Knowledge of Enemies

[aruon]

An intersteller empire's signals intelligence service would have entire departments of very smart people devoted to understanding, replicating and ultimately being able to interact with the computer technologies of alien empires. Figuring out the basis of the alien's IT technology wouldn't be that hard, you just get hold of civilian examples and get your engineers to figure out how it works. Working out how their communications technologies work would be harder and breaking their cryptography would be extremely challenging. But you throw enough brains, money and computer processing power at it, you could at least have a crack at it.

this. in Halo both the UNSC and Covenant knowledge of each other came exclusively from intelligence and salvaging components. AIs would also help massively considering (once again from halo) they fly entire capital ships single-handedly while figuring out how an unfamiliar FTL drive works, building a plasma-shaping weapon system from scratch, fending off onboard enemies and an enemy AI all at the same time.

AIs are the underappreciated badasses of sci-fi.

 

[FelixG]

this. in Halo both the UNSC and Covenant knowledge of each other came exclusively from intelligence and salvaging components. AIs would also help massively considering (once again from halo) they fly entire capital ships single-handedly while figuring out how an unfamiliar FTL drive works, building a plasma-shaping weapon system from scratch, fending off onboard enemies and an enemy AI all at the same time.

AIs are the underappreciated badasses of sci-fi.

I think AIs tend to be downplayed because really, once you have them they can overshadow the humans they work with in every single way. Heck, they never even bothered to make AI automated weapons to fight the covies in HALO even though they could be small, fast, deadly, and synced together by a single driving intelligence that would let them work together perfectly!

 

[aruon]

I think AIs tend to be downplayed because really, once you have them they can overshadow the humans they work with in every single way. Heck, they never even bothered to make AI automated weapons to fight the covies in HALO even though they could be small, fast, deadly, and synced together by a single driving intelligence that would let them work together perfectly!

now here's a funny thought: an AI technology in stellaris that allows for AIs to be added to a ship which means the AI performs better in battle than the standard game AI. MIND FREAK!

 

[FelixG]

now here's a funny thought: an AI technology in stellaris that allows for AIs to be added to a ship which means the AI performs better in battle than the standard game AI. MIND FREAK!

I really do hope they let us give full control of our ships over to AIs as a research (The Culture ships for example!) and it would be cool if they got little bonuses to things like reaction time and accuracy since they work faster than fleshy crews.

 

[Chat]

I like the idea of learning much of what we know about other empires though espionage. I suppose a lot of it comes down to more passive methods though, like sensor arrays/ships watching the other empires.

The idea of being able to mark something as secret, even towards your allies is pretty appealing. Allow them to know much about your worlds, but keeping a facility or even an entire world hidden, labelled as an uninhabitable world in a dangerous region of space on their books. Secretly housing one of your main weapon research and development labs, and where your latest and greatest dreadnoughts slumber. Anyone watching long enough would realise that it's not so uninhabited as your government proclaims as they watch supplies being delivered to nowhere, but the hostile nature of the region and in particular the defenders, limit how much anyone can learn about what goes on there.

I really do hope they let us give full control of our ships over to AIs as a research (The Culture ships for example!) and it would be cool if they got little bonuses to things like reaction time and accuracy since they work faster than fleshy crews.

That's how an AI module worked in SotS. Slap one onto your ship as the command section, and it would operate considerably better than a flesh driven ship.

Downside? If your AI rebelled, they took every ship with an AI module installed with them. It made installing even half of your warships with AI command modules a very risky proposition, unless you had researched AI Virus or AI Slave to preempt any potential rebellion.

 

[FelixG]

I like the idea of learning much of what we know about other empires though espionage. I suppose a lot of it comes down to more passive methods though, like sensor arrays/ships watching the other empires.

The idea of being able to mark something as secret, even towards your allies is pretty appealing. Allow them to know much about your worlds, but keeping a facility or even an entire world hidden, labelled as an uninhabitable world in a dangerous region of space on their books. Secretly housing one of your main weapon research and development labs, and where your latest and greatest dreadnoughts slumber. Anyone watching long enough would realise that it's not so uninhabited as your government proclaims as they watch supplies being delivered to nowhere, but the hostile nature of the region and in particular the defenders, limit how much anyone can learn about what goes on there.

This is a highly appealing idea, I love the thought of having a system that looks uninhabited while you store things you don't want others to know about, say setting up a fairly useless system ahead of time right next to your boarder with a power where you start to store ships knowing you will be going to war, while your spies and espionage team makes those ships and material look like they are going elsewhere and nothing is actually in that system.

 

[Quaade]

I like the idea of learning much of what we know about other empires though espionage. I suppose a lot of it comes down to more passive methods though, like sensor arrays/ships watching the other empires.

The idea of being able to mark something as secret, even towards your allies is pretty appealing. Allow them to know much about your worlds, but keeping a facility or even an entire world hidden, labelled as an uninhabitable world in a dangerous region of space on their books. Secretly housing one of your main weapon research and development labs, and where your latest and greatest dreadnoughts slumber. Anyone watching long enough would realise that it's not so uninhabited as your government proclaims as they watch supplies being delivered to nowhere, but the hostile nature of the region and in particular the defenders, limit how much anyone can learn about what goes on there.

Really neat idea to have a choice between keeping all nation closed down, being open for all to see or keeping some informations hidden even to allies and best friends. Should differ about what you can learn simply by observing and interactions with a specie which would bring up some knowledge of them and to what they keep away from you and you must discover through spies, being discovered would of course have it´s consequenses especially if they had nothing to hide or you didn´t discover anything (yet). But making it a necessity to even spy on allies in order to gain knowledge of a potential threat down the line.

Having it divided into obersvations and espionage would be interesting if observations could be false, due to past interactions where you think a specie is friendly only later to find out that they had plans to conquer vital resources in you systems and mobilizing on a hidden millitary base on an inhabitable planet.

 

[Tim_Ward]

I think that would only be relevant to multiplayer. I can't see the AI caring one way or another about your secret moon base.

 

[FelixG]

I think that would only be relevant to multiplayer. I can't see the AI caring one way or another about your secret moon base.

That would depend on how they are programmed to respond to force buildups. If they are oblivious to obvious buildup of forces on their boarders then yeah, it doesn't make a difference.

If the AI reacts with a buildup of its own fleet however in response to you however keeping the secret would be highly beneficial.

 

[Tim_Ward]

That is true, but think that would work on a slightly different set of game-play mechanics than espionage and political secrecy. That would be more the sensor/fleet movement/cloaking dynamic, if there is one.

I suppose one area that might be AI relevant is your ship designs - it is not uncommon for AIs to react to your designs by changing it's own designs, especially if the combat mechanics are very rock/paper/scissors-ish. I think Endless Space did that. So, if it could nick information about your designs it would be able to build ships to counter them without having to go through the process of losing to them a bunch of times first.

Speaking of the AI, here's a feature I've always wanted to see: you can use spies to get information about the AI's internal state. I'm not really sure how 4x AI programming works exactly, or even if there are consistent techniques used across the industry but I imagine there's some kind of top level process which decides what the AI should be doing (expanding, focusing on research, attacking someone, etc) and others which actually issue orders and so on.

What it would be cool to be able to do is use espionage to actually find out what the AI's plans are; is it planning to attack you? is it planning to attack someone else? does it want to expand into the same area you do? does it just want to sit on it's ass and act defensive?

Another area is diplomacy. In the real world, when allies spy on one another, 9 times out of 10 it's because they want to know the negotiating position of their ally on some deal that's being proposed. It'd be cool to be able to use spies to find out what, exactly, an AI would be willing to accept in order to accept certain treaties.

 

[Quaade]

I think that would only be relevant to multiplayer. I can't see the AI caring one way or another about your secret moon base

That would depend on how they are programmed to respond to force buildups. If they are oblivious to obvious buildup of forces on their boarders then yeah, it doesn't make a difference.

AI is supposed to react to armies on borders in HOI4 when deciding who to trade with, favouring the one who threatens them over those who don´t. But since the game hasn´t been released yet, we cannot know if this system works and the AI is capable of doing so in a meaningful way. However it seems like they are at least playing with the idea of having an AI reacting to build-ups, but the main issue is really fog of war, since AI has proved less capable of handling fog of war this has been cancelled for AI in order to adjust to player in action better. This should be improved or at least give the AI the ability to only react to this when it´s discovered.

What it would be cool to be able to do is use espionage to actually find out what the AI's plans are; is it planning to attack you? is it planning to attack someone else? does it want to expand into the same area you do? does it just want to sit on it's ass and act defensive?

This would require AI to have longterm goals aswell as shortterm goals which might not be aspiring for the same or at least in a more logical way, ie. expanding one way in order to build up the power to tackle player who holds a few planets they desire. But I am in full support of this one

Another area is diplomacy. In the real world, when allies spy on one another, 9 times out of 10 it's because they want to know the negotiating position of their ally on some deal that's being proposed. It'd be cool to be able to use spies to find out what, exactly, an AI would be willing to accept in order to accept certain treaties.

A good idea, to have agreements be more of a guess unless you know a bit of their desires. Make you some time overpay for a deal, but should still be some kind of tell like having "maybe" in a great range instead of guessing all the time. Say yes would be 100 space money, it would say "maybe" in the range of 50-150 where going over means paying more than they wanted and going under would result in a no. Having knowledge of this, would make the range into 90-110 perhaps. Though given how players are able to exploit these number and simple guess the by using a system like this, it would either have to be sophisticated or change it´s variable each time.

 

[TheAtreides84]

Isn't that the kind of thinking that created the Japanese-American interment camps? Unless the polity is Tokugawa Shogunate level of xenophobia or higher (and there was still a small number of foreign merchants in Edo Japan at any given time), other species' tourists/merchants/dignitaries/missionaries/migrants/descendents of migrants/mercenaries/et al aren't going to be so rare that the immediate conclusion of bystanders at seeing an alien is "Probable spy, better call the police."

Also, there could be traitors. In a game where ethics are a thing, there will be individuals willing to put ideology (or personal gain) above nation or race.

 

[Chyll]

Unless there are no greedy alliens, no migration and no trade, I don't see why it would be impossible.

Difficult vs. impossible

 

[TheAtreides84]

That would depend on how they are programmed to respond to force buildups. If they are oblivious to obvious buildup of forces on their boarders then yeah, it doesn't make a difference.

If the AI reacts with a buildup of its own fleet however in response to you however keeping the secret would be highly beneficial.

In these kind of games the AI usually doesn't abite to fog of war rules. I guess it's almost impossibile to make it somewhat smart with limited informations.

 

[FelixG]

In these kind of games the AI usually doesn't abite to fog of war rules. I guess it's almost impossibile to make it somewhat smart with limited informations.

If/then statement?

If non friendly faction moves forces within 1-5 (or whatever) hexes/parsecs/ect of shared boarder that do not have the [HIDDEN/STEALTH/WHATEVER] tag then move appropriate countering force to same shared border.

After that you provide a few ways fior ships to remain tagged as hidden or stealthed as well as a few ways for factions to remove those tags?

 

[Chat]

This would require AI to have longterm goals aswell as shortterm goals which might not be aspiring for the same or at least in a more logical way, ie. expanding one way in order to build up the power to tackle player who holds a few planets they desire. But I am in full support of this one

The AI having long-term goals which it typically generates short-term goals to help fulfill the long-term ones sounds like a really good system for the AI. I'm guessing a big part of the reason why we don't see it already is the difficulty in fluidly generating reasonably achievable medium and short term goals to advance the long term ones, and recognising an opportunity to advance long term goals when it appears, even if it hurts the original short term ones. Especially for cheap when it comes to memory and processing power.

 

[Quaade]

The AI having long-term goals which it typically generates short-term goals to help fulfill the long-term ones sounds like a really good system for the AI. I'm guessing a big part of the reason why we don't see it already is the difficulty in fluidly generating reasonably achievable medium and short term goals to advance the long term ones, and recognising an opportunity to advance long term goals when it appears, even if it hurts the original short term ones. Especially for cheap when it comes to memory and processing power.

I believe it´s due to over-complicating, but I´m no expert, but it a common mistakes when doing lots of thing. Make the long term goals as simple as possible and not the main goal but only a, check_if_true. Then go down and the list and make them a bit more complex, like need resources or gold could be one if less than an amount equal to the target, less troops, less provinces. To fulfill these it determines if it can do so within it´s limit, like force limit or less provinces. A simple check as to, is target weak if no, is any neighbour weak (since a long bloody war would mean getting attacked from your rival) could do it in theory.

Having a more complex AI is wanted and seems like there are improvements in general to this end

 

[Chat]

I believe it´s due to over-complicating, but I´m no expert, but it a common mistakes when doing lots of thing. Make the long term goals as simple as possible and not the main goal but only a, check_if_true. Then go down and the list and make them a bit more complex, like need resources or gold could be one if less than an amount equal to the target, less troops, less provinces. To fulfill these it determines if it can do so within it´s limit, like force limit or less provinces. A simple check as to, is target weak if no, is any neighbour weak (since a long bloody war would mean getting attacked from your rival) could do it in theory.

Having a more complex AI is wanted and seems like there are improvements in general to this end

I'm no expert either, but when I'm thinking about how we humans think and do our long term goals, it seems like it might be fairly intensive when it's expanded to hundreds of AIs. Like wanting to take over France as Ulm, you normally might pick either conquest of smaller neighbours, or aim to become the HRE Emperor. Then you're looking at who's around you, what they can contribute to your goals. Who has the power to protect you, who's a threat, who will go to war with you, who will ally you, who will accept diplo-annexation, can I take advantage of a strong country that has sent 3/4 of their troops to the other side of the world to fight a war there, etc. All of that I think is achievable, but how well does it scale with the number of AI normally found in PDS titles? How good and natural can you make it feel with that restriction?

 

[Quaade]

I'm no expert either, but when I'm thinking about how we humans think and do our long term goals, it seems like it might be fairly intensive when it's expanded to hundreds of AIs. Like wanting to take over France as Ulm, you normally might pick either conquest of smaller neighbours, or aim to become the HRE Emperor. Then you're looking at who's around you, what they can contribute to your goals. Who has the power to protect you, who's a threat, who will go to war with you, who will ally you, who will accept diplo-annexation, can I take advantage of a strong country that has sent 3/4 of their troops to the other side of the world to fight a war there, etc. All of that I think is achievable, but how well does it scale with the number of AI normally found in PDS titles? How good and natural can you make it feel with that restriction?

To some extent the AI in EU4 do these calculations quite well, having them attack at weak moments both due to weak allies or a present war. What they need to improve is having the AI become more active to it´s neighbours and rivals, like Mamluks not expanding south or attacking Otto but sits there waiting for them to attack and conquer them. But much of these calculations have proved in EU4 to be made succesful, they only need to improve it further making it less random and overpowered for some.

But yeah, much of the devious calculations made of humans would either be too dificult to make an AI use or simply take up too much power. But I have faith in PDX and their abilities in their use of AI and has seen the AI being much more improved than before, so hopefully they will get it somewhat right eventually giving us the illusion that they are like us.

 

[Tim_Ward]

This would require AI to have longterm goals aswell as shortterm goals which might not be aspiring for the same or at least in a more logical way, ie. expanding one way in order to build up the power to tackle player who holds a few planets they desire. But I am in full support of this one

Oh either's good, I just think it would be neat to be able to get information about the AI's current behaviour via spies; doesn't matter specifically what they are; obviously that depends heavily on how the AI actually works.