[1.12w] Aircraft has 2 attacks for no clear reason

[Elver]

I'm assuming this is a bug b/c I've never seen this happen before and the numbers are weird and not uniformly applied. It's from a slightly older patch b/c I only later noticed what had happened to the stats after the (routine, unimportant-to-me) upgrade.

I upgraded a fighter-bomber (and didn't change its weapons, nor did the Weapons Design roll improve - the only change WRT weapons was lvl 0 AP -> lvl 79 AP) and it went from 1 attacks to 2 attacks with the second model being much, much weaker against ground targets... but also flat-out twice as good against air targets. Looking at the respective design logs was not helpful; they went along perfectly normally in the second case until it got to attack number divisor, which was 2 instead of 1 for no stated reason. To further confuse things, the actual divisor used was 2.67, and no divisor was used for the air-to-air attack.

Is this a bug, or is there an obscure mechanic I've never seen before at work here?

[Saves available]

[Vic]

Yes the 2 attacks result in /2 of attack values but also *0.75. This is because of the, for better words machine-gun effect, a single bullet fire has full focus of the operator, but a lot of bullets fired have reduced focus per bullet. Anyway it is working as intended. I invite you to actually compare the 2 models effective performance against especially ground targets.

[Vic]

Aircraft that have stronger ground than air attack get 1 attack per 2500 attack points. You crossed the treshhold.

[Elver]

[Redacted my math to double-check it]

--

The other area of concern is that the Air-to-Air attack appears to be getting two attacks as well, but it doesn't get divided before being doubled. It's possible that's just the air design details page is showing this but in combat they won't actually get two attacks, but if I take the design details page at face value, crossing the 2.5k ground attack threshold has the odd effect of making much better interceptors.

[Vic]

That is a good point. They should ideally indeed have only 1 attack in combat against other aircraft. .. Will have to double check too. putting it on the list at the moment, because i don't want to dangle with these algorithms just before release.

[Elver]

I absolutely understand not wanting to mess with this with Oceania's release pending. Having said that, I double-checked my math, and splitting attacks will perform worse against well-armored targets.

--

To the best of my understanding from reading the manual and combat details (and double-checking and graphing outcomes b/c I second-guessed myself), combat works as follows (with everything unmentioned assumed to be equal):

If combat occurs between 1x2500 attack vs. 2500 HP, there is a 50% chance the defender takes a hit and 50% chance they are entirely unharmed.

If it's 1x1250 vs. 2500, there's a 25% chance of taking a hit and 75% chance of being unharmed.

If it's 2x1250 vs 2500, there's a 75% chance the first attack does nothing. If this is the case, there's then a 75% chance the second attack does nothing. 0.75*0.75 = 56.25% chance of being unharmed. There's a 37.5% chance one attack will hit, and a 6.25% chance both will.

If we add in the machine-gun factor, we end up with 2x938 vs 2500, for an 81.26% chance a given attack will miss, yielding 0.81*0.81 = ~65.6% chance of being unharmed. There's a ~30.8% chance one attack will hit, and ~3.6% chance both will.

[ETA: if it's 1x2500 vs. 1250, that's 25% of the defender being unscathed, and that stays the same if you go to 2x1250 vs. 1250. Attack needs to be at least double the defender's HP for splitting into two attacks to have the same chance of scoring at least one hit, and that's only if there's no additional machine-gunning penalty.]

The overall weird effect of this is that AP Optimization can actually make it harder for aircraft to damage well-armored targets if it pushes the pre-division attack value over 2.5k. OTOH, having AP push you over 2.5k will make your aircraft able to damage a greater number of poorly-armored targets. That outcome feels entirely reasonable to me for planes performing saturation bombing, but it's rather odd for those using precision bombs, rockets, and missiles.