at level 4, and then the best chance decryptor and you have an 80% or so chance of success inventing SPR IIs. Seems a bit high now that I think about it--we're not even factoring meta item in that. Eh, that's why we test.
To the singularity server!
----------------------------------------------- Adm Vladimir Tinakin CFO Hadean Drive Yards |
Hermes Massai
Gallente Coreli Corporation Corelum Syndicate
|
Posted - 2007.03.21 08:08:00 -
[17]
From my testing on SiSi, the runs on the tech 1 BPC only affect the runs on the resulting tech 2 BPC.
I used a Small Armor Repairer blueprint, War Strategon as decryptor (increases success rate), and Small Accommodation Armor Repairer as meta-item. No ME/PE on the BPC.
1-run tech 1 BPC = 4-run tech 2 BPC (4 success out of 5)
100-run tech 1 BPC = 7-run tech 2 BPC (3 success out of 5)
200-run tech 1 BPC = 10-run tech 2 BPC (4 success out of 5)
|
Artmedis Valben
Gallente Lobster of Babel
|
Posted - 2007.03.21 11:52:00 -
[18]
Edited by: Artmedis Valben on 21/03/2007 11:51:16 I think there is a base chance. modified with the encryptor 40-130%
I would assume that the meta item either gives a add on modifier or a percentage modifier, that meta-level 0 items give a base modifier and that higer metalevels add 10% or so to that modifier.
The skill modifier may work similiar to the gang links, with the supporing science skills modifing the encryption skill by 10% per level. Again either an add-on modifier or a percentage modifier.
If we assume that all skills at level 5, meta-item level 4 and best chance encryptor on a module give a chance under 100%. We can sort of see what this means.
module base chance 30% (that Sisi test seems to suggest no more than 35% base chance). Modified with encryptor this yields a chance from 12-39% lets say add 7% for the meta-item, and an extra percentage for each meta-level yielding from 19-50% base chance before skills. Assume skills give an add on effect of 4% per encryption level, and that each level of support skills gives 10% modification to skill effect. that means a minimum of 4,84% (with all skills at level 1) to a maximum of 45% add-on chance.
If this is correct no invention job can ever have a better chance than 95% with modules. Unless you can use high meta-level items like officer or 0.0-faction stuff (meta levels of 6-14).
obviously we need more testing, but the few jobs I have made do not contradict this thesis. ___________________________________________ Selling PERFECT PRINTS of almost all seeded BPOs. Lobster of Babel currently holds 666 of the 723 Tech 1 BPOs seeded. 7 ships and Capital are missing. |
Quutar
Caldari Auraxian Irregulars The Sundering
|
Posted - 2007.03.21 16:31:00 -
[19]
http://oldforums.eveonline.com/?a=topic&threadID=494598&page=1#3
I checked in my own copy of the SQL... those results are valid for the drycptors... not sure what it means though...
Not finding research slots in Empire Space? Try Quutar Research Services. |
Vladimir Tinakin
Caldari Hadean Drive Yards Archaean Cooperative
|
Posted - 2007.03.21 17:40:00 -
[20]
The decryptors appear to multiply your percentage chance of success--the best one gives a 1.3 multiplier, while the 'worst' one more than halves it with a 0.4 multiplier--but if it succeeds you get a huge amount of runs on the BPC.
Testing on SiSi with me0/pe0 1 run covetor BPCs...
With stolen fomulas (best chance decryptor), no covetor as a base item, and the 4/3/3 skills I have on the test server I got 3 successes out of 5 attempts. Obviously not enough to form any conclusions, but it looks like hulks have a low base chance of success. Doing another round of testing now, with no decryptor.
----------------------------------------------- Adm Vladimir Tinakin CFO Hadean Drive Yards |
|
Ruffio Sepico
Minmatar Hidden Agenda Deep Space Engineering
|
Posted - 2007.03.21 18:11:00 -
[21]
Originally by: Vladimir Tinakin The decryptors appear to multiply your percentage chance of success--the best one gives a 1.3 multiplier, while the 'worst' one more than halves it with a 0.4 multiplier--but if it succeeds you get a huge amount of runs on the BPC.
What is a HUGE? amount of runs? I did some jobs last night and I still get the same amount of runs as I did pre-patch. 19 runs of a module using a 300 run t1 bpc, and the circurity schematics decryptor that is supposed to give the most runs. 1 job succeed, 4 failed, and all skills at level 4 + using best metha items.
Home: http://www.hidden-agenda.co.uk
|
Vladimir Tinakin
Caldari Hadean Drive Yards Archaean Cooperative
|
Posted - 2007.03.21 18:14:00 -
[22]
Originally by: Ruffio Sepico
Originally by: Vladimir Tinakin The decryptors appear to multiply your percentage chance of success--the best one gives a 1.3 multiplier, while the 'worst' one more than halves it with a 0.4 multiplier--but if it succeeds you get a huge amount of runs on the BPC.
What is a HUGE? amount of runs? I did some jobs last night and I still get the same amount of runs as I did pre-patch. 19 runs of a module using a 300 run t1 bpc, and the circurity schematics decryptor that is supposed to give the most runs. 1 job succeed, 4 failed, and all skills at level 4 + using best metha items.
Hedging my bets since I wasn't sure if they had changed it to a multiplier.
In any case, at least 5 more runs than the "best chance" decryptors, which in turn add 4 runs to the output. ----------------------------------------------- Adm Vladimir Tinakin CFO Hadean Drive Yards |
Fergus Runkle
Minmatar
|
Posted - 2007.03.21 19:35:00 -
[23]
Originally by: Vladimir Tinakin
In any case, at least 5 more runs than the "best chance" decryptors, which in turn add 4 runs to the output.
Yes but the patch notes said there would be changes, and so far no one is seeing any changes to the number of runs that we have been getting since Rev 1.3.1 (?).
Which was when the runs were seriously nerfed down from 40-90 to 14-19.
|
Supreme Feather
|
Posted - 2007.03.21 20:35:00 -
[24]
So where on gods green earth can I find any info about what decrypters to get?
Please help me, I need the one for aux nano pump I = so armor rigs.
I got no freaking clue! Why doesnt CCP give a freaking database instead of wasting everyones time and ISK.. its ******* fustrating..
Signature: I only take direct trade or Item Exchange after autions please (to prevent scams). When ever I sell or buy! |
Dyeadmheet
Caldari Khanid Aerospace Group
|
Posted - 2007.03.21 20:41:00 -
[25]
One interesting thing I did notice on the test server is that with some 1 run BPCs that I used with a +4 run decryptor, the resulting BPC ended up with 4 runs. This seems to imply that with a really low-run input BPC you can end up with a base runs of 0, which then must get rounded up to 1 if there's no decrypter to add to the 0.
I saw some CCC II BPCs on contract in Jita on the real server that also seemed to show the same sort of behavior.
|
Vishnej
Demonic Retribution Pure.
|
Posted - 2007.03.21 21:49:00 -
[26]
Edited by: Vishnej on 21/03/2007 21:47:37 -
|
Vladimir Tinakin
Caldari Hadean Drive Yards Archaean Cooperative
|
Posted - 2007.03.22 00:01:00 -
[27]
Originally by: Dyeadmheet One interesting thing I did notice on the test server is that with some 1 run BPCs that I used with a +4 run decryptor, the resulting BPC ended up with 4 runs. This seems to imply that with a really low-run input BPC you can end up with a base runs of 0, which then must get rounded up to 1 if there's no decrypter to add to the 0.
I saw some CCC II BPCs on contract in Jita on the real server that also seemed to show the same sort of behavior.
Yes, I'm seeing the same with the hulk tests that I'm doing.
Of course, it could also be x4, but I doubt thats the case. ----------------------------------------------- Adm Vladimir Tinakin CFO Hadean Drive Yards |
Little Girly
|
Posted - 2007.03.22 01:35:00 -
[28]
Used max run covetor bpcs, stolen formulas, covetor ship, skills all level 5 - i only get 25% success rate making hulks.
|
Vladimir Tinakin
Caldari Hadean Drive Yards Archaean Cooperative
|
Posted - 2007.03.22 07:05:00 -
[29]
Originally by: Little Girly Used max run covetor bpcs, stolen formulas, covetor ship, skills all level 5 - i only get 25% success rate making hulks.
Over how many attempts?
The actual success rate with everything maxed--what you're using--has to be higher, otherwise ship invention is an utter joke and not the "future of T2 production in Eve." ----------------------------------------------- Adm Vladimir Tinakin CFO Hadean Drive Yards |
Umbriele
Gallente Natural Inventions
|
Posted - 2007.03.22 08:08:00 -
[30]
My last invention experience, after 1.4 patch:
10x invention jobs of cap rec II all skills lvl4 formation layout decryptor eutetic cap rec I BPC ME140 PE10 300 runs
results: 8 successfull, 2 failed, 1 was 19 runs, 1 was 11 runs, the others all 10 runs.
|
|
|
|
|
Pages: [1] 2 3 4 5 6 7 8 9 10 11 12 13 14 .. 14 :: one page |
First page | Previous page | Next page | Last page |