Skill Ratings & Point Value - Feedback Wanted

[Descronan]

Wow! 8 Pages of comments. I haven't been able to read through all of them so my comments may be redundant.

First, skill mostly affects offensive value, not defensive. Yes, you can evade, but that rarely gives you a significant benefit. So this issue really comes down to an executive decision if the Offensive Value should be printed on the cards. If the answer is yes, then we can accurately attack the issue (pun intended). If not, then we are forced to use the PV of the entire unit.

Regardless, the majority are pointing out the odds of hitting reduce vs. the lowest likely target on the board. However, I think you've got that backwards. We should be looking at the odds of missing in comparison. Instead of starting with skill 4, start with skill 0 and work backwards.

Also, what is the average lifespan of a mech in combat? How many shots do you expect to take before your unit is combat ineffective? My bet is 2-3 rounds.

So my ideal cost table would look something like this:

Skill   PV Percent
0   209
1   159
2   127
3   109
4   100
5   92
6   77
7   58


This assumes 3 attacks based on the percentages below:
Skill   Combined
0   27.76
1   16.65
2   8.32
3   2.77
4   0
5   -2.77
6   -8.32
7   -16.65

[Joel47]

Also, what is the average lifespan of a mech in combat? How many shots do you expect to take before your unit is combat ineffective? My bet is 2-3 rounds.

But I think that's the point some of us are making -- a unit that's harder to hit lives longer, and is therefore worth more. While that longer life is accounted for in the point system, live longer -> shoot more -> benefit more from skill. Thus, the whole unit's PV should be affected by the skill improvement. While an argument could certainly be made that the contributions are unequal, the requirement that these calculations be simple means that we can't do anything fractionally (i.e., base it off some of the defense and all the offense), but must instead use an existing number or numbers.

As a sidenote, I record the longevity of a unit in my script as battle length, and could re-do the "unit list against itself" run to generate those numbers. While I think it's no more than tangentially relevant to this discussion, if someone thinks otherwise let me know.

Speaking of graphs, today's XKCD seems relevant:

[Elmoth]

Something real simple like +TM cost per 1 level, TM+2*TM (TM*3) for an increase of 2 levels (skill 2) and TM+2*TM+TM*3 (TM*6) for an increase of 3 levels (skill 1) could work if you want something simple.

[Scotty]

But that doesn't account for offense at all, which is obviously wrong (unless TM doesn't mean 'target modifier' in which case I'm totally lost).

Skill increases only modify the offensive capability of a unit, therefore it should be based entirely on the offensive output of a unit.  You can make the argument that defense makes that last longer, but a unit has already paid for its defensive component.  Two 'Mechs that do (for arbitrary example reasons, of course) 4/4/2 damage and differ only by durability (let's call it 5/4 and 10/8 at 6"/+1 TMM for comparison) are going to experience the same relative offensive increase from a skill upgrade.  The one that's more durable will last longer but already costs more in the first place by a significant margin.  By my count, by a minimum of 14 points assuming no Specials.  A unit's defenses otherwise have no effect on its offensive PV during the PV conversion process, and we shouldn't be trying to model that now.

I re-submit and would like to see testing in the direction of the method that derives the cost of skill increases from the offensive capability of the unit.  Every other method is trying to account for too much, or significantly overvalues skill increases (and would lead us back exactly to where we are now, where extremely high skill is worse than worthless).

[Elmoth]

Quite right. 1 AM around here. Disregard my previous post :p

[Joel47]

Then I'd propose something along the lines of:
x(aS + bM + cL)
where x, a, b, and c are all constants, and S/M/L are the damages at those ranges. Set those numbers so that a small sampling of hand-selected "representative units" have the same cost increase for one point of skill as they would under the current system, provide a spreadsheet with new PVs, and I'll run it and see what it looks like.

While values for the constants are, as they say, left as an exercise for the reader  8) I'd propose* something along the lines of:
b = 1 (a baseline, M range combat dominating)
a = 0.5 (less frequent, but devastating due to its accuracy)
c = 0.2 (less frequent and less accurate, but one might argue it benefits from skill
x = Dunno, but that's where the "scale it so it's equivalent" comes in. Start with 1 and see how close it is.


* In this case, "propose" is code for "pull from sphincter."

[Scotty]

I think it might actually need to vary based on which skill increase is happening.  If testing determined that the jump from 4 -> 3 and 3 -> 2 was meaningfully close to effective, but 2 -> 1 and 1 -> 0 were demonstrably overpriced.

By default I'd use Medium range because it's far and away the most common range.  If a unit doesn't have a Medium range, it'd need to use Short of course.  If a unit doesn't have any damage then I'd fall back on its Size since most units are still capable of physical attacks of some stripe.

[nckestrel]

And an artillery unit?  Or a unit with HT?
I think that’s going to become much more complicated.

[Scotty]

Consider a unit with any ART or non-interactive offensive special (meaning, it's automatic and meaningfully impacts offense) to have that much additional damage.  Off the top of my head, that includes ART, HT, and maybe MEL/TSM if the unit's Size + MEL (+ TSM) is higher than its ranged damage.

Even if it is more complicated than I'm giving it credit for, I think it's a much better approach than trying to modify a unit's whole PV, and at the end of the day it's still pretty simple arithmetic.

[Joel47]

I think it might actually need to vary based on which skill increase is happening.

I suspect the factor there might be X^0.8 or something like that; given the reaction from the powers-that-be the last time a non-integer exponent was proposed, I think we might end up leaving out that correction. (Though if we can correct for that, I'd love to put it in.)

By default I'd use Medium range because it's far and away the most common range.  If a unit doesn't have a Medium range, it'd need to use Short of course.  If a unit doesn't have any damage then I'd fall back on its Size since most units are still capable of physical attacks of some stripe.

That's where simplicity is the enemy of being correct, I think, as leaving out long range may well underpoint extreme skill. While I doubt it's possible to have a unit whose long range damage is greater than its medium range damage, I'd argue that there's significant benefit to having improved skill in a unit with L = M as compared to L = 0. The question is, of course, how much "incorrectness" we can get away with without materially affecting the balance. If you want to try something like "skill increases cost 3M" (triple the M value), we can try it and see what happens -- especially between units with wildly differing S or L values.

And an artillery unit?  Or a unit with HT?

Ooh, artillery -- thank you, that's an excellent reason to use the whole PV and not just a subset.

Scotty's variance suggestion, though, continues to intrigue me. Since I think we can all agree that the current values are pretty close for skill 3, and it's only at skill 1 or 0 where they're overpriced, what if we just had each increase after the first cost 1 less (minimum 1)?

[NeonKnight]

And an artillery unit?  Or a unit with HT?
I think that’s going to become much more complicated.

Shouldn't their PV's have been determined already in the actual unit cost?

I'd hate to think my Flea's SKILL POINT increase is contingent upon how much more offensive an artillery unit that is hardly used in play is costed for a skill increase.

[Scotty]

That's where simplicity is the enemy of being correct, I think, as leaving out long range may well underpoint extreme skill. While I doubt it's possible to have a unit whose long range damage is greater than its medium range damage, I'd argue that there's significant benefit to having improved skill in a unit with L = M as compared to L = 0. The question is, of course, how much "incorrectness" we can get away with without materially affecting the balance. If you want to try something like "skill increases cost 3M" (triple the M value), we can try it and see what happens -- especially between units with wildly differing S or L values.
Ooh, artillery -- thank you, that's an excellent reason to use the whole PV and not just a subset.

Scotty's variance suggestion, though, continues to intrigue me. Since I think we can all agree that the current values are pretty close for skill 3, and it's only at skill 1 or 0 where they're overpriced, what if we just had each increase after the first cost 1 less (minimum 1)?

I mentioned it way upthread (page 2 I think?) but my hypothesized method was to use whatever the longest ranged damage value was going from 2 -> 1 and 1 -> 0.  The initial thought was actually 2x Medium for 4 -> 3 and 3 -> 2, then 1x Long for the higher skills.  The multipliers I'm not sold on, but I think the concept is reasonable.

[nckestrel]

Shouldn't their PV's have been determined already in the actual unit cost?

I'd hate to think my Flea's SKILL POINT increase is contingent upon how much more offensive an artillery unit that is hardly used in play is costed for a skill increase.

I was mentioning artillery in reference to the proposal of just using the M damage value (or S if no M) for calculating PV increases.
A unit with 0 M and 0 S, but three Arrow IVs, needs a PV increase for better skill.

[NeonKnight]

I was mentioning artillery in reference to the proposal of just using the M damage value (or S if no M) for calculating PV increases.
A unit with 0 M and 0 S, but three Arrow IVs, needs a PV increase for better skill.

Maybe so, (and this is NOT directed at you personally, so please don't take it that way), but I'm looking at the comments throughout this entire thread and shaking my head, because right now, after the previous round of POINT VALUE adjustment, one would think  all the units are now balanced at a default skill of 4. Which to my thinking is why I cannot understand why now we need to think of "what if the unit is artillery, what if it's hiding in the woods, what if the mini is painted green instead of purple?" None of those, to my thinking, should matter, as they should have been accounted for already in the first re-calculation.

[sadlerbw]

Question: should there be a minimum 1PV cost per skill increase?

Some units, especially transports and some other unarmed vehicles, are so cheap and have no damage that any offensive formula is likely going to create a situation where multiple skill increases could end up costing the same. Take a look at the lowly hover APC (sensor):

http://www.masterunitlist.info/Unit/Details/3731/armored-personnel-carrier-hover-sensors

10PV, no damage in any bracket or any specials that do damage. So, there are a variety of possible calculations that could end up giving this vehicle a cost of 0PV for one or more skill increases. Regardless of the calculation, is this something you would be OK with, or do you think EVERY skill increase should cost a minimum of 1PV? Personally, I am in the ‘everything should cost at least 1PV’ camp.

[Scotty]

Maybe so, (and this is NOT directed at you personally, so please don't take it that way), but I'm looking at the comments throughout this entire thread and shaking my head, because right now, after the previous round of POINT VALUE adjustment, one would think  all the units are now balanced at a default skill of 4. Which to my thinking is why I cannot understand why now we need to think of "what if the unit is artillery, what if it's hiding in the woods, what if the mini is painted green instead of purple?" None of those, to my thinking, should matter, as they should have been accounted for already in the first re-calculation.

If we take the skill increases as a proportion of total unit cost, yes.

I'm in the middle of arguing that we shouldn't do that and should instead focus on the offensive characteristics only.  Since not all of a PV calculation is offense, this naturally lends itself to some instances of "but what about...?" that are entirely valid.

[nckestrel]

If a unit has armor, movement, etc, but not a single point of attack, how much is a higher skill actually worth?
That’s an extreme, but it illustrates the point.
I don’t disagree with the logic, but I think the effect would be too complicated (M or maybe S or maybe artillery or maybe physical attack), and that perhaps we can accept that the more extreme cases should assume their own logic (don’t make your transport skill 0) in favor of a simpler “full PV” basis for skill increases.

[Scotty]

I'm against, mostly because the "extreme cases" go both ways, especially when it comes to things that punch above their weight.  A unit with high offense and low(ish/relatively) durability is going to get significantly more out of a skill increase that's based on unit total worth than any of our metrics could possibly indicate.  It ends up with one of two outcomes: 1) costing skill too conservatively and ending back up exactly where we are now, or 2) creating a new breakpoint where there's an exploitable best-use for points.  Neither of those are good for the game.

[sadlerbw]

I re-submit and would like to see testing in the direction of the method that derives the cost of skill increases from the offensive capability of the unit.

I would submit that virtually ALL of a units PV is related to its offensive power. Imagine a large, armored mech chassis with no engine. It does no damage, cannot move or make melee attacks, but it does have an armor and structure value. We could, with the conversion rules, assign the lump a PV for just that armor and structure. However, the lump is never going to win any fight against a unit with a non-zero damage output. So, despite the fact that we calculated a PV, it is essentially worth nothing on its own because it is impossible for it to win a fight against anything; it cannot cause any damage. In a one-on-one fight, the lump’s PV is effectively zero because you cannot defend your opponent to death.

In my opinion, the only reason we assign a value to defensive stats is because they improve your ability to use offensive stats. The defensive value is nothing more than a convenient way to express how much armor and structure are going to increase the offensive ability of a unit. I feel like a unit with 3/3/3 and and 10a/5s is worth more than 3/3/3 5a/3s not because it is harder to kill, but because it keeps that 3/3/3 around longer. Basically, from my perspective, the only reason the defensive stats have any value is because they keep the offensive stats alive. They act as a multiplier for offensive value, not as a source of value on their own. So, from my perspective ALL PV is functionally about calculating offensive power, so I don’t feel units will naturally end up over-priced by simply taking some fraction of overall PV.

This is a big part of why I don’t want to look only at offensive stats.

Also, I don’t think whole number multiples of damage produces good results. I think the increases generally end up being inconsistent, and not granular enough. Some examples:

* Dire Wolf Prime: 55PV stock, 6/6/4. So, 2x medium range damage is 12PV for a total of 67. That is more than a stock Turkina E at 10/10/3. That feels high to me, and I think the fight-o-matic would bear out that the Turkina would murder the Dire Wolf, even at a 1pt skill disadvantage.

* LRM carrier vs Atlas K. LRM is 22pv 2/3/3, Atlas is 45pv 3/3/3. Should they both really go up the same amount per skill level? Their base damage is almost identical, but giving them both the same cost for skill increases seems wrong to me.

* Dasher H vs Crimson Langur Prime: Dasher is 5/0/0 25pv. It is fast, but only takes 2 points of damage to kill. The Crimson Langur is 55PV 4/4/2, and has a total of 9 health. Oh, and it is TMM3 and can jump 12” while the Dasher is TMM 4 and earthbound. So, 2x medium or 2x short if there is no medium means it is CHEAPER to up-skill the Langur than the Dasher. That can’t be right. Even if you say “don’t double the damage value used if it isn’t medium” that is still 5PV for the Dasher and 8 for the Langur. That Still seems off.

In all of these instances, I think the problem is that only looking at offense ignores all the defensive PV which will amplify, sometimes greatly, the base offensive power of a unit. If you look at the whole PV, there is no escaping how good a unit is. Doesn’t matter if it is good because it’s fast and has loads of damage, or has buckets of armor and mediocre damage. If it costs a lot of PV, it pays for the upgrade to skill, regardless of how it got to be expensive. Sure this is slightly unfair to units that have little or no offensive power, but they are generally dirt cheap in terms of PV anyway, so they wouldn’t pay much for skill upgrades even if it was based on a percentage. Honestly, anything at 10PV or less is probably over-paying even if it only costs 1PV per skill level!

[Scotty]

Anything with Hardened armor that's of appreciable size disagrees strongly with your thesis there.  Things like the Great Turtle or Grand Titan "Vengeance" to name the biggest offenders.  Making skill increases based on entire-PV is deliberately discarding not just those two units in particular, but that entire kind of unit in calculations.

Arguing that the defensive value of a unit is merely a way to express its offensive longevity flies in the face of how we actually calculate PVs, because with few exceptions the two are wholly unrelated in the PV calculation process.

I've said it at least three times before in this thread: We should not be trying to introduce new controls to a unit's total PV (which that absolutely is) in the skill value assessment.

[sadlerbw]

Still don’t agree. I feel like those sorts of units prove the point, not refute it. The only reason they cost so much is because they get to keep throwing out their piddly damage turn after turn. I mean, the whole reason the Great Turtle 1 is 66pv and the Javelin 11D is 26pv is that the great turtle gets to throw around its 3/3/0 OV1 damage for way more turns than the Javelin. Ok, so TMM is involved a little as well.

Maybe me using the word offensive isn’t helping when the PV calculation uses the same word for something specific, but it’s more likely that we just don’t agree on this. I can live with that!

So, lemme ask a different question: what, other than armor and structure, makes a significant contribution to DEFENSIVE PV? I can’t remember what else can make a large contribution besides A/S. I’m wondering, maybe it would be easier to subtract off the major defensive part of the PV to come up with an ‘offense’ number we can look at instead of abstracting all the way out to a single damage bracket and having to add a bunch of ‘if’ checks for units without medium range damage.

EDIT: Nevermind. I dug up my copy of ASC, and defense has enough modifiers and stuff that I wouldn’t call backing it out of total PV easy.

[Scotty]

But that's exactly what I'm saying: A Great Turtle gets to throw its 3/3/0 OV1 damage for way more turns than a Javelin (I do note with some amusement you picked a C3 variant which artificially pumps up its PV in a single-comparison).  The problem is that a Great Turtle that takes a skill upgrade is going to be paying significantly more than double what the Javelin is, for what is basically the same amount of extra offensive firepower.

The level of a unit's firepower is never considered during the defensive portion of a unit's PV calculation.  I see absolutely zero reason to start now.

[KCmasterpiece]

A Great Turtle gets to throw its 3/3/0 OV1 damage for way more turns than a Javelin.

This is exactly why sadlerbtw is suggesting the skill upgrade for the Great Turtle 1 should cost more than a skill upgrade for a Javelin 11D.  The outcome of a game is a combination of results from multiple turns... which is why the simulations show W/L/T and average number of turns.

[Scotty]

My counterpoint is that the Great Turtle has already paid for that ability.

The amount of PV that your A/S is worth is completely unchanged by what your offensive output is.  The comparison is just flat out not made or taken into account whatsoever.  The defensive PV component for a unit that is 20/16 A/S with 3/3/1 damage and the defensive PV component for a unit that is 20/16 A/S with 15/15/15 damage is unchanged.  Their armor and structure both cost the same regardless of their offensive output.

I've said it at least three times before in this thread: We should not be trying to introduce new controls to a unit's total PV in the skill value assessment.

[KCmasterpiece]

All it would be is keeping skill costs as a percentage of total PV.  Just increase the percentage from what it is now.  Really just need to simulate the percentage of total PV method and the percentage of offensive PV method (or factor of damage).  These seem like the two best options.

[Elmoth]

Maybe we can go for really quick and dirty and make each skill increase cost 0.1 of point value, round up or down normally. So a 26 PV unit will pay 3 points for each increase and a 54 PV unit will pay 5 points.

[sadlerbw]

(I do note with some amusement you picked a C3 variant which artificially pumps up its PV in a single-comparison).

You give me too much credit. I just picked the first great turtle in the list, and then looked for a light mech with the same damage profile. Didn't even notice the C3!

Anyway, a couple posts back, I laid out a few examples where just looking at damage values looked like it was going to produce undesirable outcomes. Go take a look at those and let me know if you think they look OK to you, or if not maybe how we should be looking at the offensive stats to produce results you would agree with. Personally, I thought all three examples seemed pretty wrong in terms of what the skill increases would cost based on damage values. As a quick reference, the comparisons were:

- Skill 3 Dire Wolf Prime vs Skill 4 Turkina E, using your 2x medium damage idea for the skill-up on the wolf. Seems too expensive to me.
- LRM Carrier vs Atlas K, should skill increases costs the same for both?
- Dasher H vs Crimson Langur Prime, should the Crimson Langur be cheaper to skill up?

Also, if you have examples where two units that have the same Skill 4 PV would end up wildly out of whack if we made them pay the same amount for skill-ups, drop them in the thread. Show me some instances where you think two units of the same or very close PV deserve to pay a different amount for skill increases. Show me some concrete examples where stuff will end up broke if we use a percentage of PV so I can see if I agree with you that whole PV isn't going to work either. Oh, and since I know tone doesn't travel well across the net I want to make sure you know I'm asking this nicely, not demanding it. Pretty please with sugar on top!

[DarkJaguar]

I still think that skill cost should be completely detached from a unit's abilities, and simply limited to a difference of 1 from a force's average skill.

Base the skill cost on say the median PV of all assault mechs increased by the percentage of effectiveness increase that the simulated scenarios are showing.  That incentivises putting high skill pilots only im the heaviest units (also the biggest targets), would lead to a very simple table (Skill 3?  that's +X, skill 2?  That's +Y, etc. etc.) . and would be easy to remember.

The pilot skill cost doesn't need to be fair or account for every ability.  It needs to be uniform and simple.

[nckestrel]

The Davion Light Guards (and the entire DCMS) objects to your requirement that the best pilots only be put in the heaviest units.
The Steiners and Smoke Jaguars heartily concur however.

[DarkJaguar]

The Davion Light Guards (and the entire DCMS) objects to your requirement that the best pilots only be put in the heaviest units.
The Steiners and Smoke Jaguars heartily concur however.

If you have two equal PV forces, and each of them takes a single skill 3 pilot, the same percentage of each force's PV is dedicated to pilots, so it doesn't really matter what unit you put it in.  It just makes the most sense to put it in the heaviest hitting units generally.