Mega New Elk WIP
-
I think it could be worth trying.
Right now the 10/3 placement thing is producing a situation where computer has significantly more placement than I was anticipating initially. On account of the inf x999 thing I'm pretty sure, I routinely see computer placing more than 3 hitpoints out of a frontline factory minor. Right now it's capped at 10 placement for inf but that is very high (it makes every minor behave more like a major), and also that's not applying to the other unit types, so it's a bit inconsistent currently. These are like the early v1 first edition rules right now, in practice, since there are effectively no hard limits on placement when doing that aside from the inf thing. I think I would just spam mech and artillery all day under these conditions.
Example, as Germany I can buy 14 artillery and place them all in E. Prussia. This is pretty distorting, since it allows the player/computer to drop its big TUV stacks in very forward positions, without first having to telegraph or a run a logistics line from their production cores (where their placement is more concentrated.) I think the Minors probably need to be kept at 3 placement regardless of unit type, for both player and computer, otherwise the strategic value of a forward minor is pretty pronounced. Even if it does seem to be producing a stronger play pattern from the computer right now. If we need more placement coming from a specific factory location, I would maybe just make that spot a Major, since those have 10 placement standard.
My concern with destroyable factories, just from having seen how they work in the other one, is that it's relatively easy to bomb a player off the board, before they have a chance to respond in a meaningful way. Also the player doesn't have the ability to damage/destroy their own factories, so I think that could be problematic. Still it might be worth trying just to see.
In the standard game purchasing production/factories is generally poor gameplay, except in the case of Japanese tank/mech drives, or for the more do-or-die formula from Classic where Allies would buy forward factories to try and wedge at China/India simultaneously. Or perhaps in SA factory to push tanks towards Suez in Classic, though I always found that one a bit weaksauce compared to the Brit Carrier buy hehe. In any case, point being that there was always the big danger of purchasing new production, only to have it used against one's own team if captured. Destroyable factories would remove that purchasing tension, or the double edged sword, although the same thing remains for Bases right now. AI I don't think is going to be purchasing these, since they're so expensive, but they're probably the most important unit type from a strategic standpoint.
I think as a human I could probably see purchasing a couple to make the naval or air logistics more effective, or to activate a scramble option, but AI is not really keyed up for that. I think I would not consider SBR vs bases a very efficient use of bombers, if they could instead be used to destroy factories or limit ground movement, I would just always only use them for that. A magnified bomber build would have them knocking off production pretty consistently. I mean AI will use 1 maybe 2 bombers to SBR. I would use 12 bombers and knock minor factories off the board all day, cause that screws the computer harder than just about anything else they could do. It's because of the whole repair thing post v3, where the ability to prevent enemy placement is just way more potent than some extra cost to place in PUs coming from the opponent's pile to repair. I haven't seen the AI showing that it has a reliable way overcome the factory destroyed from the air, either with AAguns, interceptor fighters, or new factory purchases to offset. Even if they are gifted them as a bonus freebie. Instead what seemed to happen is that the computer's placement would just diminish over time till eventually they get to the point where they cannot place. Since so many of their factories can be nailed by bombers at the m6/7 distance, I think many factories would get knocked off, especially the critical coastal factories, which are the only spots where ships can spawn.
In general the AI does what I would do as a player, shucking units from the existing backfield factories which are more secure, rather than purchasing forward production. The problem there is that the AI sucks at protecting it's Transport TUV, so it just can't do this anywhere nearly as effectively as I can as the human. It moves/positions it's Transports for defense as if they were v1/v2 style transports, the transports which still had a hitpoint and defense power of 1, rather than the v3 style defenseless transport which has no hitpoint.
I don't know that we can really go back in time on that one, since the Classic transport was very overpowered. There was no need to purchase any other ship type really, since a large enough transport stack could operate independently there. In the post v3 ruleset they get wiped continually trying to deplay those tactics. Sure, in the standard game, a player may risk a transport to push 2 hitpoints, and then draw enemy aircraft or ships into a predictable attack response, but the AI will do that with an entire transport stack. I'll see stuff like computer sending a single cruiser to defend like 10 transports = 70 TUV, or else I'll see some edged combats from the attacker where they will go very light into such an attack. Like where allowing a single return fire shot from the defender cruiser may allow those 10 transports to escape if they get a lucky roll, whereas a human would probably just bring a second fighter or whatever.
Computer will likewise do something like group 6 transports together, (not a bad move in classic since that's 6 hitpoints and a reliable 1 hit. Even rolled at a 1, x6 transports firing is a pretty scarry battle for the lone fighter, bomber, cruiser etc) but when transports have no hitpoint and no defense power, it's more like a death wish from enemy bombers. AI will still group or place transports pointlessly into locations where they're immediately at risk from such attacks. Sometimes computer gets lucky, on account of the turn order sequence, and a computer teammate will float in to protect them, but this is pretty inconsistent. The computer may still overwhelm itself eventually via the transport spam (even defenseless transports with heavy attrition will ultimately push the stacks into position), but it takes a long time, and a human I think would just cut them to ribbons with aircraft on approach.
There is still a divide in the player base I think, between peeps who prefer the classic transport vs the anniversary transport, but I think that ship has sorta sailed by now. In PvP the v3/v5/global defenseless transport is currently the norm.
If HardAI would pair it's defenseless transports with a destroyer/carrier combo the way human players typically do, we could probably get something cooking, but it just doesn't appear to be set up that way. Even if it has some warships that could be used to protect it's transport stacks, it will frequently send these roving instead, or leave the transports unprotected in some vulnerable spot nearby. I think also the way the computer edges it's attacks on the water is less efficient there than it would be on land, say with tank attrition.
Here the computer I think loses sight of it's fundamentals, as it will risk it's more valuable 2 hit naval unit to attrition pretty regularly. Example would be like when the computer sends a naked carrier with zero attack power and like 1 sub, to attack enemy 1 HPs worth of enemy naval TUV. Sure Computer might be more likely to win the fight in narrow terms, or eek out the pyrrhic victory there, but a human would just never be making those sorts of moves when they could preserve the carrier absorption for defense. Again they may occasionally be saved by their teammates, landing fighters or moving warships to protect a damaged carrier deck, but this is somewhat random and often down to like luck of the draw.
Other main thing that computer fails to do, would be positioning it's destroyer blockers to prevent enemy naval movement into attacks. So hardAI will not do something like put 1 destroyer between their main fleet and the enemy's main fleet, just to shut down an attack lane. They won't do this to cover their own fleet movements, or to send a blocker unit to save a teammate's fleet from being attacked, even though I would always do that in PvP. It's debatable sometimes whether that is efficient use of destroyer TUV round to round, but for a big press, if one can prevent some massive loss of TUV, or prevent an attack from occurring in the first place, that's often a pretty good use of an 8 TUV destroyer sacrificed.
The obvious solution to me would be to somehow fix the transport itself, or to simply get rid of it in favor of warships that can transport ground units directly, but that is such a departure from the standard G40 expectations now, that I think it would introduce confusion into the roster pricing balance, which is otherwise pretty much identical to OOB. I don't anticipate that we will see improvements in the HardAI behavior for it's transports, which is why I was trying to mitigate that somewhat by having more factories in place at the intermediate spots, and probably more air bases where scrambling can happen, since that does help with the defenseless transports quite a bit, provided there are fighters flying around those coastal production hotspots.
-
I think you're right. You have more experience in these matters, and know better what the AI is capable of, or lack there. If harbors and airfields are strategically placed, with a good number of fictory_minors, as long as these cannot be destroyed. Then you're looking at the cost of repair, 6PUs maximum, which the computer will do. Versus the cost of producing at 15/12 PUs, which the AI tends not to do.
So, play to the AI strength. Give it plenty of distribution points, it's just giving to have to pay to keep them open.
As to the transport AI problem. I don't know. I've tried to keep compatible with 2.5, so I've not used 'isAI'. We'll see.
Cheers...
-
[Whoops I edited this into the above by accident, meant to reply. There I managed to grab the previous edit to restore my ramblethon above hehe]
Yeah it's always such a bind. It's like I would love to see the computer doing some wild stuff like I imagine 2 players might, but then for the more solo experience I definitely would rather build to what the hardAI will actually use reasonably well. But then for transports, there's a lot to untangle there lol.
Broadly speaking I've seen a couple possible approaches that have been adopted over the years, but the issue remains sorta intractable even in the latest iterations. These are the sorts of things that I've seen adopted, more in the realm of House Rules or customized/specialized mods...
Revert to Classic transport, with the old values a0/d1 m2 with 1 hitpoint. Works but then introduces the old dilemmas for transport spam, also nerfs the air umbrella/dark skies method for covering vs enemy fleets, so sorta not the most desirable probably.
Or use the post v3 transport no hitpoint, but give it some form of defense power - whether through anti air or anti ship, restricting the attacks/defense rounds there, or finding some way to give them defensive punch, but without the fodder aspect.
Some hybrid of those, where cost/capacity there is sorta indeterminate, but I have seen everything from 7/8 standard costs up to something more like 10 depending on the power mix or whether it has a hitpoint.
Then the last would be to introduce the concept of Warships with transport capacity. Has some precedent from Revised National Advantages. Frequently by the introduction of a new infantry type. So say the Marine of some boards, or the more generic elite unit. We have graphics for this. Of these the most familiar I would say is maybe the traditional unit from Iron Blitz.
In that game they cost 5 and amphib attack 3. TripleA has a version where they cost 4 and attack amphib at 2. I think these are probably equally unfamiliar in the grand scheme, though the latter prob more familiar now. I think an easy rule to grasp would be to allow warships to transport an 1 Elite or something similar. In the Revised NA this ability was given to Destroyers for Japan's Tokyo express, otherwise haven't seen much in terms of OOB precedent, but it does have a logic to in. Old destroyers were basically the current cruiser cost 12 attack 3 possible that sort of thing. In that case the question is how much to restrict them. In the standard-ish games where these units types have made appearances there may have been limits on the total number, though that's tough for a board on this scale. Introducing a supped up elite infantry unit I think is workable.
We have graphics to give basically some version of Marine or Commando or Elite type unit to each faction, but this introduces a fair bit of crowd control since Inf types tend to proliferate. I have graphics as well for paras, but the air transport I think can be problematic for similar reasons to all the transports I guess, like the AI just sorta doesn't do what one would hope with those. Although in G40 with bases, the Para ability can be attached more to ABs. It's not something I had considered much since I thought first to try and capture the more vanilla roster, but I think there is a more general issue with transports for tripleA that it's hard to account for when dealing with the computer. Mainly cause it just seems like it's making it's moves in the Classic way, but the newer games have sorta moved on into the defenseless still somewhat uncharted waters.
I think for everything else that's going on here, an adjustment to how transports work could be advisable. Although I would try to keep it somewhat limited, just because invariably a change to one thing has the knock on effects, especially for how ships work.
There have been adjustments to the cost and behavior of both carriers and transports in the standard, so I think these units are probably the first candidates for a tweak. Example the v5 carrier, compared to the global carrier. The hardAI I think uses the older ones somewhat better, but not by much. I think the main issue for the AI is just that the carrier is such a relatively weak attack unit by itself, but the hit absorption makes the calc tell the computer it's going to be worth it in the TUV exchange immediately, and they don't consider as much the counter attack or how it will limit fighter range in subsequent turns to have the carriers getting dropped. For the standard game I just feel like carriers should probably have had their own air unit types baked into the attack/defense value but we've seen quite a few different approaches in the various HRs or across different boards. The vanilla OOB carriers work pretty well in pvp since players can adapt their behavior and how they defend, but the AI sorta whiffs it, alas hehe
-
I think probably for now I would keep the roster as is, even if the AI is somewhat deficient. I'm not sure it's worth the potential player confusion if we tweak how transports or carriers work too hardcore. Perhaps something simple like an AAfire shot for transports we could get some buy in there, but it probably needs to be pretty limited just for ease of use. Even though the computer whiffs it's transports, there is a way to compensate a bit for the challenge by giving them more cash via resource modifier. Which may end up being more satisfying/simple way to help them with general transport attrition. Although they sometimes goof and leave their damaged carriers in spots where they're likely to be sunk, they still purchase and move them around a fair bit. They often will place with just 1 fighter instead of the pair, but over time they tend to purchase quite a few so it sorta works out. I think in a PvP type contest, the push would be faster and more furious, but I don't know that the computer has to necessarily be all that effective to still provide some entertainment value for what it is. Like usually it's kinda fun to stomp and slay transports, try to pick the computer apart in that way, so maybe it doesn't matter as much. PvP vs PvE Solo type dynamic will be pretty different either way.
-
Ok I think I have a sense for how to approach it now.
I've been going back and forth on it a fair bit, and as ever I'm sorta non-committal. Or where I'll try something on a lark, just to see what happens, but then have to probably rein it back in after the fact. I do the same when playing, I'm both impulsive and likely to second guess myself, then try and achieve some sort of internal consensus I guess, or just how my brain likes to process information through the many trials I suppose hehe. I think we're on the cusp though, like it feels pretty close. When I see that the computer can stalemate itself a bit, or see how the board resets after capital collapse where those production lines tend to end up it's a pretty good proxy for the PvP endgame or stalemate I think. Computer just takes longer to get there.
I think for game resolution Capital capture is the most straightforward, it sorta supersedes everything else and is the most hardcore way to get the leg up towards game over. The first time it happens this is very powerful, but it's more the second time purse stealing that I think starts to throw a wrench. This is because then the player has to consider the risk of gifting recovered money to the enemy a second time. Or when deciding whether to restore liberated territories to their rightful teammate, where it could negatively impact their own front line placement. Example being like most of France or North Africa under USA/British control where they can drop the big ticket buys, vs liberating Paris and giving all that forward placement to a faction with less money to burn. Although here France is more powerful generally now, and goes last in standard sequence, so there is a bit of a momentum build on that. They get to place rather than the delay on place by 1 round. I mean moreso than the vanilla situation where France has a lot less cash, or where things like Objectives also enter into the equation and G goes right after interrupt the French placement after Paris is restored.
I think the best option may be to highball the number of minor factories, keep the placement on them individually lowballed (3 to the minor) and then see how impactful it is, if I add a few more in locations where they make sense just purely from the movement standpoint. I think it's easy to understand them abstractly as the endpoints for some form of logistics line, whether that's themed more like rail or just a kind of outpost. To me they can represent different things depending on the theater/location. For example, maybe Siberia and Irkutsk is meant to be Tran Siberian rail, something like Chungking is more like Burma Road, Truk is like Tokyo Express, Alaska might be like ALCAN Highway or ALSIB air lifts. The stuff in North Africa is maybe interpreted differently than these, similarly a logistics hub or an outpost but we can imagine what it represents differently, depending on whether that hub is on an island or at the end of some long road or whatever makes sense thematically. For the main industrial complexes those would be more like where the Major Factories are located, the minors more abstract than that.
I'm heading up the mountain to watch my dad's hounds this weekend, so I think will just run a bunch of games and start trying to see how the computer responds to what I'd do when controlling a single faction. I suspect something like a modifier 110-25% resource to the opposing team's AI would be pretty fun. It makes a pretty significant difference, where suddenly the big takes from wiping enemy transports are less decisive, and the player cannot delay as long with efficient trading, since computer will outpace them the longer the game goes on. I think there's a point regardless where, because of the production spread, the ascendant power will continue to ascend. So it's more like can the human battle against the clock/recurring income bonus to the machine, to some midpoint where they can crescendo into a big stomp down. At that point I think it's fun if the computer, even on the back foot, has a way to keep it's life line going. I mean if it cried uncle and quit, say in round 20 or whatever, I wouldn't want to deny the committed player that last battle to control Washington DC or Berlin or the like hehe.
Anyhow, here was my latest thought. I added several minor factories and bases and a few combat units. The opener is largely similar, but the production fronts are a bit deeper.
Here's the edit save
-
UHD WIP 1940-45 1.38.8.zip
New starting setup @Black_Elk@black_elk said in Mega New Elk WIP:
I'm heading up the mountain to watch my dad's hounds this weekend
You have fun.
Cheers...
-
@wc_sumpton Cool digs!
Just got up here, fired up the first game to poke around haha
For now I think the issue I'm running into is that the $AllPlayersNoCaps$ attachment seems to be allowing all the players to over place at their factories.
So for example, Japan can place a bunch of dudes into Peking and just slam, or Germany can drop a big stack in Algiers, making it pretty hard for Allies to snap up a foothold. Rather than just the 3 units at a minor, they might be placing like a dozen units, so that can be pretty distorting for the play pattern that emerges. Probably cause I kept going back and forth on the desire for China rules hehe, but I think it crept into just the regular purchasing too. Anyhow wasn't sure if I nixed the line if it would go haywire or spit off an error when national capitals start falling? I think for now we could prob just go back to the vanilla situation for that, now that we got China behaving more like all the other factions.
Just ordered Pizza!
-
Hope you enjoyed the pizza! I've made a few changes to the "China" style recruitment. I'm going to run more test, but I'm sending you this new change to test.
mega_new_elk_1940.xml
Changes to purchaseNoPUCheers...
-
-
Just finished a 'quick' 5 round all Hard AI (not really 'quick', but you get the idea). After which I scanned the placements. Everything seemed in order.
Hope your test run as well!
BTW good catch on that placement bug!
Cheers...
-
Nothing major:
UHD WIP 1940-45 1.38.10.zipTech Development is uncheck, can be checked in Game Options
@beelee "Welcome" message will now display if any player is not 'AI'Cheers...
-
@wc_sumpton I'm just coming up on round 5 over here too haha
This is actually quite nice! I was concerned it might be a blowout for one side or the other, but so far so smooth. The lines shaped up really well! Must have been some lightning in that bottle!
Here my first trial HardAI vs itself at the end of the 4th round, pretty solid!
2025-3-29-UHD-WIP-1940-45_France_4.tsvg
Good times!
ps. Just grabbed the new one. Will use that one for tonight!
Thanks dude -
Nice! Just managed to log back in. I tried this morning couldn't reach the forums, but it seems Roi was on it! Coming clutch haha
I did have some time to play a few games during the downtime. I was curious to see how the computer would purchase position now, and think I have a good idea for how to cinch things up. There were a few spots that surprised me, like how the Computer was using some factories. Example would be like in India how they'd place at New Delhi but then move and stack Punjab, where I just figured those units would advance towards Burma, instead they hang back. Few other situations like that across the map, where including an additional starting minor factory sorta shifted the placement logistics pretty markedly. So I think I may take this into account for the next round of tweaks. Sorta same deal for the fleets, like to fine tune any German breakouts into the North Atlantic or Axis control of the Med. Or similarly spots like New Guinea where I may have made it a bit too easy on Anzac to hold the line there. But for the most part things were feeling good.
I just played openers to round like 5-10 or thereabouts. Here's one to round 7 HardAI vs Itself, on China's turn...
2025-3-29-UHD-WIP-1940-45_China_7.tsvg
One thing I had not fully considered with the extra factories, is whether having more targets might result in a smaller faction getting bombed beyond their income to repair. I think this could be a factor for nations like China, Italy, Anzac, where their purse is smaller, so that might be something I need to crank on. Although I was pleased to see just how often the computer attempts to SBR. Good times!
Will check back in when I get home tomorrow
-
Alright I ran a couple dozen games and made about as many edits at least haha
The last one I was running felt pretty cool. I repositioned the IJN and tried to go lighter footprint with their starting aircraft so it wouldn't take as long to calc it's first turn combat move. Just did a lot of trial and error trying different combination of tuv till they started to make the moves.
Here's the Edit Mode save with starting unit tweaks, there are quite a few...
2025-3-31-UHD-WIP-1940-45_test_27.tsvg
I concentrated on Japan for the most part. I think this should produce a somewhat more consistent J1 playpattern. Has HardAI achieving lines that roughly suggest Dec 1941 to late January/Early Feb of 42 on Japan's opening turn. Goal was to have them advancing on Hong Kong, Philippines, Borneo, Guam, New Guinea, Gilberts, Wake etc. The usual suspects. It's hard to get them to make a complete push on J1, so think a 2 turn sprawls is somewhat simpler to work out.
To achieve this required several transports, mainly on account of how many different target territories there are on this board compared to a vanilla G40, and hard AI using them somewhat inefficiently or losing them to attrition, To spread around quickly Japan needs something like a dozen plus transports, but then hardAI will almost always try to hold a transport or two in reserve to make a series of noncoms, so in order for them to move out efficiently I went with 15 transports ready to roll. With this I get them to do something that felt pretty satisfying, but that is also a lot of transports for a PvP where the player is more likely to concentrate them and magnify them in consolidated attacks.
This means I needed a pretty good excuse (or a pretty high value target, something that will be consequential or annoying if the player just leaves it alone to focus soley on something else). For that I think the most consistent was the bomber or the aagun(depending on how variable the defense). AAguns alone are somewhat less interesting for the combat drama, but sometimes to just to get the HardAI to push they seemed to do the trick.
I tried to go with a bit of a balance where it wouldn't just always be a walk in type thing, but it's hard to model everything. Where the defender might have a chance to roll hot and have an impact. So out of I don't know maybe 14 times I ran it I think I saw it go down maybe twice or three times where the a bomber survived the round. Then on account of high transport attrition there can definitely be some impacts from the variability but for the most part seems to work out how I'd imagined it might. Should have several pockets, and something interesting to do in each region of the board now, emphasis more on the central pacific.
It's tricky to fully model the exact attack route for Japan, also because we tried to keep a few force groupings familiar from the G40 arrangement, I had to modify many of those, but basically where they have a cluster around Formosa and then another around Truk as sorta the main staging points in terms of transport actions, with a more limited target range up North vs USSR, since they will just go ham up there otherwise. USA is effectively in the opposite position, fewer transports at the start and having to react in both theaters. This made me really want to put down another pocket point in Central America at panama, both as a target and to give the player some flexibility in which theater to enforce, which seemed to recall the classic in my mind. I tried to fill out their eastern seaboard with a few more hitpoints so that the delay on USA entry into either theater is not as pronounced. I think goal would be a round 3 where they are making move outs.
Since I don't think I've ever really seen the computer purchase subs, I also tried to add a few more of these. It's a challenge since each addition may change the attack pattern so I've been trying to keep it focused on the first two rounds how the AI will position, but basically with Germany hopefully having a couple uboats surviving the initial fireworks.
This may change slightly based on the results of individual battles but broad strokes trying to get the computer into an alright position. Beyond the opener they may need an income modifier to be competitive for the solo, but pretty good for a start. Now the challenge is trying to anticipate what players would do with all the stuff we've laid down to get the computer to open well, but that will take a while I think haha.
I had considered using a combat move/purchase/combat phase sequence, but I do think having the purchase screen first will be helpful. And also gives the player a slight pause to survey the board after each hardAI turn, while it's making it's calcs. We can always do both or adjust later, but I found it helpful like a little breather after the pretty involved turns. Also because the income is generally higher, I think it's somewhat entertaining to have to prognosticate in advance and really puzzle over what to buy. The HardAI takes longer than it would if we went straight into the combat move and then had the purchase sandwiched between the move and the battle phase, but this is mostly a feature of the first round, which is also the most involved. We'll keep pushing it, but I think it's getting pretty close.
Quickie Screens
Example HardAI J1 combat
Then their position on USA1
There's some variation in whether they fully take FIC or which secondary spot in Celebes or Borneo they might snap up, but for the most part pretty close. I tried to tamp down the early slam into USSR, they may skirmish but just not as hardcore, since their transports and main fleets are grouped more around Truk now. It will probably need a little tuning for the Allied response USA/UK 1, but hopefully a bit more on theme
-
-
Haha right on! I got the bonus for staying up extra late
I suspect that when removing a lot of the Air around Tokyo I may have given USSR a bit too much breathing room, but it should be easier to balance now that I got a better idea of how to get that spread vs the Dutch.
New Guinea and Guadalcanal is a bit of a balancing act, if I let Japan in with too much they stomp, if I don't give them enough than Anzac can get pretty insane hehe. Not that it isn't fun to see Allies bang wild after the last few Axis monster crushes, but I think the ticket might be to have Japan break through and take Guadalcanal, but Anzac hangs onto New Guinea production so they can sorta tit for tat that way, till USA regroups.
I was curious to see whether having more of a presence in central USA would help USA to position vs the Med, and I think it sorta does the trick, though it puts a bit more focus on Sub Saharan Africa than I think is really needed. I think USA calculates that it's path of least resistance is to push up through the Free French Zone, and into the USSR backfield where it sorta slowly creeps down towards Manchuria. I think the ideal would have that remain a factor/possibility but not necessarily the go to for stackfest push. I'd like to see then in North Africa a bit sooner, but that's also a bit of a dance since Axis need a round or two before they're just getting pressed from both directions. I tried to sorta connect the Atlantic gap all the way down, so that the eye would follow and not lose track of too many subs down south. I felt like Germany just needed more to work with or the battle of the atlantic in round 2 sorta lackluster hehe.
I'll keep chipping away at it though. As I've tried to find pattern that kicks alright, just also trying to include some thematic bases where they made sense. Everything is an abstraction, but you know to sorta capture the whole AB/NB push and pull on top of the minor factory contest. Hopefully should feel G40-ish enough, even without a bunch of new bases being bought necessarily. Starting TUV will be a bit deceptive since it's more the TUV that survives into the second round, but I think production capacity is a good proxy and also like how many total naval hitpoints for Axis vs Allies, at least in the main zones of conflict need to be comparable.I'll probably have to add back in a Japanese fighter here or there to counter China/USSR potentially gunning forward, but they definitely add to the calc time for each air unit, strat bombers a bit moreso. I think it might make sense to recalibrate Manchuria and Coastal China with a bit more ground to offset what I gave to Allies, but that's sorta the pendulum back and forth each iteration I guess. Long as it gets a bit closer to the vibe into the second round I think shouldn't be too hard to carry it over the line.
Once the starting forces are sorta keyed out, we can probably rework the place a bit to beautify it and get some of the bases into spots that look a bit nicer for the quick read. Then maybe start thinking on a tech tree or token scheme, or potentially simplified objectives and things of that sort. I think since there is so much going on generally in terms of map scale, it might be helpful if doing objectives to just have it be like a single objective per nation so it's easy to track. Or perhaps something more generic related to VC control, but nothing too complex there I wouldn't think. Like having 2:1 Axis compared to Allies for objective haul might make sense, since there are more on team Allies. Right now I think it would just be to add a slight boost to pattern that already emerges, just so the computer can hit the right beats on that.
Anyhow, should be fun. I got tomorrow free so I imagine I'll just do the same thing going down the turn sequence with the rest of the gang. G1 and J1 I think are the hardest to key up, so hopefully the rest goes a bit faster with the quickie trials now that the balls rolling
-
ps. Oh cool! Using that one I was finally able to get computer Japan to go the full distance! hehe
Not bad for like only one iteration out.
Hit on an idea to add some infrastructure to the Dutch East Indies, which I think might help Japan stabilize in the central Pacific. What tends to happen is the fleets will sort break apart and strand themselves, so I think having these extra spawn point will help them from getting overwhelmed by transport clips. Sorta works the same way for both sides eventually. I think It's pretty close at any rate. For the soviet far east I added back a pair of fighters so Japan has a bit more to hold the line vs counter. Didn't seem to add too much time to the calc as they just sorta use them forward. I removed 1 USA transport from Alaska so their push over isn't as immediate and dropped one more central closer to home. Few other minor tweaks, couple tanks to see if USA would drop into Morocco a round sooner, it tends to depend on some of the early casualties and how Paris shakes out I think. Anyhow, just sorta riffing but I think it was pretty nice hit on something that quickly. Took me like an hour last night to get them on Rabaul lol. Much steadier progress this time haha
Quick Edit save with those adjustments. I'll trial that one tomorrow see how it shakes out to 10 rounds.
-
Working with @beelee, but I have made your changes:
UHD WIP 1940-45 1.38.12.zipCheers...
-
Fantastic! Looks good!
I feel asleep on the couch running computer games last night lol. I did the whole like nod and yawn just straight into the full snooze.
Running that last was pretty enjoyable. I feel like every iteration I learn a little bit more about what HardAI is trying to do for itself, or where it stalls out typically. Sometimes when I've tried this on more moderately scaled maps the puzzle is a bit of a black box. So I'll ask myself like ok 'is the reason HardAI does something consistently, because of the opposing TUV, distance from capital or production, something in the map design geometry, some other X factor, like a butterfly flapping it's wings air battle?' etc. hehe But since this map isn't tethered to an OOB and it's pretty easy to make adjustments on the fly, I've just been sorta running it all like a continuous experiment to see what can be achieved using very simple means.
One behavior which seems fairly consistent across the board and regardless of faction, is that the computer really doesn't want to run a Naval battle and then an Amphibious assault as part of the same combat move (into the same sz). So where a human player might go for broke and do a joint naval/amphib action, the computer seems to really prefer only running the naval battle or only the amphib battle. It will wait until the lanes are cleared of ships before actually positioning it's transports more foreword vs the enemy. This means that a single enemy destroyer can deter the Computer from making an otherwise very sensible attack, say vs a starting factory territory. Or they'll clear a sea lane, then wait for a teammate to link up with them. This was a little tricky, because initially I wanted to use Destroyer blockers to limit fleet movements since it's the cheapest unit that will create a hostile sea zone, but the computer will get hung up, and often fail to run amphibs in round 1 under those conditions> It really wants a clear path for it's transports.
Similarly, I've noticed that when the computer manages to stalemate itself on the water, it can then fall into a pretty insane naval arms race, where amphib and airblitz starts to take a backseat, to like mass fleet spams. This is actually pretty interesting when it happens, since we then start to see a literal ton of naval TUV entering the fray each turn, but then what happens is the computer will move it's fleet away from coastal production and then get a bit marooned.
Sometimes having a starting factory near a main warfront, ends up complicating the logistics and working against the computer's play pattern. So this can also happen when a lot of naval TUV is staring the enemy down in a hotspot, that the computer will sorta foot drag and then get snafu'd by it's own purchase placement. To me it just indicates that I may need to give the Allies more/less naval TUV based less on the production capacity balance for reinforcement, and more on direct pressure from ships and air. So like rather than shortening the reinforcement logistics in a given spot, we may need to get an earlier breakthrough, or at least where that doesn't take dozen turns to materialize.
Here the balance in the Med and Baltic is particularly narrow, mainly on account of Germany now stepping into a Vichy-esque role for the Axis ships that coalesce in the Western Med. Vichy is hard to model under these very simplistic rules, because I think the basic game has that sorta angle where it wants to make Free France the deal, as opposed to France becoming a neutral Axis aligned power, which would be more accurate. I think it still works reasonably well, but I think the clapback vs North Africa needs to be a bit more immediate, just so USA doesn't get too hamstrung or where it's no longer building out the Eastern Seaboard for it's main push. I think after G1/J1, the timeline will become more extended. So maybe round 1 covers a full year, but then each year after that the actual timeline is much more flexible and based more on what players are doing, rather than an external clock keeping time per game round. I think ideally by round 5 we start to see some big maneuvers and the lines sorta 42/43 vibes where the Allies strike back.
Should be fun!
-
heh heh you actually slept. I haven't got around to it yet
