British shipbuilding AI is screwed (compared to the American one)

[DicRoNero]

This might have been reported already, but since it applies to the ever-suffering navy, no attention can be excessive.

Some entry words first, as this comes directly from my current playthrough (so normal stuff, nothing artificial).

As I usually do, I go for heavy surface raiding tactics, with the same single fleet simultaneously moving convoys down and luring the enemy for good ol' Jutland-style brawl (and I've got proper naval artillery to bring dignity into it).

With Japan preoccupied with USSR, and Germany at war with me (Sweden) and Allies (fascist Italy feels there just fine, UK/USA aren't bothered at all), there are no regular Germany/USSR and Japan/USA tensions and the Allies can pretty much focus on Atlantic theatre alone, without messing in the Pacific, let alone Mediterranean.

So it was nice to see American strikeforce of a decent caliber trying to protect American shipping into Europe.

Note they use 1940 carriers, which is a welcome upgrade compared to what we used to observe all the time with full racks of obsolete ships built by the AI. After recieving some beatings, I go back to ports with one of my SHBBs needing months to repair.

Then I go further to the North, where the Americans for some reason don't mind me sinking their British fellows. Who keep reinfocing their battered convoys with the same old 0/10 or 2/8 CL/DD composition, just sending more of them as if that could help.

Now those barely scratch my paint.

Apparently, the British AI 'thinks' this way:
- look, our convoys in the Atlantics are being eradicated and we need to protect them!

- let's build more convoy-protecting DDs, then!

...
- I said BUILD MORE! Cancel the Capitals! [do note the NIC allocation]

But just how many do they think they need?!

Now what the USA builds, having half as many dockyards:

A far more balanced and I would say equally far more potent battlefleet. With the historical UK needing (or thinking they need) new battleships as late as 1942 (and thus rushing Vanguard), I don't think it's appropriate ingame to see the mighty British Empire baking tier 1/2 destroyers, when in fact capitals are required. If they don't happen to have new designs researched, I guess USA could just license them those.

 

[Evethor]

Post your SHBB design please. Im just to curious

 

[DicRoNero]

Post your SHBB design please. Im just to curious

There's a bit of RP'ing on my side (i.e. "technical specification requires the designers to come up with 30+ knots BBs, fully operational by the end of 1942"), and AA is probably excessive, but I've done some assessment and realized I win major battles through heavy firepower alone (by smashing the AI's capitals), so elimitating the screen line is not a priority (and which the AI overstacks anyway).

At the same time, anticipating these final clashes vs USA and Japan, I need as much AA as possible put into these very ships targeted by planes (even with me sometimes having the carriers of my own, like in this run).

 

[el nora]

Heavy batteries are super cost-inefficient. And the raiding fleet designer's -10% damage doesn't help (which doesn't affect heavy cruisers because reasons). But SHBB are so tanky that they will still kill an equal ic navy of heavy attack CA. If they don't bring any torps, ofc. I'm not a fan of researching line AA. So if I make SHBB, I just put AA2 in the designated slot and the two front slots, and fill the rest with DPS. That frees up some speed so you don't need to take raiding fleet designer to break 30 kn.

I would highly recommend going with the cost reduction designer. -25% cost = +33% ships. Even when you factor in the -20% heavy damage, that's still +10.8% damage output. CR designer is the best offensive ship designer in the game for light attack, and second best for heavy attack, behind only Japan's battlefleet designer. It is also the best defensive designer, the only other designer that comes close in raiding. But if you're already using TI doctrine, which you are, and have -25% visibility there, then the -10% visibility is actually -7.5% on BB because it's a multiplicative modifier. And if you manage to get concealment expert on your admiral, that's another -20% which is also multiplicative, so the effective benefit that raiding is bringing drops to only -6% visibility.

The speed boost from the raiding designer doesn't get penalized though. It's additive with the modules you fit onto the ship, so it's providing you with the full 3.4 kn you would expect it to. Which allows you to stack -21% speed from modules instead of restricting yourself to only -11% in order to remain over 30 kn. I still don't think the heavy batteries are worth it, but if you do both want to have them and to remain over 30 kn, you'd have to take the raiding designer.

 

[DicRoNero]

I would highly recommend going with the cost reduction designer. -25% cost = +33% ships. Even when you factor in the -20% heavy damage, that's still +10.8% damage output. CR designer is the best offensive ship designer in the game for light attack, and second best for heavy attack, behind only Japan's battlefleet designer. It is also the best defensive designer, the only other designer that comes close in raiding. But if you're already using TI doctrine, which you are, and have -25% visibility there, then the -10% visibility is actually -7.5% on BB because it's a multiplicative modifier. And if you manage to get concealment expert on your admiral, that's another -20% which is also multiplicative, so the effective benefit that raiding is bringing drops to only -6% visibility.

The effect of the reduced visibility is also multiplicative, so it's still beneficial to stack these reductive modifiers.

For instance, a ship of 27.2 speed and 16.2 visibility has the 0.55 hit factor for enemy heavy guns; if those are changed into 27.2 and 12.1 [concealment expert takes the lead of the fleet in fjords], the hit factor drops to 0.31. There's a world of difference over there.

Coastal Designer has a whole bunch of downsides too, aside from the obvious ones you also tie 33% more manpower, take more damage, commit a bigger fleet and thus receive bigger positioning penalties and so on.

That said, you're probably right on many points, but the thing is, given the AI and the way the game is designed overall, it's a hell lot more fun to pick some weird stuff and then develop it to its maximum, than just search for the most optimal solution right from the start (medium tanks, paradrops, w/e). I.e. I recently discovered that in my particular games, the most cost-efficient way of conducting the air war is... sending unescorted Strategic Bombers III (rushed ahead of time) against the hostile Light Fighters, and those trade in a far more favourable IC-ratio than any fighters of mine ever would, while also damaging industries and contributing to the war score.

 

[el nora]

Yes, when you stack another -25% visibility on them, that makes a big difference. And the benefit of visibility is quadratic, so -25% visibility provides -43.75% chance to be hit. But consider the relative influence that the designer applies in that case is only -4.5% visibility, from -55% to -59.5%. You went from a -79.75% chance to be hit down to -83.6%, with a relative gain of 4.82% when measured from the base hit rate. At a certain point, the extra damage output from cost reduction is more useful that the diminishing returns that you get from visibility reduction.

Naval manpower is a joke though. It's never going to be more than a couple hundred thousand. I'd rather have a 200k manpower navy than a 150k anyway because that extra 33% ships is providing 76.9% more combat ability than the smaller fleet does per Lanchester's Square Law. This ties in nicely to the positioning penalties. They're also a joke. You get -25% positioning, otherwise known as -12.5% damage, for having a fleet that is double the size of your opponents. Again, per LSL, having double the fleet size is providing x4 relative combat ability. The positioning penalty also stops growing at 3x fleet size, so once you reach that number, you can go ham with no downside.

It doesn't actually take more casualties either, because by presenting your opponent with more targets they can't focus fire down any one ship as well. It's very unlikely that a single shot will knock out one of your ships, so if we were to compute the chances that any two shots that land hit the same ship, we would get that any ship in the larger fleet is protected from getting hit twice by the mere presence of the other ships by a factor of 33% more than they are in the smaller, and from getting hit thrice by 78%, etc. That's more of a defense than the base -10% visibility provides because that only gives a flat 19% increase in survivability from the reduced hit factor. Yes, you will end up with more repairs, but repairs are cheap. You will lose fewer ships sunk, which are expensive to replace.

I agree that on the strategic level it's fun to try new things, but not on the tactical level. I will go ahistorical all day, but I'm not going to do artillery-only for "fun."

 

[sekelsenmat]

Naval manpower is a joke though. It's never going to be more than a couple hundred thousand. I'd rather have a 200k manpower navy than a 150k anyway because that extra 33% ships is providing 76.9% more combat ability than the smaller fleet does per Lanchester's Square Law. This ties in nicely to the positioning penalties. They're also a joke. You get -25% positioning, otherwise known as -12.5% damage, for having a fleet that is double the size of your opponents. Again, per LSL, having double the fleet size is providing x4 relative combat ability. The positioning penalty also stops growing at 3x fleet size, so once you reach that number, you can go ham with no downside.

That's another unrealistic thing in the naval combat system which can't be modded away.

IMHO ships above 3:1 ratio should not be allowed to join the battle (excluding carriers) since IMHO it is impossible to have like 20 ships shooting at 1 ship since they prevent accurate shots since you can't tell anymore whose splash is from who, they also block each others lines of sight and cause severe confusion about who is enemy and who is friend, as seen in the Battle of the Duisburg Convoy, when italians had a huge numeric advantage but didn't shoot because they didn't know if the enemy was the enemy or friend.

 

[el nora]

But you can kind of circumvent it by changing NDefines.NNavy.HIGHER_SHIP_RATIO_POSITIONING_PENALTY_FACTOR and NDefines.NNavy.MAX_POSITIONING_PENALTY_FROM_HIGHER_SHIP_RATIO to be more reflective of reality. Set the former to 0.5 and the latter to 1.0. Or 0.75 and 1.5 to reflect the possible boosts to positioning that a good admiral gets.

And you can also change NDefines.NNavy.DAMAGE_PENALTY_ON_MINIMUM_POSITIONING up to 1.0 as well.

 

[Evethor]

After thinking it trough I recall some notes in a dev. diary.

The UK naval AI is weighted towards producing screens to counter the submarine warfare.
The weighting is obviosly out of whack. But hey what about navy is not in this game?

 

[sekelsenmat]

After thinking it trough I recall some notes in a dev. diary.

The UK naval AI is weighted towards producing screens to counter the submarine warfare.
The weighting is obviosly out of whack. But hey what about navy is not in this game?

Although ahistorical, I can see that in the game the UK could benefit from building only DDs, CLs and escort carriers. After all they already start with more then enough heavy attack in old BBs.

Of course they need to fix the AI cancelling capital ships under construction, but besides this maybe the major issue that makes AI navies weak is that they seem to build crappy old designs? Has anyone seen the AI use a tier 3 design?

 

[kettyo]

But you can kind of circumvent it by changing NDefines.NNavy.HIGHER_SHIP_RATIO_POSITIONING_PENALTY_FACTOR and NDefines.NNavy.MAX_POSITIONING_PENALTY_FROM_HIGHER_SHIP_RATIO to be more reflective of reality. Set the former to 0.5 and the latter to 1.0. Or 0.75 and 1.5 to reflect the possible boosts to positioning that a good admiral gets.

And you can also change NDefines.NNavy.DAMAGE_PENALTY_ON_MINIMUM_POSITIONING up to 1.0 as well.

Very good but the AI would probably still bring hordes of weak screens to be obliterated even more easily with these penalties, wouldn't it?

 

[kettyo]

Of course they need to fix the AI cancelling capital ships under construction

Isn't it possible to fix this by modding?

 

[DicRoNero]

Has anyone seen the AI use a tier 3 design?

Literally the picture #1 in the OP. 1940 carriers. 1940/1936 designs - 2+2 for that task force, if you're interested. Aside from subs, those are their only 1940 ships, though (as of March 1944 for now).

 

[Paul.Ketcham]

This is a problem that's magnified quite massively when you increase the size of the enemy fleet and naval industry. I've modded the UK to have 3 times as many ships as normal, and still saw them losing literally hundreds of destroyers and cruisers on convoy escort duty from mostly the same few seazones (chokepoints like South Africa that technically aren't by their own ports, so strike forces never reach them). I leveled up Gunther Lutjens to skill 7 by sticking him with 3 panzerschiffe and 12 light cruisers near South Africa, yet was able to sink well over a hundred ships with just that singular task force. Raiding with larger capital ships results in even worse stats, with Japan and Britain being spectacularly-vulnerable to these sort of raids and thoroughly-inept at stopping them outside of coastal waters.

The production changes are a bandaid solution to their tendency to suffer massive problems due to attrition, but fail to address the underlying causes of these problems; generally if you are losing dozens of destroyers in multiple engagements, the problem is not a lack of screen production so much as poor usage, and the AI will continue to dump destroyers into losing battles for far too long before withdrawing. The AI is also far too slow to react to surface raiders in general (their failure to react is almost game-breaking, given that they will literally never change any escort composition and certain seazones will never receive strike forces regardless of losses). The best part is that sometimes the screens are needed more than capital ships anyways due to enemy screens overpowering them, but in other circumstances the capital ships are actually more impactful due to their ability to inflict bonus damage from being screened.

Carriers are also particularly useful for major fleet battles, and the AI's use of carriers tend to be to doom-stack them all in one spot (I've seen way too many 8-carrier fleets from Japan or the USA). Speaking of carriers and poor AI, these carriers often are sent out empty due to attrition rather than consolidating airwings and retreating carriers with no airwings (and some nations do this while using carriers aircraft in land-based airwings). Oftentimes the AI has many factories queued to replace these aircraft...but higher-priority production has consumed all of their factories anyways (usually 40+ factories on rifles when they have an 800,000 surplus of rifles).

One last note regarding the AI's specific failure to make capital ships, and the suggestion that the AI needs to avoid them to build more valuable units: the reason Britain and France historically built battleships was to counter those of Germany, Italy, and Japan. This could be very easily represented in-game through the espionage system, which could tell a player in theory how many dockyards are assigned to capital ships as opposed to screens, convoys, or subs (which each would require different ships to replace). This system could apply a nation's shipbuilding to be based on its main opponents, so that Britain would build differently for the historic scenario (which would ideally be balanced), as opposed to a fascist/communist USA (which ideally would be heavier in fleet units). That, and the AI ideally should be hardcoded to at-least make a couple of its historic ships (and stop scrapping nearly-built ships or historical refits of existing ships). I.e. Ark Royal, KGV and Prince of Wales for Britain, and the Dunkerques for France).

Sorry for text wall.

 

[HugsAndSnuggles]

(I've seen way too many 8-carrier fleets from Japan or the USA). Speaking of carriers and poor AI, these carriers often are sent out empty

To be fair, having those 8-carrier fleets loaded with airwings wouldn't help much, so there is some logic there

 

[sekelsenmat]

The production changes are a bandaid solution to their tendency to suffer massive problems due to attrition, but fail to address the underlying causes of these problems; generally if you are losing dozens of destroyers in multiple engagements, the problem is not a lack of screen production so much as poor usage, and the AI will continue to dump destroyers into losing battles for far too long before withdrawing. The AI is also far too slow to react to surface raiders in general (their failure to react is almost game-breaking, given that they will literally never change any escort composition and certain seazones will never receive strike forces regardless of losses). The best part is that sometimes the screens are needed more than capital ships anyways due to enemy screens overpowering them, but in other circumstances the capital ships are actually more impactful due to their ability to inflict bonus damage from being screened.

I still think that the problem would be greatly alleviated if the AI had a non-brain-dead ship design.

If instead of 8 early-trash DDs + 2 early trash CLs with 4 light attack, you would face 2 1940's CLs with 30 light attack and 8 1940's DDs with torpedoes + 1 escort carrier with 20 naval bombers (if the game would let you make a cheap escort carrier which doesn't consume BB-like oil), then I'm quite sure that a player surface raiding force would take significantly bigger loses from the encounter. The AI also should make small TAC wings and put them to cover every single sea zone where they have convoys.

The ship designer also needs to be tweaked to ban abusive and absurd ahistorical designs which a player can do. Using a 1940's CL with 2 medium turrets and 2 secondaries, so a non-abusive build, I can already 1-shot early DDs with 25 HP and regularly kill 20 AI DDs per encounter with 4 ships (1-BB + 3 CL). Using the abusive light-attack CA, I'm not sure what I could accomplish, I think that players using that can destroy the AI fleets as if they were machine gunning a Banzai Charge.

 

[kettyo]

The ship designer also needs to be tweaked to ban abusive and absurd ahistorical designs which a player can do.

As a starter a CA without armour shouldn't be allowed

 

[Paul.Ketcham]

As a starter a CA without armour shouldn't be allowed

Technically you can blame France for that, the Duquesne-class has about as little armor as possible (~60mm give-or-take). That said, the problem here comes from the whole London Naval Treaty being the only real defining difference between light and heavy cruisers (otherwise ships like the Duquesne and Pensacola-classes would be light cruisers, while ships like the Brooklyn or Town class would be heavies--on account of both armor, and volume of firepower).

The real ban should be mixed 6-and-8 inch guns, which is literal pre-dreadnought design philosophy that had been obsolete in 1906 (mainly due to shell splashes being nearly-impossible to distinguish, causing incredibly poor targeting). Particularly since basically every mixed-gun CA is a light cruiser with a "don't shoot me please" 8-inch gun mount that exists solely to classify the ship as a heavy cruiser yet wildly changes its stats and combat rules.

 

[sekelsenmat]

The real ban should be mixed 6-and-8 inch guns, which is literal pre-dreadnought design philosophy that had been obsolete in 1906 (mainly due to shell splashes being nearly-impossible to distinguish, causing incredibly poor targeting). Particularly since basically every mixed-gun CA is a light cruiser with a "don't shoot me please" 8-inch gun mount that exists solely to classify the ship as a heavy cruiser yet wildly changes its stats and combat rules.

Not only that, but also the fact that you can put who knows how many L Cruiser main turrets, or DP secondaries in a single cruiser. If 1 main turret is 6 guns, then 4 main turrets is a ship with 24 main guns? A max of 3 slots should be allowed for main guns in all hull, regardless of tech level.

Also the speeds are wrong, too high at 1940's tech, you can build a BB or Cruiser with the speed of the fastest DD ever built, maybe even higher.

 

[kettyo]

Technically you can blame France for that, the Duquesne-class has about as little armor as possible (~60mm give-or-take).

You're right but it's most weird to see a CA (armoured cruiser) with literally zero armour on it in the game. It's like calling a pure truck division an armoured division