Skill gain rates and ratio: a test

[Doer]

I have been trying out a couple of the uber scanners (ES400 and ES500) through the H&R COT rental service, and used the opportunity to study the ratio of skills gained by performing a specific profession.

The advantage of using scanning for this test is that very few other professions are overlapping, especially in actions performed during the time of the scanning. Another advantage was that i had nearly zero skill in most of the contributors at the start (exceptions: ~200 scan robot, a lot of perception, and some mechanics and electronics from crafting).

We see from Jimmy's thread that the Robot Investigator profession is comprised of these skills:

Intelligence 	 2.0
Perception 	 5.0
Probing 	 8.0
Scan Robot 	50.0
Analysis 	10.0
Computer 	 5.0
Electronics 	 5.0
Mechanics 	 3.0
Robotology 	 8.0 (unlocked at level 30)
Total sum:      96.0

Presumably Scientist makes up the final 4% contribution.

I have tracked each of these skills during the scanning sessions and then calculated the skill volume change (in tt) to compare. Here are the percentages of the total tt skill gain for the profession for each of those skills (vs the % contribution):

%tt gain       % contribution
 1st    2nd
 6.5%   6.3%  5% 
 9.1%   9.5%  8% 
57.6%  57.2% 50% 
11.4%  12.1% 10% 
 6.0%   5.5%  5% 
 5.9%   5.6%  5% 
 3.6%   3.7%  3%

I have left out Robotology and Intelligence because i don't have Robotology unlocked and there is currently no known mapping from levels of attributes to volume. The correlation isn't very good: all of the gains were a higher % of the total, but there is something to consider here: the distribution is different because of the locked skills. If i count intelligence, there are only 88% of 100% accounted for. If i don't count intelligence, there is 86% accounted for. Therefore, there should actually be 113.6% or 116.3% greater gain than the original unmodified % account for.

Let's see how the numbers look when we adjust for these two possibilities:

gain%(1|2) 88%  86%   % skill contribution
 6.5  6.3  5.68 5.82  5 -- (Perc was also gained from hunting)
 9.1  9.5  9.09 9.30  8
57.6 57.2 56.8 58.2  50
11.4 12.1 11.4 11.6  10
 6.0  5.5  5.68 5.82  5
 5.9  5.6  5.68 5.82  5
 3.6  3.7  3.41 3.49  3

It's impossible to state whether attributes represent a normal portion of the tt skills awarded or not. However, it is interesting to see that the skills are gained in very nearly the proportions that they contribute.

I will update this thread with my results from the next scanning session to see if there's an obvious slowdown in the gains of the highest level skills compared to the lower-leveled skills.

Edit: After the second session there's no sign that the skill gains slow down according to their own levels.

 

[Jimmy B]

gain%  86% skill contribution
 6.5   5.59  5 -- (Perc was also gained somewhat from hunting)
 9.1   7.83  8
57.6   49.54 50
11.4   9.80  10
 6.0   5.16  5
 5.9   5.07  5
 3.6   3.10  3

Hmm, seems to me if you multiply the numbers by 0.86 you get very close to the percentages on the right, given perc was gained a bit during hunting. I would say its pretty good evidence that you gain skills in proportion to the amount they contribute to the profession. Obviously in a short run the randomness skews it a bit but I bet if you'd done it 10 times longer it'd have been very close indeed.

Nice test Doer.

 

[club]

ty very much for that study very useful

+rep









__________________________________________________ ____________

EFD is welcome , donate please fill me with EFD

 

[Doer]

I added the percentages from the second scanning session to the first post. There were no significant changes, and the results seem to support the idea that skills don't slow down at higher levels in gains (just in levels due to the increase in volume/level), because the relative % gain stayed constant. Obviously this isn't conclusive; all skills involved are below 4k still.

There are some more interesting things to pull out from my results but i'm not going to crunch any more or report it now.

 

[Xen]

Same thing happens with defense skills. So same thing probably happens with all activities.

 

[Xen]

BTW, I'd be interested if you could run a comparison between the 400/500 and maybe crafted scanner with higher decay. Skills gained vs. scanner decay.

 

[KP708]

I would say its pretty good evidence that you gain skills in proportion to the amount they contribute to the profession.

Definitely a nice project, and useful results.

But...

According to the SG profession details, contributions are:

SG 80%
conc 10%
some others 10%

Conc only comes from sweat gathering and other MF activities. So if you never use an MF chip, all your conc comes from sweat gathering.

This would imply, based on the above figures, a ratio of 8:1 for SG:conc

But it's more like 4:1. (I have around 3.8 times as much SG as conc and never used an MF chip)

So I'm not sure that the skill gain rule shown by this study works for every single prof.

Am I missing something?

 

[Doer]

Definitely a nice project, and useful results.

But...

According to the SG profession details, contributions are:

SG 80%
conc 10%
some others 10%

Conc only comes from sweat gathering and other MF activities. So if you never use an MF chip, all your conc comes from sweat gathering.

This would imply, based on the above figures, a ratio of 8:1 for SG:conc

But it's more like 4:1. (I have around 3.8 times as much SG as conc and never used an MF chip)

So I'm not sure that the skill gain rule shown by this study works for every single prof.

Am I missing something?

Assuming your sig is reasonably up-to-date you have over 500 SG and less than 200 Concentration. In terms of skill volume that means you have less than 0.28 PED tt Concentration and over 1.20 PED tt SG. Besides being almost too small a volume difference for accurate numbers, in the worst case possibility is a ratio of over 4.28:1 so i'm assuming you're making the mistake of comparing levels instead of skill volume. Levels are compressed into more and more volume and the mapping of level to volume skills looks like this.

 

[KP708]

Assuming your sig is reasonably up-to-date you have over 500 SG and less than 200 Concentration. In terms of skill volume that means you have less than 0.28 PED tt Concentration and over 1.20 PED tt SG. Besides being almost too small a volume difference for accurate numbers, in the worst case possibility is a ratio of over 4.28:1 so i'm assuming you're making the mistake of comparing levels instead of skill volume. Levels are compressed into more and more volume and the mapping of level to volume skills looks like this.

Yup you're absolutely right, and thanks for pointing out where I went wrong. I wondered what 'skill volume' was. Now I know. Hehe

And you'll be pleased to know it works pretty close to perfectly

conc=190
SG= about 730 but I can't remember exactly now and can't login from work

Gives 2.13 / 0.26 = 8.19 which is pretty close to 8:1

In fact, it's such a good match, I get that feeling...you know the one when you have a theory and the results bear it out and you say wow maybe this is actually right.

So spot on and well done guys

 

[Witte]

A subject brought up before, how about rifle vs LWT? I heard people gain rifle skill faster, while in total LWT should gain faster. Or does damage professions gain much slower then hit profession?

 

[KP708]

in my experience LWT is way faster than rifle. Something like 30-40% more skills gained at LWT. But when you translate into SI value, LWT is about 80% more than rifle for me

 

[Witte]

in my experience LWT is way faster than rifle. Something like 30-40% more skills gained at LWT. But when you translate into SI value, LWT is about 80% more than rifle for me

My experience too, but I can remeber people who only skilled laser rifle, and never chipped, reported they have higher rifle skill. But maybe its my bad memory .

 

[silvernight]

My experience too, but I can remeber people who only skilled laser rifle, and never chipped, reported they have higher rifle skill. But maybe its my bad memory .

Thats not true, I haven't chipped anythiong but Ana. And my Rifle was alot lower then my lwt when I use my first hg. So Lwt IS faster gained then rifle. =)

 

[Doer]

A subject brought up before, how about rifle vs LWT? I heard people gain rifle skill faster, while in total LWT should gain faster. Or does damage professions gain much slower then hit profession?

LWT is tricky because it does seem that dmg and hit professions skill at different speeds. It would be interesting to compare some skill sets of people who only hunted with laser weapons and only with rifle (or only with handgun) to see if we can figure out the relative skill speed of dmg and hit.

 

[Jimmy B]

It could be the case that the skills in the (Dmg) profession rise faster in proportion to their percentages because the high-level unlocks (Wounding, KS ~ 21%) contribute more than those for the (Hit) profession (CS, Commando ~ 8%).

Maybe you get the same amount of gain in ESI tt roughly for each profession, and this is spread over the skills you can gain it in. This would mean the damage skills get a boost over the (Hit) skills.

That doesn't necessarily disagree with the fact most people observe their (Dmg) profession to increase slower than their (Hit) profession, due to skill slowdown.

 

[Agar]

A subject brought up before, how about rifle vs LWT? I heard people gain rifle skill faster, while in total LWT should gain faster. Or does damage professions gain much slower then hit profession?

I don't know everything about the recent VU's and changes in the skill system yet, but until november 2006 it was definitely like that.

Until I took a break, I skilled only with Opalo, Breers 1&2 and J. MK II. Besides those, only Willard Heatray and TT Longblade. So all my LWT were gained by laser carbines.
One or two months ago, as I logged in again after my break, the new profession standing overview showed me about lvl 15 in Laser Sniper (Hit) and only lvl 12 in Ranged Laser (Dmg). Rifle skill was ~2100, and LWT was ~2600.

With a view on unlocks of new skills, this makes a bit sense. For the typical hunter:
- Marksmanship at 10 (Hit)
- Ranged Damage A. at 18 (Dmg)
- Coolness at 40 (Hit)
- Wounding at 45 (Dmg)
...
Until the RDA unlock, dmg gains are slower than hit gains. After RDA, I don't know. And I guess, it will be hard to find someone who skilled up naturally between RDA and Coolness and could tell us...

First a bit hit bonus, resulting in a higher higher hit profession standing, then a bit dmg bonus, then nothing for a long time...
I don't know if the VU's changed something related to rifle vs. LWT or dmg vs. hit profession as well, but if the skill unlock levels are still the same, nothing has been changed to this.

Well, I never chipped and won't do it. I will reach RDA soon, and then nothing for years...
And now I'm crawling back killing some Snables not to loose to much

edit:

hmm, I write to slow...

LWT is tricky because it does seem that dmg and hit professions skill at different speeds. It would be interesting to compare some skill sets of people who only hunted with laser weapons and only with rifle (or only with handgun) to see if we can figure out the relative skill speed of dmg and hit.

Well, read my post
Somewhere I still have a note of all my old skills without disturbing BLP and plasma skills... *searching*

 

[Oleg]

I don't know if laser works out the same (I would guess yes) but for BLP the hit professions certainly go up at a faster rate than the damage up to RDA unlock. When I unlocked RDA (Level 18 BLP Ranged Damage) my BLP Pistoleer Hit was around about 23, if memory serves. Now with RDA it has caught up a lot, it's now about 2.5 levels instead of 5, that's with me just about to hit Serendipity (hopefully tonight!) - so that's 30 Hit, 27.5 Damage.

The Serendipity should, in theory, help the Hit to kick back in again, but I don't know yet how quickly it comes so don't know what the overall effect will be.

All of this is natural but I have done other skilling with laser rifles (<2k rifle) and melee (powerfist about 1k and everything else lower) and a small amount on plasma pistols.

Good research Doer, +rep.

 

[Jimmy B]

Until the RDA unlock, dmg gains are slower than hit gains. After RDA, I don't know. And I guess, it will be hard to find someone who skilled up naturally between RDA and Coolness and could tell us...

(Dmg) is still slower. I've skilled up naturally and am a few weeks away from Wounding. It is so much slower that you can skill both handgun and rifle and have Pistoleer (Hit) and Sniper (Hit) both higher than Ranged Laser (Dmg).

 

[Doer]

Here's something else quite interesting i noticed. By considering the volume of the other skills gained, i can estimate the volume of the attribute Intelligence that would have been gained, were attributes weighted the same. That comes out to almost exactly 1 ped tt volume of Intelligence i gained over the entire period of scanning.

Now, there is probably some equivalent mapping from attribute level-> skill level, some factor, that when the multiplying the start and end levels of the attribute, will give 1 ped tt volume over the equivalent range of skill level increase. I'm not going to spell it out, but here's the program, along with its output, that i used to investigate that possibility. Keep in mind the amount attributes contribute to health and that there is a certain amount of error in the numbers due to rounding &tc, and there are some possible tentative conclusions to be drawn here:

for factor in range(10, 1000):
	d_tt = tt(int_end*factor/10.0)-tt(int_start*factor/10.0)
	if d_tt > 0.99 and d_tt < 1.01:
		print factor/10.0

Which outputs:
41.3
46.7
46.8
48.8
48.9

 

[Coop_]

(Dmg) is still slower. I've skilled up naturally and am a few weeks away from Wounding. It is so much slower that you can skill both handgun and rifle and have Pistoleer (Hit) and Sniper (Hit) both higher than Ranged Laser (Dmg).

Do the skills come slower, or just the professional levels? Since so much of the damage profession is locked at the start, an equal tt in skills gained would be jammed into fewer skill buckets, thus reaching wider sections of those buckets. Then those buckets are of couse balanced with a very big fat 0 for all the locked skills.

 

[Coop_]

I'm not going to spell it out, but here's the program, along with its output, that i used to investigate that possibility. Keep in mind the amount attributes contribute to health and that there is a certain amount of error in the numbers due to rounding &tc, and there are some possible tentative conclusions to be drawn here:

Oh please do spell it out for the dense among us

 

[CycoKick]

well
I went from 1998 rifle to 2168
and from 0 ltw to 814 very recently

Dont know which is more though but i think ltw is faster

 

[Doer]

Oh please do spell it out for the dense among us

Nothing particularly useful. Just that it seems that the factor for mapping between attribute level and equivalent skill level is around 40-50 by my skilling calculations. Now, if we consider that 40 also happens to be the levels of an attribute needed for a complete hp, and that if we multiply 40 by 40 we get 1600, which also happens to be a very significant number when it comes to health gains, the rather incredible coincidence it all adds up to leads me to believe that 40 is indeed the factor used internally. It also leads extra support to MGMighty's theory (which has already all but proved itself, anyway) and ties in the attribute value very well.

I never was very good at waiting to let people figure things out for themselves.

Here's a chart that represents attribute level in equivalent effort by mapping it to the skill level->tt function with a factor of 40.

 

[Witte]

It was my understanding that attributes are a factor 20 more significant then normal skills

 

[Doer]

It was my understanding that attributes are a factor 20 more significant then normal skills

That seems to be the case for their effective role in Professional Standings calculations, yes.

When considering their role in health, there seems to be a factor of 40 difference, and that seems to hold for calculating the effective volume, too. There's only circumstantial evidence, but it seems to make sense.

I'm pondering the 20 vs 40 roles, though. It doesn't seem likely that my estimation of "equivalent tt" gain in Int could be far enough off to fit 20 as the factor, but there may be some things i'm overlooking to rationalize it.

 

[Witte]

That seems to be the case for their effective role in Professional Standings calculations, yes.

When considering their role in health, there seems to be a factor of 40 difference, and that seems to hold for calculating the effective volume, too. There's only circumstantial evidence, but it seems to make sense.

I'm pondering the 20 vs 40 roles, though. It doesn't seem likely that my estimation of "equivalent tt" gain in Int could be far enough off to fit 20 as the factor, but there may be some things i'm overlooking to rationalize it.

I dont realy see the relation. It is simply 1 hp gain each 40 levels. Or in skill terms, 1 hp gain in 20*40 = 800 levels, aka 2hp gains in 1600 levels.

 

[Doer]

I dont realy see the relation. It is simply 1 hp gain each 40 levels. Or in skill terms, 1 hp gain in 20*40 = 800 levels, aka 2hp gains in 1600 levels.

Yes, that's certainly a possibility. This came up as a result of the calculations i did from the original topic of this thread, and those calculations point to 40 as being the factor between attributes and skills. It would be nice to see someone repeat this type of experiment, especially with professions that use other attributes, to see whether the ~40 is repeatable.

 

[Jimmy B]

Do the skills come slower, or just the professional levels? Since so much of the damage profession is locked at the start, an equal tt in skills gained would be jammed into fewer skill buckets, thus reaching wider sections of those buckets. Then those buckets are of couse balanced with a very big fat 0 for all the locked skills.

The (Dmg) skills come faster imo, or at least my HG+Rifle<LWT in ESI-worth. And that's ignoring Plasma&BLP too.

The pro-levels come slower I guess because there's loads of it still locked up.

 

[Jimmy B]

I'm pondering the 20 vs 40 roles, though. It doesn't seem likely that my estimation of "equivalent tt" gain in Int could be far enough off to fit 20 as the factor, but there may be some things i'm overlooking to rationalize it.

Forgive me for suggesting the possibility Doer but I can't be sure:

Intelligence contributes 2%.

 

[Witte]

Yes, that's certainly a possibility. This came up as a result of the calculations i did from the original topic of this thread, and those calculations point to 40 as being the factor between attributes and skills. It would be nice to see someone repeat this type of experiment, especially with professions that use other attributes, to see whether the ~40 is repeatable.

How did you come to the conclusion that intell should have gained 1PED?