| Author |
Topic |
Deileon
Minmatar
Sebiestor tribe
|
Posted - 2005.12.25 02:38:00 - [ 1]
Edited by: Deileon on 25/12/2005 02:42:06During my non-existent free time today, I decided to write a program which would be a virtual trainer, calculating how long it would take to train a "reasonable" skillset from scratch, which would evaluate every possible attribute combination, to see just how important it is to be "balanced." Certainly there are many skillsets and ways to play, (and I would welcome trying some others), I chose one which is pretty heavy on initial learnings, strong in combat but with also plenty of science and moderate skill in other stuff to round it out: (Captain Training: initial skills = Frig 4, Mechanic 1, Spaceship Command 2, Engineering 1, Iron Will 1) - Train non-charisma learnings to L2 (optimal ordering) - Get some essential initial skills: 2.5K Drones, 8.2K Elec/Eng/Mechanic stuff, 11.6K SpaceComm/Gunnery/Missile, 1.4K Mining, 1.4K Navigation - Train science 3/cybernetics 1 - Achieve +1 implants - Train non-charisma learnings to L4, Empathy to L2 - Second significant non-learning skillset: 90K Drones, 317K Elec/Engineering/Mech/Sci, 453K Gun/Missile/Spacecomm, 45K Industry/Mining, 45K Navigation, 16K Social - Achieve +3 implants - Finish all learnings, optimal ordering (advanced to L4 except Presence to L3) - Final skillset: 710K Drones, 3.5M Elec/Eng/Mech, 12M Gun/Missile/SpaceComm (T1 and T2 ships included), 1.5M Industry, 1M Nav, 2M Science, 600K Social, 300K Trade, 500K Leadership Attribute results? First I tried a theoretical one - what if you could distribute the initial 39 points ANY way you wanted? (Minimum of 1 point in any attribute). The results: 12I, 20P, 1C, 1W, 5M (17.174 months) Amusing.  OK, let's raise the minimum to 3: 10I, 18P, 3C, 3W, 5M (17.536 months) No real change, char and will still minimal, mem remains optimal at 5. What if we restrict minimum and maximum for each attribute to what is min/max for ALL the 8 subclasses? 12I, 14P, 3C, 4W, 6M (17.686 months) Char and will still minimum, restricting perc to 14 max throws 2 points into int and 1 into mem, oddly enough. Later I will try actual achievable attributes for each of the 8 (well 12) subclasses, and try different skillsets, such as CEO, trader, producer etc. But so far it seems to confirm that Perc and Int rule the day. Even in a learning-heavy regimen with plenty of science, 5 or 6 starting mem is all you need. Well, off to do Christmas related stuff. |
j0sephine
Caldari
Reikoku
Band of Brothers |
Posted - 2005.12.25 03:09:00 - [ 2]
"12I, 20P, 1C, 1W, 5M (17.174 months)
10I, 18P, 3C, 3W, 5M (17.536 months)
12I, 14P, 3C, 4W, 6M (17.686 months)
(..) But so far it seems to confirm that Perc and Int rule the day."
Wouldn't this sort of claim need to provide time calculations done with attribute sets that are actually 'light' on perception/intelligence, or balanced attribute-wise... so a comparison can be made?
I mean, seeing how 6 points of difference in perception (reduction by nearly 1/3rd) made whoopin' 3% difference in total training time, i.e. ~2 weeks over 1.5 year period... talk of how important perception is appears premature? ^^;; |
Deileon
Minmatar
Sebiestor tribe
|
Posted - 2005.12.25 03:38:00 - [ 3]
Edited by: Deileon on 25/12/2005 03:43:06
I don't... think you understood. I wrote a program which tries ALL POSSIBLE attribute combinations within the defined parameters, and calculates which one (picked from the many thousands possible) completes the given skill sequence in the shortest time. The 3 listed sets of attributes are the ones that produced the shortest time, given each of their respective parameters.
But you are right in saying that at least given this particular skill sequence, adjusting the attributes as shown does not have a massive impact on the total train time. But the program is always choosing the best one it can, given the restrictions.
It will be even more informative once I set up cases specifically for each subclass, and try a variety of skill sets/sequences. |
j0sephine
Caldari
Reikoku
Band of Brothers |
Posted - 2005.12.25 05:06:00 - [ 4]
"I don't... think you understood. I wrote a program which tries ALL POSSIBLE attribute combinations within the defined parameters, and calculates which one (picked from the many thousands possible) completes the given skill sequence in the shortest time. The 3 listed sets of attributes are the ones that produced the shortest time, given each of their respective parameters."
I understand that ^^ my point was more to the effect, without getting to see how large impact on skill time would it have if the character was more balanced attribute-wise, or maybe even lacking in perception/intelligence area... declaration of some particular build as 'best' has little practical meaning. If the impact of attribute combination is couple weeks over the period of 1.5 year... then it's pretty much just splitting hair ^^;; |
Deileon
Minmatar
Sebiestor tribe
|
Posted - 2005.12.25 05:49:00 - [ 5]
Well it's splitting hairs until it's enough weeks of training time to make you stop and re-evaluate the usefulness of the attributes. For instance an Intaki that tried to balance with 10I, 6C, 6P, 11M, 6W would take 19.14 months for this skillset. A pure Gallente with 8I, 8C, 13P, 4M, 6W would take 18.67 months. Some would say 4 memory is horrendous, but in fact even with a learning-heavy skillset that has 2M in science, the 4 memory one pays off in just over a year. Maybe you don't care. I just think it helps to give another angle to the importance of the attributes. I have to do more testing though. |
Ryzolette
Deer Hunters
|
Posted - 2005.12.25 16:58:00 - [ 6]
Interesting. What attribute distribution trains the worst?
What happens if you restrict spaceship command to 2 race's ships and also when ship training is restricted only to starting race's ships? |
ShadowStrike
Shadow Tech Industries
|
Posted - 2005.12.25 17:48:00 - [ 7]
Can you do it taking ALL skills in the game to lvl 5? (yeah, i know, that would take almost 20 years to do) but just to know which attributes are the best if you want to train all skills Also, do it taking only one race ships, that should reduce perception and willpower importance alot I would also be interested to know what the worst attributes would be  |
Deileon
Minmatar
Sebiestor tribe
|
Posted - 2005.12.26 04:04:00 - [ 8]
Without even trying, I am quite confident the worst would be the theoretical 1I/1P/35C/1W/1M  , but trying all possibilities again for max time... Yep, as I said, with a time of 28.63 months. My distribution was already restricted to 1 races ships. Though a variety of T1&T2ships, Bship 5, as well as high T2 gunneries, good missile skills etc. If I increased to 2 races ships that would just be a matter of increasing the perc/will and will/perc requirements in the final stage, order doesn't matter after learnings are done. Undoubtedly Perc would just skew up even more, nothing amazing there. I did try the new "broken" caldari with 9/9/3/9/9. It does provide a good time, 18.17 months. Maybe balance isn't dead after all. I mean, shifting to perc and int from that point DOES help, but I'm not sure any of the races can actually achieve a better result with the available choices. None of the ones I tried yet. I'll have to look at it thoroughly when I have time.  Originally by: Ryzolette
Interesting. What attribute distribution trains the worst?
What happens if you restrict spaceship command to 2 race's ships and also when ship training is restricted only to starting race's ships?
|
Delacho |
Posted - 2005.12.26 16:58:00 - [ 9]
you`r doing it for an all round character.
if you would it again for a fighter char or an industrial char your optimal stuff would be soo different |
Deileon
Minmatar
Sebiestor tribe
|
Posted - 2005.12.26 19:15:00 - [ 10]
Originally by: Delacho
you`r doing it for an all round character.
Yes, and perc and int are still favorable to the point of excluding everything else except for 5 or 6 memory. |
Now Mary |
Posted - 2005.12.27 00:50:00 - [ 11]
Originally by: Deileon
Originally by: Delacho
you`r doing it for an all round character.
Yes, and perc and int are still favorable to the point of excluding everything else except for 5 or 6 memory.
Uh... Willpower is the secondary attribute on all gunnery and most spaceship command skills? Infact, it's primary on the t2 ship spaceship command skills. Giving yourself 1 willpower for a fighter character seems like a very foolish thing to do. I beleive you'd want your attributes in this order (from most to least) Perception, Willpower, Intelligence, Memory, Charisma. |
Tony Fats |
Posted - 2005.12.27 05:13:00 - [ 12]
I happen to take the long view on this subject, not just whats out right now, but what will be out in the future, and also how my own interests as a player may mutate.
If you have 10 Per, adding 1 more Per is %10 of your total.
If you have 15 Per, adding 1 more Per is %6.66 of your total.
By having balanced stats, you protect yourself from the diminishing returns of further training that stat, and leave yourself open to being able to EVOLVE as a player and to evolve with the game. |
Deileon
Minmatar
Sebiestor tribe
|
Posted - 2005.12.27 06:26:00 - [ 13]
Originally by: Now Mary
Uh... Willpower is the secondary attribute on all gunnery and most spaceship command skills? Infact, it's primary on the t2 ship spaceship command skills. Giving yourself 1 willpower for a fighter character seems like a very foolish thing to do. I beleive you'd want your attributes in this order (from most to least) Perception, Willpower, Intelligence, Memory, Charisma.
I always said the results were dependant on the skillset, and qualified them as such. Nevertheless, I would never relegate int to be below will, except for the most specialized combat character. You focus on T2 ships but you are not factoring in everything else. Above mem? Probably, even though the results don't bear that out. Eventually (especially with new T2 BShips on the horizon) will should surpass mem in importance for most people, even though the results for my skillset didn't reflect that. It may take quite a few years for will to win out though. |
Tripoli
XenTech |
Posted - 2005.12.27 07:10:00 - [ 14]
Right now (and probably forever) for a pilot who intends to focus on combat...
Perception > Intelligence > Willpower > Memory > Charisma.
Period. |
Stanley VanSantz |
Posted - 2005.12.27 17:15:00 - [ 15]
I'd like to see the results when you drop science and industry ( and possibly drones ), Using the same Maximum skillpoints you used in that build, max out all gunnery and missile skills along with spaceship command in 2 races. Then nav/engineering/electronics/mechanic skills that are combat related ONLY. Of course, dont leave out the learning skills.
Also lets not forget that increasing an attribute from 8 to 9 will result in a larger time difference than increasing it from 12 to 13. Factor in implants, and I'll bet a more balanced attribute distribution would be better. |
Joerd Toastius
Octavian Vanguard |
Posted - 2005.12.27 19:57:00 - [ 16]
Originally by: Tripoli
Right now (and probably forever) for a pilot who intends to focus on combat...
Perception > Intelligence > Willpower > Memory > Charisma.
Period.
You've got solid numbers for that? Forgive me for being slightly skeptical but you're coming across as kinda dogmatic there :P |
Tripoli
XenTech |
Posted - 2005.12.28 02:49:00 - [ 17]
Originally by: Joerd Toastius
Originally by: Tripoli
Right now (and probably forever) for a pilot who intends to focus on combat...
Perception > Intelligence > Willpower > Memory > Charisma.
Period.
You've got solid numbers for that? Forgive me for being slightly skeptical but you're coming across as kinda dogmatic there :P
Haven't read too many of my posts in this forum yet, have you?  Seriously, though...Deileon's run the numbers, his math and mine have a way of agreeing. His proven conclusions confirm what many have suspected for years. |
Joerd Toastius
Octavian Vanguard |
Posted - 2005.12.28 12:38:00 - [ 18]
No, I'm fully aware of your activity here, which is why I'm being curious rather than dismissive. The numbers in post one suggest that M > W, and you're putting them the other way round. |
Tripoli
XenTech |
Posted - 2005.12.29 01:04:00 - [ 19]
Edited by: Tripoli on 29/12/2005 01:04:05It really all depends on how much emphasis you intend to put on Gunnery/Missile/Command skills versus Engineering/Electronics skills (for a combat character). It it were me, I'd weight willpower over memory any day if I were a combat character. Deileon's skill set he used in these calculations is fairly limited, so this skews the optimal attribute distribution. I believe in the long-term, Willpower will greatly outweigh Memory. Memory probably only got such a good rating in his calculation because of its use while training the learning skills. Oh Deileon......... Why don't you be a dear and do these calculations for, let's say, a capable HAC pilot using T2 guns. |
Deileon
Minmatar
Sebiestor tribe
|
Posted - 2005.12.29 03:07:00 - [ 20]
I think it just depends how long the "long run" is. And as you said, the skillset. The results did show a small boost in memory to be worth it, but it's hard to say what will happen with other types of characters. I'd have to look at how many skills are required for HAC, I can also remove the 2M in science and see what happens. First I just want to redesign it to work for each of the 12 bloodlines and each of their 3 professions, after that I can start making more varied skillsets. Just been a bit busy lately... |
Deileon
Minmatar
Sebiestor tribe
|
Posted - 2005.12.29 08:43:00 - [ 21]
Edited by: Deileon on 29/12/2005 09:47:52FINALLY coded in (and debugged) all 12 bloodlines, and put them through the wringer. Here are the results for best choice for each bloodline, in order of best to worst time (same skillset still): Ca/Achura: I=10, C=3, P=12, M=6, W=8: 17.9034 months Am/Khanid: I=9, C=5, P=13, M=4, W=8: 18.2306 months Ca/Civire: I=7, C=6, P=12, M=8, W=6: 18.3594 months Mi/Sebies: I=9, C=6, P=11, M=6, W=7: 18.4103 months Mi/Brutor: I=6, C=6, P=12, M=8, W=7: 18.4268 months Am/Amarr : I=9, C=3, P=7, M=6, W=14: 18.4586 months Am/Ni-Kun: I=8, C=8, P=13, M=6, W=4: 18.5544 months Ga/Jin-Me: I=8, C=7, P=12, M=5, W=7: 18.5587 months Ca/Deteis: I=10, C=6, P=9, M=7, W=7: 18.5763 months Ga/Gallen: I=8, C=8, P=13, M=4, W=6: 18.6715 months Mi/Vherok: I=9, C=8, P=9, M=8, W=5: 18.9234 months Ga/Intaki: I=9, C=8, P=8, M=7, W=7: 19.0239 months Some are truly fascinating. As I imagined, restricting the available choices ala profession really shakes things up. Apparently Intaki is so starved for perception that taking the +2C +2P is the best choice for an all-rounder. Vherokior puts 2 free points into WILL instead of Int. So does Deteis. Amarr takes the +4 Will profession. Brutor takes +4 Mem. So there IS balancing... to a POINT. And for each bloodline that point shifts around, sometimes unexpectedly. Tomorrow I'll screw around with other skillsets (specific suggestions would be helpful.) And for the record, I had correctly guessed the distributions for Achura, Khanid, Civire, Sebiestor, Ni-Kunni, Jin-Mei, and Gallente. Guess 7/12 isn't too bad. And also for the record... my OWN starting attribute results? Mi/Sebies: I=11, C=6, P=6, M=8, W=8: 18.9874 months  (Almost as bad as the best possible Intaki  ) |
Nova Cat |
Posted - 2005.12.29 19:54:00 - [ 22]
my 3 chars have the following starting attributes:
Me = I: 6, P: 14, C: 6, W: 9, M: 4
Alt 1 = I: 12, P: 9, C: 6, W: 5, M: 7
Alt 2 = I: 13, P: 8, C: 6, W: 5, M: 7
How do they do? |
Deileon
Minmatar
Sebiestor tribe
|
Posted - 2005.12.29 21:13:00 - [ 23]
6,14,6,9,4 time: 18.6113 months
12,9,6,5,7 time: 18.6080 months
13,8,6,5,7 time: 18.7722 months
|
Grace Hopper |
Posted - 2005.12.30 00:02:00 - [ 24]
This is really cool, and shows that they are all pretty well balanced, because of implants and diminishing returns. What if you take off the +3 implants and put +2 implants on int and per and +1 on everyhting but cha? |
Deileon
Minmatar
Sebiestor tribe
|
Posted - 2005.12.30 01:07:00 - [ 25]
First, how about examining the WORST possible starting choices (for the same skillset): WORST case scenarios: Ca/Achura: I=14, C=6, P=7, M=6, W=6: 18.9032 months Am/Khanid: I=5, C=8, P=9, M=5, W=12: 19.3232 months Am/Amarr : I=7, C=9, P=4, M=6, W=13: 19.8223 months Ca/Civire: I=5, C=11, P=9, M=4, W=10: 19.8310 months Mi/Brutor: I=4, C=10, P=9, M=4, W=12: 19.8748 months Ca/Deteis: I=7, C=9, P=5, M=13, W=5: 19.9589 months Ga/Jin-Me: I=5, C=11, P=6, M=5, W=12: 20.1008 months Mi/Sebies: I=7, C=13, P=5, M=8, W=6: 20.4589 months Ga/Gallen: I=6, C=15, P=8, M=4, W=6: 20.4758 months Am/Ni-Kun: I=5, C=15, P=7, M=6, W=6: 20.5947 months Mi/Vherok: I=7, C=12, P=4, M=13, W=3: 20.9386 months Ga/Intaki: I=8, C=13, P=3, M=9, W=6: 20.9546 months Some similarities in ordering to the first set. I love how the WORST possible Achura is still better than the BEST possible Intaki. Achura actually dumps its points all into int (the ones it can't cram into charisma). Guess that's a case where imbalancing towards int slows things down. As for your request (not sure why one would use said implants only, but): LIMITED Implant set, Best times: (+2I +2P +1W +1M) Ca/Achura: I=10, C=3, P=12, M=6, W=8: 19.2824 months Am/Khanid: I=9, C=5, P=12, M=5, W=8: 19.6543 months Ca/Civire: I=7, C=6, P=12, M=8, W=6: 19.7696 months Mi/Sebies: I=9, C=6, P=11, M=6, W=7: 19.8297 months Mi/Brutor: I=6, C=6, P=12, M=8, W=7: 19.8442 months Am/Amarr : I=9, C=3, P=7, M=6, W=14: 19.8876 months Am/Ni-Kun: I=8, C=8, P=13, M=6, W=4: 19.9984 months Ga/Jin-Me: I=8, C=7, P=12, M=5, W=7: 20.0026 months Ca/Deteis: I=10, C=6, P=9, M=7, W=7: 20.0101 months Ga/Gallen: I=8, C=8, P=13, M=4, W=6: 20.1395 months Mi/Vherok: I=9, C=8, P=9, M=8, W=5: 20.3989 months Ga/Intaki: I=9, C=8, P=8, M=7, W=7: 20.5050 months Limiting the implants as you requested produces exactly the same ordering of bloodlines as Optimal/+3, and the solitary change to starting stats is to shift one of Khanid's Perc points into Mem. Comparison to the worst-case set shows that for weaker bloodlines, starting choices matter more than implants. And vice-versa for stronger bloodlines. |
Tripoli
XenTech |
Posted - 2005.12.30 03:24:00 - [ 26]
|
Deileon
Minmatar
Sebiestor tribe
|
Posted - 2005.12.30 04:02:00 - [ 27]
No agendas. (*COUGH*Tripoli's Intaki 19.53 months*COUGH*) Just cold, hard, unfortunate facts. Just kidding.. well.. I'm sure Intaki will be on the top of a list once I run a producer or scientist skillset through. I should start delineating all of the popular fields: - Balanced/some combat focus (the set I have) - Agent Runner Combat-specialized w/ implants - PvP Combat-specialized w/o implants (after a point) - Full-time CEO - Market trader - Industrialist/Producer - Scientist (is there really any such viable thing?) Any I missed? |
Tripoli
XenTech |
Posted - 2005.12.30 05:06:00 - [ 28]
Originally by: Deileon
- Scientist (is there really any such viable thing?)
* Tripoli points at Dr Caymus. |
Stephen HB
Mystical Knights
Malum Exuro |
Posted - 2005.12.30 14:51:00 - [ 29]
Any chance of running my starting set through the wringer?
Int 6
Mem 4
Per 12
Wil 11
Cha 6
Trains gunnery like a charm, shame about the engineering. Maxed learning skills help though. (Yes, I'm aware that breakeven is still barely on the metaphorical horizon) |
Night Swordstrike
Amarr
Viziam
|
Posted - 2005.12.30 18:52:00 - [ 30]
Very fascinating insight into how the attributes work out for the various races. Makes me wonder how my Amarr Int 8, Perc 7, Cha 4, Mem 10, Will 10 would fair if run through your program. |
|
All Channels
The end of balance?
Screenshots & Art
