Update on Progress

[Bisc8]

Worth noting here that there will be many AIs (ship, fleet, system, population, etc.  ) all working within the strategic AI. 

Will those low level AIs also be available for the player? For example, if the player don't want to manage the colonies, can he order the population AI to take care of it for him?

[firsal]

These changes sound absolutely amazing! However, I do have a few questions. Will the AI be able to change its doctrine or design philosophy in reaction to its experiences in combat? For example, if it finds its current beam-based designs ineffective versus an enemy, will it be able to create new designs in response to these experiences, or even change to a more effective design philosophy? Will it also prioritize certain research fields and/or create new components based on experiences in combat in order to counter specific threats?

[Steve Walmsley (OP)]

These changes sound absolutely amazing! However, I do have a few questions. Will the AI be able to change its doctrine or design philosophy in reaction to its experiences in combat? For example, if it finds its current beam-based designs ineffective versus an enemy, will it be able to create new designs in response to these experiences, or even change to a more effective design philosophy? Will it also prioritize certain research fields and/or create new components based on experiences in combat in order to counter specific threats?

Not initially (or I will never get the mythical test campaign started)

I will continue to develop the AI post-launch though and these are definitely areas I will tackle at some point.

[Profugo Barbatus]

I'm certainly ok without to start. Just having to retool my fleet to deal with potential countering AI compositions will be very interesting, and help protract AI warfare. Combine that with ground combat and the AI doing ground combat, and I can see myself finally playing AI campaigns instead of pure RP campaigns again.

Question, in the interest of combating potential slowdown, is there any way we could enforce the expansion limits? Having the AI not spawn more AI helps, but being able to force them to set themselves a systems-from-homeworld limit would be nice for long term campaigns.

[Froggiest1982]

These changes sound absolutely amazing! However, I do have a few questions. Will the AI be able to change its doctrine or design philosophy in reaction to its experiences in combat? For example, if it finds its current beam-based designs ineffective versus an enemy, will it be able to create new designs in response to these experiences, or even change to a more effective design philosophy? Will it also prioritize certain research fields and/or create new components based on experiences in combat in order to counter specific threats?

Not initially (or I will never get the mythical test campaign started)

I will continue to develop the AI post-launch though and these are definitely areas I will tackle at some point.

Adaptive AI is a nightmare as this will impact the whole production chain with the risk of looping production each time it encounters a new race plus the research tree component factor. I think you could tackle this right away just adding "standard" designs of multi-role ships and if possible just rather than have 100% research and ship production focused on the philosophy balance it to 80-20 or even 70-30 making sure there are always a good base tech and ships able to serve the purpose. I think the actual V6 races already use something like that, so would be a good starting point. At that point, you need to focus more on the new Strategic AI behaviour (assign the right ships to the right enemy etc) which I believe would be easier than constantly balancing it.

However, I do really love the concept therefore if you manage to integrate it...

[Father Tim]

These changes sound absolutely amazing! However, I do have a few questions. Will the AI be able to change its doctrine or design philosophy in reaction to its experiences in combat? For example, if it finds its current beam-based designs ineffective versus an enemy, will it be able to create new designs in response to these experiences, or even change to a more effective design philosophy? Will it also prioritize certain research fields and/or create new components based on experiences in combat in order to counter specific threats?

Not initially (or I will never get the mythical test campaign started)

I will continue to develop the AI post-launch though and these are definitely areas I will tackle at some point.

Adaptive AI is a nightmare as this will impact the whole production chain with the risk of looping production each time it encounters a new race plus the research tree component factor. I think you could tackle this right away just adding "standard" designs of multi-role ships and if possible just rather than have 100% research and ship production focused on the philosophy balance it to 80-20 or even 70-30 making sure there are always a good base tech and ships able to serve the purpose. I think the actual V6 races already use something like that, so would be a good starting point. At that point, you need to focus more on the new Strategic AI behaviour (assign the right ships to the right enemy etc) which I believe would be easier than constantly balancing it.

However, I do really love the concept therefore if you manage to integrate it...

That sounds like a largely terrible idea.  If an NPR's beam ships aren't good enough, it's AI should seek to build better beam ships, not abandon beam ships entirely.  I don't want to see every enemy evolving towards the same end design.

[Panopticon]

These changes sound absolutely amazing! However, I do have a few questions. Will the AI be able to change its doctrine or design philosophy in reaction to its experiences in combat? For example, if it finds its current beam-based designs ineffective versus an enemy, will it be able to create new designs in response to these experiences, or even change to a more effective design philosophy? Will it also prioritize certain research fields and/or create new components based on experiences in combat in order to counter specific threats?

Not initially (or I will never get the mythical test campaign started)

I will continue to develop the AI post-launch though and these are definitely areas I will tackle at some point.

Adaptive AI is a nightmare as this will impact the whole production chain with the risk of looping production each time it encounters a new race plus the research tree component factor. I think you could tackle this right away just adding "standard" designs of multi-role ships and if possible just rather than have 100% research and ship production focused on the philosophy balance it to 80-20 or even 70-30 making sure there are always a good base tech and ships able to serve the purpose. I think the actual V6 races already use something like that, so would be a good starting point. At that point, you need to focus more on the new Strategic AI behaviour (assign the right ships to the right enemy etc) which I believe would be easier than constantly balancing it.

However, I do really love the concept therefore if you manage to integrate it...

That sounds like a largely terrible idea.  If an NPR's beam ships aren't good enough, it's AI should seek to build better beam ships, not abandon beam ships entirely.  I don't want to see every enemy evolving towards the same end design.

I don't disagree, but there should be some sort of balance, optimally played missile ships are pretty much always gonna beat optimally played beam ships, assuming the logistics are there, so an energy loving fleet will eventually need to make some nod to the missile tech trees, if only to produce effective antimissiles.

[Graham]

If an entire empire is geared towards beams in ships, infrastructure and tech. Then switching to missiles in the middle of a war is not going to help it. It will just end up with a janky composition of beam ships and terribly outdated missile designs.
Changing design philosophies to counter enemies is important but switching to a completely new tech basis and doctrine is just insanity unless perhaps due to salvage you are able to get a giant tech boost in the right direction.

Even if you only want to develop anti missiles, getting them to the point where they are close to as effective as your beam pd will still be a very sizeable tech investment, which probably won’t be worth it.

[Father Tim]

I don't disagree, but there should be some sort of balance, optimally played missile ships are pretty much always gonna beat optimally played beam ships, assuming the logistics are there, so an energy loving fleet will eventually need to make some nod to the missile tech trees, if only to produce effective antimissiles.

In my experience, optimally played missile ships never beat optimally played beam ships -- strategically --  because of logistics.  Armour and shields and point defense can be cheaper than the missiles needed to destroy them.

[TCD]

I don't disagree, but there should be some sort of balance, optimally played missile ships are pretty much always gonna beat optimally played beam ships, assuming the logistics are there, so an energy loving fleet will eventually need to make some nod to the missile tech trees, if only to produce effective antimissiles.

In my experience, optimally played missile ships never beat optimally played beam ships -- strategically --  because of logistics.  Armour and shields and point defense can be cheaper than the missiles needed to destroy them.

In any case, even ignoring logistics I don't think we actually know that missiles will be superior to beams for C# Aurora. With all the changes going through balance is going to be pretty different.

[Panopticon]

Not getting into the missiles vs beams argument, because even if you are right then my point remains, let's flip the script then and say its a missile focused NPR vs a beam optimized player, wouldn't the AI still need to start making adjustments rather than focusing on it's current outclassed strategy?

Yeah it's a tech investment and might be hard to catch up in, but what are the other options if you are in a losing war with your current tech? Stay the course and go down fighting? Sue for peace or a truce and get time to change? Surrender?

the last couple options would be cool if the AI is flexible enough to do that, but the stay the course and go down fighting option seems like maybe no the thing to commit to every time, in an ideal world at least.

[Father Tim]

Yes; stay the course and go down fighting, every time.  I am not interested in playing against a thousand NPRs that end up as perfect-tech clones.  I want to fight the empire that commits to size 13 missile launchers and plasma carronades, or the empire that loves FACs and massed Gauss cannons.

To extend the metaphor, let's abandon Aurora and think Star Fleet Battles instead.  I don't want an AI that figures out Photon Torpedoes are the best heavy weapon and abandons Disruptors and Plasma Torpedoes and Hellbore Cannons and PPDs, etc.  I don't want my Klingon and Romulan and Kzinti and Gorn and Hydran and ISC enemies all turning into Orion Pirates because photons & engine doubling equal an "I win" button.

It is more important to me that computer-controlled enemies be flavourful, and noticeably different form each other, than that they provide the theoretical best possible tactical challenge.

An NPR that builds a fleet of 60% missile cruisers and 40% beam fighters should strive to build the best damn possible fleet of 60% missile cruisers and 40% beam fighters, not a fleet of 100% supercarriers loaded with fighters equipped for long-range missile strikes.

[Froggiest1982]

Yes; stay the course and go down fighting, every time.  I am not interested in playing against a thousand NPRs that end up as perfect-tech clones.  I want to fight the empire that commits to size 13 missile launchers and plasma carronades, or the empire that loves FACs and massed Gauss cannons.

To extend the metaphor, let's abandon Aurora and think Star Fleet Battles instead.  I don't want an AI that figures out Photon Torpedoes are the best heavy weapon and abandons Disruptors and Plasma Torpedoes and Hellbore Cannons and PPDs, etc.  I don't want my Klingon and Romulan and Kzinti and Gorn and Hydran and ISC enemies all turning into Orion Pirates because photons & engine doubling equal an "I win" button.

It is more important to me that computer-controlled enemies be flavourful, and noticeably different form each other, than that they provide the theoretical best possible tactical challenge.

An NPR that builds a fleet of 60% missile cruisers and 40% beam fighters should strive to build the best damn possible fleet of 60% missile cruisers and 40% beam fighters, not a fleet of 100% supercarriers loaded with fighters equipped for long-range missile strikes.

Well that was my point quoting myself """adding "standard" designs of multi-role ships and if possible just rather than have 100% research and ship production focused on the philosophy balance it to 80-20 or even 70-30"""

You have 70% ships that follow the philosophy and another 30% fishing from the standard pool that will be then managed and assigned by the new Strategic AI to tackle the right threat.

I believe same as you do those computer-controlled enemies have to be flavourful, and noticeably different from each other.

EDIT: for standard, I meant not philosophy related rather than a prebuilt blueprint ship.

[the obelisk]

I think we should look for a middle ground here.  I do agree that the NPR's should have some capacity for adapting in the face of their doctrine completely failing, but at the same time I agree that they should retain their flavor, rather than just sinking massive amounts of RP into whatever the optimal ship builds are.  Seems to me that what makes sense is for an NPR in that situation to alter it's doctrine based on what technology it has available.  So basically, it would be adapting by using more or less of a certain kind of ship, using ships with load-outs it wasn't previously using, building types of ships it hadn't previously built, and/or using ships in ways it hadn't previously used them.  So, for example, if an NPR that shows a strong tendency towards large capital ships starts losing pretty badly, it might decide to start focusing on building a larger number of smaller ships, and an NPR that generally prefers large fleet on fleet battles that starts losing badly might switch to a more raiding/hit-and-run oriented strategy.  In either case, they'd largely do this using the technology they have, rather than trying to totally optimize the tech aspect of whatever it is they're trying.

[Panopticon]

Yes; stay the course and go down fighting, every time.  I am not interested in playing against a thousand NPRs that end up as perfect-tech clones.  I want to fight the empire that commits to size 13 missile launchers and plasma carronades, or the empire that loves FACs and massed Gauss cannons.

To extend the metaphor, let's abandon Aurora and think Star Fleet Battles instead.  I don't want an AI that figures out Photon Torpedoes are the best heavy weapon and abandons Disruptors and Plasma Torpedoes and Hellbore Cannons and PPDs, etc.  I don't want my Klingon and Romulan and Kzinti and Gorn and Hydran and ISC enemies all turning into Orion Pirates because photons & engine doubling equal an "I win" button.

It is more important to me that computer-controlled enemies be flavourful, and noticeably different form each other, than that they provide the theoretical best possible tactical challenge.

An NPR that builds a fleet of 60% missile cruisers and 40% beam fighters should strive to build the best damn possible fleet of 60% missile cruisers and 40% beam fighters, not a fleet of 100% supercarriers loaded with fighters equipped for long-range missile strikes.

You have strong opinions with good reasons for them and I respect that, it may not be exactly what I want to see in an ideal game but I'll surely play that and have fun with it either way.