pointers on influence calculation with negative influence actions

[goemon]

as we do know from positive influence actions:
- a faction of higher influence gains less from an action, than a faction of lower influence from the same action.

e.g.
Faction A:
tick 0 10%
redeeming 2 mio bounties.
tick 1: Faction A at 18,4% (+8,4%)

Faction B:
tick 0 72,2%
redeeming 2 mio bounties
tick 1: Faction B at 74,8% (+2,6%)


now - how is that for negative influence actions? any informed guesses, observations, tests?

does a faction of higher influence looses less from same negative influence actions, than a faction of lower influence?
or
does a faction of lower influence looses less from same negative influence actions, than a faction of higher influence?

 

[Jmanis]

As far as I'm aware, it's the normal normalising rules, only in reverse. That is, it takes away rather than adding, which is why combining positive actions in support of your allies plus negative actions against the target is so effective.

In the experiments I've done, the maths checks out, but there's a few things that aren't clear.

• is the positive bucket applied first, then the negative, or vice versa, or are they both applied at the same time? This will have an impact on edge cases like 1% influence etc
• is there a minimum threshold for when normalising influence that can't be moved past? You could end up in a weird state if all influence positions were moved to 0
• do negative and positive effects for the same faction increase some raw, "net change" bucket? I've not seen evidence of this, but it could prevent runaway influence loss.

EDIT:
Running the math as well... it would suggest there's more negative effect against low influence factions than there is against high influence factions. Thing is, that's actually what I've seen... and there's plenty of factions which get nailed to 1%... but it's much harder for factions to stay high up.

 

[DragoCubX]

• do negative and positive effects for the same faction increase some raw, "net change" bucket? I've not seen evidence of this, but it could prevent runaway influence loss.

This is far from a scientific test I've done, and just a general observation, but it would seem to me that the positive and negative actions have separate buckets. The observation was:
To begin, I did only positive work for all factions in a system until I reached a point where they would all "balance" each other out to the degree of only causing 0.3% influence changes or less for all factions. Then, I did that exact same amount of positive work for all factions again, but added in a tiny amount of negative actions for one specific faction. The next tick, that faction absolutely tanked by an absurd amount that should be impossible if it was a combined "raw" bucket for net change for that faction, because the negative work done was only worth about 1/30th of the positive work done for them.

At least, that was my conclusion. Maybe my thought process was garbage in this and the conclusion should be different, or the test had no value at all?

• is there a minimum threshold for when normalising influence that can't be moved past? You could end up in a weird state if all influence positions were moved to 0

Can you elaborate what you mean by this? I'm not sure what hypothetical scenario you're drawing up there.

 

[Screemonster]

I've noticed that negative actions against a controlling faction, those that are already ante'ing up the most influence each day, are generally less effective than those against a faction with low influence. Conversely, running positive actions in a system will take the most influence away from the controlling faction, and very little from someone that's sitting at 3%.

The easiest analogy I can think of is that every day, every faction puts a certain amount of influence into the pot, proportional to their starting influence and any negative actions, then everyone takes out of the pot proportionally to their starting influence and any positive actions. If no actions take place, then everyone takes out exactly what they put in. In high population systems, the proportional-to-influence part has a lot more weight (and thus, more inf+ or inf- actions are needed to make a significant swing).
A faction with low starting influence can be made to shed it by carrying out negative actions against them. A faction with high starting influence is already putting a lot of points into the pot, and if you don't run positives for the other factions, they'll take the majority of those points right back out again.

 

[angusmadmac1]

You may Want to read Jane Turner BGS thoughts .

Influence caps/gains and the wine analogy

This was written in 3.2 - the basic principles seem to be consistent in 3.3, but there are differences, notably in the amount of work required to hit the caps NOTE - the Carriers update has changed the way influence is redistributed. It is no longer pro-rata to starting influence but appears to...

forums.frontier.co.uk

 

[Jmanis]

Can you elaborate what you mean by this? I'm not sure what hypothetical scenario you're drawing up there.

So, quite possibly my hypothetical is answered by your observation above... but to draw out ye olde normalisation example (in the spoiler)

...

Spoiler

we have three factions as follows, and to keep the maths easy, let's just say it's some system with a population equates to 100 "points" (i.e 1% influence = 1 point)

Faction A: 20% (20 points)
Faction B: 30% (30 points)
Faction C: 50% (50 points)
(Total 100 points)

You do 5 points of negative work against faction A, so we get:
A: 15, B:30, C:50
Normalising, that becomes:
A: 15.7%, B: 31.4%, C: 52.6%

Straightforward so far.

But if you utterly smashed all the factions to, you'd end up with
A: 1, B:1, C:1, normalising to A:33%, B:33%, C:33%

My hypothetical was "What if just two positive actions were then added, of about 10 points".... and then you'd end up with:
A:1, B:11, C:1, i.e
A:7%, B:86%, C:7%
.... considering you'd smashed B and just done a tiny piece of positive work for them, for them to jump over 50% in a single tick would be crazy.

However!

Based on what you're saying above, it sounds like positive influence is applied, then negative influence...given they're two separate buckets this kinda makes sense.

Let's go with the numbers you provided above, but let's apply some larger numbers (because who knows how big the base buckets are).

A: 200, B: 300, C: 500 (1000)
A: 20%, B: 30%, C: 50%

Now, let's say you do a roughly proportional amount of work for each faction, we could get:
A: 405, B: 595, C: 1010 (2010)
A: 20.1%, B: 29.6%, C: 50.2%
A: 201, B: 296, C:502

But then if the negative bucket is applied afterwards, separately, because the single negative action is considered in isolation, the proportion of the whole bucket that one change has is almost doubly effective, even though it's 1/30th of the work
A: 196 (-5), B: 296, C: 502
A: 19.7%, B:29.7%, C: 50.4%

So even though you did 30 positive actions for Faction A, that one negative makes it come unstuck.

Huh... that would actually explain a lot of experiences I've had... where I've smashed missions for my faction, and then failed a small proportion (roughly 2-5%) and still had influence loss the next day.

 

[DragoCubX]

So, quite possibly my hypothetical is answered by your observation above... but to draw out ye olde normalisation example (in the spoiler)

...

Spoiler

we have three factions as follows, and to keep the maths easy, let's just say it's some system with a population equates to 100 "points" (i.e 1% influence = 1 point)

Faction A: 20% (20 points)
Faction B: 30% (30 points)
Faction C: 50% (50 points)
(Total 100 points)

You do 5 points of negative work against faction A, so we get:
A: 15, B:30, C:50
Normalising, that becomes:
A: 15.7%, B: 31.4%, C: 52.6%

Straightforward so far.

But if you utterly smashed all the factions to, you'd end up with
A: 1, B:1, C:1, normalising to A:33%, B:33%, C:33%

My hypothetical was "What if just two positive actions were then added, of about 10 points".... and then you'd end up with:
A:1, B:11, C:1, i.e
A:7%, B:86%, C:7%
.... considering you'd smashed B and just done a tiny piece of positive work for them, for them to jump over 50% in a single tick would be crazy.

However!

Based on what you're saying above, it sounds like positive influence is applied, then negative influence...given they're two separate buckets this kinda makes sense.

Let's go with the numbers you provided above, but let's apply some larger numbers (because who knows how big the base buckets are).

A: 200, B: 300, C: 500 (1000)
A: 20%, B: 30%, C: 50%

Now, let's say you do a roughly proportional amount of work for each faction, we could get:
A: 405, B: 595, C: 1010 (2010)
A: 20.1%, B: 29.6%, C: 50.2%
A: 201, B: 296, C:502

But then if the negative bucket is applied afterwards, separately, because the single negative action is considered in isolation, the proportion of the whole bucket that one change has is almost doubly effective, even though it's 1/30th of the work
A: 196 (-5), B: 296, C: 502
A: 19.7%, B:29.7%, C: 50.4%

So even though you did 30 positive actions for Faction A, that one negative makes it come unstuck.

Huh... that would actually explain a lot of experiences I've had... where I've smashed missions for my faction, and then failed a small proportion (roughly 2-5%) and still had influence loss the next day.

The math you're doing for this is more than I ever have, but the numbers certainly seem to match up with my intuition. We've had similar experiences

 

[angusmadmac1]

You have to mix your missions , Inf accounts for a % of overall gains
You need to do a mix to get max increase .
Trade
Inf
BH
Exploration .
Also we have diminishing returns .
So say
1 mil BH = 5 points
2 mil BH = 7 points
100 mil BH = 12 points
200 mil BH 13 points
( Points are made up )
So if you push faction A which have no station INF is the only option
Faction B has a station so you need to do inf plus trade plus BH plus exploration.
That's why it's easy to move the smaller factions . Less work required overall

 

[goemon]

You may Want to read Jane Turner BGS thoughts .

Influence caps/gains and the wine analogy

This was written in 3.2 - the basic principles seem to be consistent in 3.3, but there are differences, notably in the amount of work required to hit the caps NOTE - the Carriers update has changed the way influence is redistributed. It is no longer pro-rata to starting influence but appears to...

forums.frontier.co.uk

i can't find anything on calculation of negative inluence in there? (beside the very general statement analogue to: losses are distributed relative to (starting inluence) - do i miss anything?

 

[goemon]

You have to mix your missions , Inf accounts for a % of overall gains
You need to do a mix to get max increase .
Trade
Inf
BH
Exploration .
Also we have diminishing returns .
So say
1 mil BH = 5 points
2 mil BH = 7 points
100 mil BH = 12 points
200 mil BH 13 points
( Points are made up )
So if you push faction A which have no station INF is the only option
Faction B has a station so you need to do inf plus trade plus BH plus exploration.
That's why it's easy to move the smaller factions . Less work required overall

as said above this thread tries to get pointers for NEGATIVE influence calculation; how postive influence is calculated is generally known to me.

 

[Jmanis]

i can't find anything on calculation of negative inluence in there? (beside the very general statement analogue to: losses are distributed relative to (starting inluence) - do i miss anything?

Yeah I had a read through and came to the same conclusion; that negative influence is just "the opposite" of positive... but as per my number-run above, if that were the case you could get some pretty wild outcomes.

That said, back when you could sell stolen mission goods to the same black market as the station's mission you stole them from, and black market sales gained you rep with the controller, I did get some pretty wild outcomes. I think I knocked one faction down by about 30-40% overnight. But try as I might with that + 1t loss trading at a station, I could never get even a very small pop system control faction down to 1% in a single tick. I could hurt them substantially, but not completely nix them.

My assumption is there must be some sort of equivalent of diminishing returns to the positive influence case. In the case of positive influence, that's simply "growing" the total bucket size to achieve that diminishing return. 10 points in a 100 point bucket is a lot, but 10 points in a 10,100 point bucket, raised by 100 of those sorts of actions, is far less effective.

It can't be a straight opposite; you shrink the point bucket enough and blam, craziness ensues. I'm wondering if there's a step I'm missing in the calculations above that is "ignorable" for the positive case, but not for the negative case... for example... instead of simply subtracting 10 points from a 100 point bucket (making it 90 for the purposes of normalising), that -10 creates a "110 point" bucket somehow, of which 10 of those points are "negative", and it's the overall proportion of negative actions against that bucket that affects the faction.

I dunno what that would look like though... could it be something like:
A: 20, B: 30, C: 50 (100 total for 100%)
Do -100 points of action against C, get
A: 20, B:30, C:50, -C:100 (200 total)
100 is 50% of the total bucket, so we get:
A: 20, B:30, C:25 (75)
Normalising, we get:
A: 26.6%, B: 40%, C: 33.3%

I mean... those numbers feel ballpark correct, but I'm just spitballing here.... a higher proportion of the loss finds itself in B's lap compared to A... needs more work...

 

[angusmadmac1]

as said above this thread tries to get pointers for NEGATIVE influence calculation; how postive influence is calculated is generally known to me.

Sorry I miss read your original post . I do apologise .