Sometimes I feel that modular space stations are not a good idea

[flywlyx]

It is possible to do this if the player's flying the lead ship in a fleet & has subordinates set to 'attack'. Then it's just a matter of choosing which module you want them to shoot at.

Players can only control their subordinates, which leads to low efficiency, especially with larger fleets like 5-10 destroyers.
Ideally, player should spread them evenly and order attacks on different station modules.
Coordinated attacks can achieve similar results, but players cannot select targets directly; it still relies on luck.

They are however all flown by exceptionally well trained pilots (5* for both piloting & morale).

Pilot level doesn't affect players' capital ship movement, especially when target is station, no target position update is needed.

[xrogaan]

UI is way too bad to micro-manage the fleet properly. Players could not target individual modules or ask the ships to stay in position.

It is possible to do this if the player's flying the lead ship in a fleet & has subordinates set to 'attack'. Then it's just a matter of choosing which module you want them to shoot at.
For example, first I ask them to shoot the module on the right: https://www.dropbox.com/scl/fi/u6r4iilx ... u03q2&dl=0
Then I select a different target & they shoot the module on the left: https://www.dropbox.com/scl/fi/llehonch ... zycny&dl=0

In practice though I prefer to have a carrier as fleet command & there's absolutely no way I'm flying a ship with a such an appalling turn rate (or lack of main guns) & find it works well enough just to setup my firing lines near the modules I want them to start with.

As for staying in position, mine mostly do, until it's time to give the orders to reposition. They are however all flown by exceptionally well trained pilots (5* for both piloting & morale).

How do you select a module? I can only select surface elements, if I squint hard enough.

[GCU Grey Area]

Players can only control their subordinates, which leads to low efficiency, especially with larger fleets like 5-10 destroyers.
Ideally, player should spread them evenly and order attacks on different station modules.
Coordinated attacks can achieve similar results, but players cannot select targets directly; it still relies on luck.

Targeting individual modules approach just presented for info purposes. It's possible to do but it's not something I do all that often, or would particularly recommend. My usual approach is to setup a pair of firing lines for my destroyers, one on each side of my ship. I then move those 2 groups around the station in opposite directions as the demolition progresses. This works well enough for my purposes. Typically they'll fire on whatever module's closest, then shoot the next nearest that's in line of sight, & so on. It's not target selection, however it's very predictable so functionally it might as well be - they almost always start with the modules I'd prefer them to prioritise (usually defence modules, storage if we're attacking a shipyard or wharf).

Have not used coordinated attack much aside from a brief test when it was first introduced. Did not work well with my usual fleet composition - hurled absolutely everything at the target, irrespective of whether they were suitable for the task, or had specific orders not to attack such targets (e.g. intercept), or were even specifically prohibited from taking off! This included a bunch of completely unarmed M freighters full of missile parts, which only had defend/docked orders so they were formally part of the fleet.

Pilot level doesn't affect players' capital ship movement, especially when target is station, no target position update is needed.

You mentioned having issues with ships remaining in position during station demolition. Mine move very little, except as needed to aim at a new module when their previous target has exploded. Clearly there's some difference between how your ships behave & mine. Was just speculating that pilot skill may be the variable responsible. May well be something else.

How do you select a module? I can only select surface elements, if I squint hard enough.

Sorry, should have been more explicit. To target a module open the map, select the station on the Object List then open the list of it's modules. Click on one (will be highlighted in white on the map), then press T to target it (or whatever you have set for Target Object In Map, don't recall if T is default or not). Screenshot to illustrate: https://www.dropbox.com/scl/fi/ttgk0yjo ... dji2l&dl=0

By the way, if you're playing the 7.0 beta & would prefer not to squint, can highly recommend the VE goggles for zooming in on a target. Makes selecting surface elements as targets MUCH more convenient. Controls for VE goggles can be configured in the Camera section of General Controls.

[xrogaan]

How do you select a module? I can only select surface elements, if I squint hard enough.

Sorry, should have been more explicit. To target a module open the map, select the station on the Object List then open the list of it's modules. Click on one (will be highlighted in white on the map), then press T to target it (or whatever you have set for Target Object In Map, don't recall if T is default or not). Screenshot to illustrate: https://www.dropbox.com/scl/fi/ttgk0yjo ... dji2l&dl=0

By the way, if you're playing the 7.0 beta & would prefer not to squint, can highly recommend the VE goggles for zooming in on a target. Makes selecting surface elements as targets MUCH more convenient. Controls for VE goggles can be configured in the Camera section of General Controls.

Gotcha. Not really useful when there are 30 defense bridge and you want to target specific ones. I'll keep using F3, zoom in and target a surface element. This methods is better due to the context: you can see where your ship is aiming at and what may be in the path.

[flywlyx]

Targeting individual modules approach just presented for info purposes. It's possible to do but it's not something I do all that often, or would particularly recommend. My usual approach is to setup a pair of firing lines for my destroyers, one on each side of my ship. I then move those 2 groups around the station in opposite directions as the demolition progresses. This works well enough for my purposes. Typically they'll fire on whatever module's closest, then shoot the next nearest that's in line of sight, & so on. It's not target selection, however it's very predictable so functionally it might as well be - they almost always start with the modules I'd prefer them to prioritise (usually defence modules, storage if we're attacking a shipyard or wharf).

The closest target is a very unreliable estimation, especially when the 1st target is gone.

Have not used coordinated attack much aside from a brief test when it was first introduced. Did not work well with my usual fleet composition - hurled absolutely everything at the target, irrespective of whether they were suitable for the task, or had specific orders not to attack such targets (e.g. intercept), or were even specifically prohibited from taking off! This included a bunch of completely unarmed M freighters full of missile parts, which only had defend/docked orders so they were formally part of the fleet.

Players could select specific units to send the coordinate attack order, it doesn't have to be the whole fleet.

You mentioned having issues with ships remaining in position during station demolition. Mine move very little, except as needed to aim at a new module when their previous target has exploded. Clearly there's some difference between how your ships behave & mine. Was just speculating that pilot skill may be the variable responsible. May well be something else.

There is no way the player could predict the movement of the ships, for example :https://youtu.be/MRmOkJDhtBg
In this video, it's clear that the Asgards should be maneuvering to the left or right of the station, as it's the closest target. However, they are attempting to move across the station instead.
KUDA AI has enhanced this functionality, enabling capital ships to orbit around the station to their next target. In contrast, the vanilla AI relies entirely on chance.

[GCU Grey Area]

In this video, it's clear that the Asgards should be maneuvering to the left or right of the station, as it's the closest target. However, they are attempting to move across the station instead.

Ahh - Asgards, that would explain the difference. Don't use them myself. Very sluggish ships. Just not worth the hassle as far as I'm concerned, even if they have got a big gun. Built a couple during my first Terran game & almost immediately regretted the completely free resources I'd expended to build them. Detached them from the fleet after trying them out on a couple of demolition jobs (during which their performance was remarkably unimpressive) & gave them guard duty at my HQ instead. Very much prefer destroyers for my demolition fleets. Currently using Rays, which are an absolute dream to work with; really quite nimble for an L ship, their instant travel drive is fantastic for rapid repositioning during station demolition.

[flywlyx]

Ahh - Asgards, that would explain the difference.

The behavior of capital ships remains consistent as long as they are equipped with front-mounted weapons.

[TroubledRabbit]

OK... CGU. Have you tried attacking that way two stations at once while conducting a skirmish (not even a battle, I 'don't do battles', it would cook my laptop ). This is not particularly complicated example - quite natural and obvious way to deal with scaterred enemy which has got numeric advantage in the light hardware and cannot mobilise hard one in quick fashion.

But it is a bit better now (in 7.3) so there is a progress. Though sometimes I suspect that the main problem is not that the 'AI' is primitive but, quite opposite - it is overcomplicated trying to archive too much.* Humans are not that complicated beasts to imitate, LLMs already have proven that.


* e.g. I have noticed that now autopilot is trying to avoid danger zones (in the Void) if possible. However - this results with some mess - it does not aim properly or overshoot.

[jlehtone]

RISK is perhaps closer to that (desired) level of "RTS"; one tells armies to take an area and they do (or die). We have no control on details, on tactics. We can only harness the logistics to send enough numbers to beat the odds. (I presume, can't remember rules of that game.)

What you said is a typical turn-based strategy game(TBS) rule, not Real-time strategy (RTS).

True. The point was that "destroy that station" sounds like what you order your troops to do for next "turn". The RTS, I assume (since I don't play those either), differs from turn-based, because the player can interrupt and issue new orders at any time.

Lets say that I order troops to "Move to B". There is enemy ambush on the route.
On TBS the effect of ambush is applied to the troops, isn't it?
On RTS I could pay attention and change orders when I spot first signs of danger? This should reduce damage.
However, I got the impression from earlier comments that on "real RTS" the AI would automatically do the ambush avoidance (just like I would re-order them) all on their own -- "without a mess". Is that true?

---

Another point is that if you ask some people to do something IRL, and they probably delegate the task, will it be done like you would have done it? How specific orders and how much followup is required in order to get the "like I had done it myself" result? Isn't it "realism" that AI does not do exactly what you ask them to do? (Then again, we don't seem to like realism.)

We are not admirals. We start as penniless privateers and hire anyone that is willing to our crew. Probably most of them are paupers and murderers. Not well-trained "units", like crew of NCC-1701, but perhaps more like that of Shinano. Overall, they perform better than one could expect. (Yes, I have lowered expectations.)

Should AI really act like we do? I do run into range of big turrets and get killed. I do attack when outnumbered and get killed. I do crash on big things (and get killed in X3). I do unexpected things. Should the AI really chase all those "achievements"?

[GCU Grey Area]

The behavior of capital ships remains consistent as long as they are equipped with front-mounted weapons.

That was not my experience at all. Would agree if we were just discussing destroyers. I have my favourites, but performance is broadly similar. Asgards on the other hand really suffer from their abysmal turn rates. Found them really quite awkward to use for station demolition. In general they were just a constant source of aggravation (always lagging behind & painfully slow to manoeuvre) so ditched them pretty quickly.

OK... CGU. Have you tried attacking that way two stations at once while conducting a skirmish (not even a battle, I 'don't do battles', it would cook my laptop

Don't tend to, unless the stations are VERY close together. Even then would rather focus my entire demolition fleet on one station at a time, rather than split my forces & take twice as long smashing them simultaneously. Entire point of conducting my battles & station demolition in high attention is so I can be there to participate & see all the fireworks. Would not want to miss out on all the fun by having a fleet attack a station I can't see.

[flywlyx]

True. The point was that "destroy that station" sounds like what you order your troops to do for next "turn". The RTS, I assume (since I don't play those either), differs from turn-based, because the player can interrupt and issue new orders at any time.

I agree with you. One of the issues with X4 is that it has an RTS-style user interface but lacks a game logic like typical RTS games. You cannot direct a ship to destroy a specific station module; instead, you have to order it to destroy the entire station, which is the root cause of all these problems.

Lets say that I order troops to "Move to B". There is enemy ambush on the route.
On TBS the effect of ambush is applied to the troops, isn't it?
On RTS I could pay attention and change orders when I spot first signs of danger? This should reduce damage.
However, I got the impression from earlier comments that on "real RTS" the AI would automatically do the ambush avoidance (just like I would re-order them) all on their own -- "without a mess". Is that true?
--

Another point is that if you ask some people to do something IRL, and they probably delegate the task, will it be done like you would have done it? How specific orders and how much followup is required in order to get the "like I had done it myself" result? Isn't it "realism" that AI does not do exactly what you ask them to do? (Then again, we don't seem to like realism.)

Games vary in their settings, but one commonality is that if players instruct a unit to move to point A but it moves to point B and gets destroyed, players will blame the AI. And that is exactly what happened in X4.

We are not admirals. We start as penniless privateers and hire anyone that is willing to our crew. Probably most of them are paupers and murderers. Not well-trained "units", like crew of NCC-1701, but perhaps more like that of Shinano. Overall, they perform better than one could expect. (Yes, I have lowered expectations.)

Should AI really act like we do? I do run into range of big turrets and get killed. I do attack when outnumbered and get killed. I do crash on big things (and get killed in X3). I do unexpected things. Should the AI really chase all those "achievements"?

I highly doubt that players expect the AI to perform tasks exactly as players would. For instance, I can clear a Xenon sector with a single Rattlesnake, but I don't expect the AI to replicate that level of performance.
As I mentioned, the RTS genre is quite common in gaming. X4's challenge lies in having an RTS user interface without a corresponding RTS game design.

The crucial aspect of an RTS game is having NPCs follow orders as expected by the player. If an ambush occurs and the player fails to notice it, that's on the player. However, if a unit fail to follow the order and falls into an ambush, players will blame the AI.

That was not my experience at all. Would agree if we were just discussing destroyers. I have my favourites, but performance is broadly similar. Asgards on the other hand really suffer from their abysmal turn rates. Found them really quite awkward to use for station demolition. In general they were just a constant source of aggravation (always lagging behind & painfully slow to manoeuvre) so ditched them pretty quickly.

Destroyers and battleships use the same AI script, turn speed doesn't affect the station demolition since the station doesn't move; any perceived differences are a result of your own bias.

[jlehtone]

Don't tend to, unless the stations are VERY close together. Even then would rather focus my entire demolition fleet on one station at a time, rather than split my forces & take twice as long smashing them simultaneously. Entire point of conducting my battles & station demolition in high attention is so I can be there to participate & see all the fireworks. Would not want to miss out on all the fun by having a fleet attack a station I can't see.

Not to mention that the opponent can send a "response fleet". It is easier to fly CAP around one point than two. Protecting multiple points is obviously a challenge that some want and NPC are not very likely to act on "all fronts" simultaneously.

Destroyers and battleships use the same AI script, turn speed doesn't affect the station demolition since the station doesn't move; any perceived differences are a result of your own bias.

This thread did start because AI is not smart enough to move around the station when "landscape" changes. GCU issues move orders manually. Whether auto or manual, there is movement and the desired "smart" movement is not straight line through station's zone of control.

The crucial aspect of an RTS game is having NPCs follow orders as expected by the player. If an ambush occurs and the player fails to notice it, that's on the player. However, if a unit fail to follow the order and falls into an ambush, players will blame the AI.

Thought so. What we discuss here is what we do expect, isn't it?

GCU issues "Move to A. Attack B. Move to D via C. Attack B'. ..." and claims that NPC perform as expected.
TroubledRabbit issues "Attack B" and is not happy when newly-dead B does "ambush" the ships by blocking fire to B' from position ~A.

[flywlyx]

Thought so. What we discuss here is what we do expect, isn't it?

GCU issues "Move to A. Attack B. Move to D via C. Attack B'. ..." and claims that NPC perform as expected.
TroubledRabbit issues "Attack B" and is not happy when newly-dead B does "ambush" the ships by blocking fire to B' from position ~A.

No one expects an AI to get itself killed needlessly; that's the main concern.
Whatever GCU claims is totally his personal experience without any solid proof, many players have pointed out that the AI sometimes selects a module on the far side of a station and ends up getting destroyed, and this was clearly an issue until version 6.2.

Egosoft aimed to make their AI "smart" by allowing it to automatically target different modules for destruction on a station.
However, this decision backfired as the AI ended up being dumb and often got itself destroyed in the process.
In most RTS games, the "dump" AI typically stops after completing the player's order, avoiding automated actions that could lead to its own demise.

The suggestion here is to simplify the destruction process to prevent the AI from inadvertently getting itself destroyed.

[xrogaan]

I would really love to be able to select a module that sits within my field of vision. Be it a new mode (like scanning or SETA), that allows you to move your cursor and highlight the various modules.

When attacking a station, I actively switch between 2 turrets modes: attack all enemies, and attack my target. The "my target" mode tends to select its own specific module and stop working whenever it doesn't have LoS. The "all enemies" mode will just spread firepower on however many target can be reached. Ideally, I would want to be able to manually fire my turrets onto whatever target I want hit. Same way we use main weapons, but show which turrets don't have LoS on the target I aim at (blue is OK, red won't fire). Turret based warfare is, after all, quite boring.

[GCU Grey Area]

Whatever GCU claims is totally his personal experience without any solid proof, many players have pointed out that the AI sometimes selects a module on the far side of a station and ends up getting destroyed, and this was clearly an issue until version 6.2.

That's over 10k hours of personal experience, during which my demolition fleets have destroyed innumerable stations using the tactics outlined above. It's been literal years since I last lost a destroyer to station turrets (3.0 Split game, had some issues with Rattlesnake's short range guns). Most of the time they don't get to fire even a single shot back at my ships. The main thing it's always been essential to watch out for is ships that have no viable targets in clear line of sight. That's the point at which they start trying to find a route to undamaged modules on the other side of the station. Those routes are often distinctly sub-optimal, which is precisely what my 'fly & wait, then attack' approach is intended to counter - i.e. give them a safe flightpath to a point from which they can shoot without ever going within range of station turrets.

I would really love to be able to select a module that sits within my field of vision. Be it a new mode (like scanning or SETA), that allows you to move your cursor and highlight the various modules.

When attacking a station, I actively switch between 2 turrets modes: attack all enemies, and attack my target. The "my target" mode tends to select its own specific module and stop working whenever it doesn't have LoS. The "all enemies" mode will just spread firepower on however many target can be reached. Ideally, I would want to be able to manually fire my turrets onto whatever target I want hit. Same way we use main weapons, but show which turrets don't have LoS on the target I aim at (blue is OK, red won't fire). Turret based warfare is, after all, quite boring.

The map-based module selection method in my earlier post also works for turrets which are set to 'attack my current enemy'.

Personally though don't like to go within turret range of hostile stations. Prefer to do all of my station demolition at destroyer main gun range instead. Many of my targets are Commonwealth stations which can potentially be armed with several dozen L plasma turrets. In an artillery duel between a station with 70+ L plasmas (e.g. this one) & a destroyer with substantially fewer, the station tends to win.

[flywlyx]

The main thing it's always been essential to watch out for is ships that have no viable targets in clear line of sight. That's the point at which they start trying to find a route to undamaged modules on the other side of the station.

That's exactly what people are complaining about: AI shouldn't autonomously plan routes that might destroy itself without player input.
Micromanagement is fine, babysitting is not.

[jlehtone]

The main thing it's always been essential to watch out for is ships that have no viable targets in clear line of sight. That's the point at which they start trying to find a route to undamaged modules on the other side of the station.

That's exactly what people are complaining about: AI shouldn't autonomously plan routes that might destroy itself without player input.
Micromanagement is fine, babysitting is not.

In other words, the AI should be changed. Two obvious options:
A. Do nothing
B. Plot route differently

[flywlyx]

In other words, the AI should be changed. Two obvious options:
A. Do nothing

And that's precisely what older RTS game AI does: they halt immediately after executing the current order.
However, X4 is better than that, it has the command queue feature, players should have the ability to leverage this function.
Station modules should be selectable from the map view, allowing players to set up move-attack order queues in advance instead of babysitting all the destroyers.

[TroubledRabbit]

I won't be discussing personal preferences - if someone thinks that the current system is fine, it is (for them).

The main complaint - and it is quite repetitive not only in this topic - is that 'AI' trying to be 'smart' is terribly dumb while being much simpler would actually be 'smarter' by the virtue of not making evidently wrong decisions (which human player would never do).

Queving orders (which is very good by itself) is not an answer: situation changes and expecting that 'AI' updates on the fly parameters of the execution is just too much to ask, and the queve won't help in any way if the basic execution logic is just not reasonable (by 'AI' standard of 'reason' ofc.). Ability to select particular modules would be great but it still won't solve the basic problem: a ship trying to reach 'optimal position' ('AI' does not understand the 'good enough' approach) may and probably will get into trouble.

I think that overcomplication and overambition are the main culprit there. It would be fantastic if each ship model has got its own combat logic (e.g. having main battery on the upper deck it would properly align itself to the target *not* forgetting to cover own ass by positioning among support units like real frigates - which we do not have in the game at all - you could see that in the 1st Homeworld as simple as the game logic was) but that is kind of deadly rabbithole for a application which needs to process so much at 'once' already.
Instead I would propose the KISS principle or 'do or do not' if you prefer such spelling. And dropping seemingly nonsense limitations like the max-pitch for destroyers which forces them to pointless 'nod-dance' and getting smashed in the process. For now we have ships unable to use directional thrusters and unaware about 'magnet tool' - that last gave me so much of motion-sickness that I will never stay on the ship with 'collect drops' order.

[charlie1024]

That's over 10k hours of personal experience, during which my demolition fleets have destroyed innumerable stations using the tactics outlined above. It's been literal years since I last lost a destroyer to station turrets (3.0 Split game, had some issues with Rattlesnake's short range guns). Most of the time they don't get to fire even a single shot back at my ships. The main thing it's always been essential to watch out for is ships that have no viable targets in clear line of sight. That's the point at which they start trying to find a route to undamaged modules on the other side of the station. Those routes are often distinctly sub-optimal, which is precisely what my 'fly & wait, then attack' approach is intended to counter - i.e. give them a safe flightpath to a point from which they can shoot without ever going within range of station turrets.

The map-based module selection method in my earlier post also works for turrets which are set to 'attack my current enemy'.

Personally though don't like to go within turret range of hostile stations. Prefer to do all of my station demolition at destroyer main gun range instead. Many of my targets are Commonwealth stations which can potentially be armed with several dozen L plasma turrets. In an artillery duel between a station with 70+ L plasmas (e.g. this one) & a destroyer with substantially fewer, the station tends to win.

At least in 6.2 we haven't lost too many capships, but even in the version there were some possibility of losing ships due to running directly to the station. It has been true for long time, and even Egosoft has been making an effort to fix the issue, it means that at least there has been some problems in capship's main battery management.

Even, actually getting the perfect LOS is not very a big issue when we can control the ships perfectly. However, for 3D space in 2D monitor, we always have to get hard times to provide the ships for perfect positions. For good LOS, it is good to get the ships below or above the station. That's true. So, manually adjusting 10~20 ships' position in hand is good? Personally, don't think so.

Even after getting their own 'perfect' position some ship(of course, just 1 or 2 from 20 ships) still tried to run directly to the station. Or, they just didn't attack anything. They were able to attack the target directly, but didn't. The bugs are reported by myself and fixed, but I still cannot understand why the ships failed to attack anything, or just ran directly to the station.

Although, I think the risk of attacking station has been gradually lowered from the release of the game.