Human vs AI

I recall in one of my last conversations as a contributor, that I had mentioned how events favored the player more than the AI. Keith, (aReclusiveMind) had replied at the time that he wasn’t convinced about that. I was stunned by this response.

In a game like ISG, all rules, mechanics, events, etc, that are equally available to both the human player and the AI, can always be better used by the player.

In a game like ISG, all rules, mechanics, events, etc, that are equally available to both the human player and the AI, and are better used by the player, will confer an advantage to the player.

This is not debatable. If this were not true, then there would be little need to give the AI bonuses and cheats.

This thread then will delve into specifics and be broken into parts, showing the difference in AI and human use of game components and mechanics. The purpose of this is two-fold

First, while I acknowledge that the AI cannot match a skilled or experienced player, it can certainly learn how to better use the assets available. Perhaps then, some of what I document can be of help in getting better performance from the AI and giving the players a richer experience.

The second purpose is to introduce new players to some of the subtleties that abound in this game and aid them in formulating their own strategies.

Part one Leaders (General)

I recently wrote in my updated reviews at various sites that leaders in ISG could almost be a mini game of their own, they can be, but perhaps in ways that aren’t obvious.

First I want you to look at the shot below. This is one of my low tech games, but that is not a factor here, what is a factor is that this is around turn 300.


Next I want you to take a look at this here

The first shot doesn’t clearly show Ruiz or Dexanna, but the relevant information is there. Start by looking at wages per turn. What do you see? My 8 leaders, all of them in the green (opinion), all of them leveled up, (most of them multiple times), have a combined cost of 103 BC per turn. Dexana, who I presume was lost by some other empire, has a cost of 71 BC per turn on her own. The link I provided shows another example of run-away costs, where Lord Matjosh was asking for 192 BC per turn.

Obviously the greedy trait plays a role (and we will get into negative traits shortly), but the main point is that the human player can keep leaders happy and keep costs down. I do this in part by almost always turning down merit raises, especially for greedy leaders, the trick is to keep them happy anyway. The AI on the other hand, cannot do this even though the tools are there. Advantage human

One way the AI could be helped here is by denying merit raises to leaders who have achieved a certain threshold in opinion.

But it isn’t only about costs, there are other things we need to discuss. In order do so, I will now talk about specific leader types.

Leaders (ship)

Here are some general hints in keeping your ship leaders happy, and alive.

Do not keep them in the same fleet, even if they don’t have the lone wolf trait, they generously get the envious trait much more quickly if they are with other leaders, especially if those leaders have similar traits.

When you first hire a leader, pay attention to the command rating

Any leader that can command a Frigate or Destroyer only, will not ask to command a ship in battle

Any leader that can command a Cruiser from the get-go, will not ask to command one in battle if assigned to the maximum command allowed. In other words if say Ruiz shows up and you hire him, you can get him to hold off on his desire by assigning him directly to a Cruiser. If you assign him to a smaller ship, he will then ask to command a Cruiser in battle (the highest class he can command). So If I put him on a Cruiser, level up his command to Battleship (without assigning him to one), his first desire will be to command a Battleship in battle, meaning his bonuses will be available when he is on a Battleship, very handy to know this. (Tamara seems to be an exception and will generally wait for a Titan)

You want to pay attention to this as it is useful to have your best leaders, have their best performance, on the largest hull possible. The AI cannot do this, advantage human

Speaking of command rating, it seems the AIs neglect this entirely. I have often seen their leaders show up in large fleets, and commanding tiny Frigates and Destroyers. Here we can help the AI, it needs to level up the command rating more often and not keep leaders on small ships throughout the game. (You did note that all my leaders can now command Battleships yes?)


Leaders (colony)

Colony leaders are a bit more straightforward and yet also require a certain finesse to get them happy, productive, and keep costs reasonable. Here is how to I do it.

Before ever hiring a single colony leader, I think about where they will go, with my play-style that is always the Homeworld for the first one. Then I think about what colony specific building I’m going to want there. (Econ cap, science cap, etc). Again, for me that is always an econ cap on the HW to get the economy going fast. My very first research choice then is the econ cap, and I do not research any other type of cap building until the following occurs.

A colony leader shows up and I hire him/her/it. I then assign that leader to my HW. Because I only have Econ cap researched, that leader will ask for that building as a request, I put the building into queue. I am now free to research a second type of cap building, usually the science cap, I do so. The next colony leader that gets hired (can be hired as soon as the first leader requests a building), gets assigned to the colony that will become my research center, that leader will then ask for the only building available, the science cap in this case.

Rinse and repeat as needed.

The advantages here are that the leaders gain opinion as you build their requests, they get bonuses to production which aid you, and you get to place the buildings where you want them, not where the leaders decide.

Advantage human

Sadly, there is not much that can be done to aid the AI here, especially on harder levels where they start out with an econ cap already built. Still by following the general suggestion about keeping leader costs down, the AI can do a bit better.

Negative traits on colony leaders can’t be contained as much, if you have some with similar skills, (and most of them do), expect someone to get envious, keep them happy and you can ignore their requests to dismiss other leaders.

This concludes part one, I have shown how the same mechanics available to both the player and the AI do not equate to the same advantage, and offered some suggestions for at least trying to close the gap. Each new release promises better performance from the AI, and while I have seen some moderate improvement, there is a lot more that can be done without re-inventing the wheel.

The next part of this series will come in a few days where I will focus on infrastructure.

 

I have no issue with that, you have my permission to reprint anything I write here.

I generally have no issue with money either, and I like to squirrel it away as it becomes very handy in rushing heavy ships. With my settings they take very long to construct.

I do not, nor can I. My settings have the player on emperor, nothing but hostile worlds (and if you're lucky tolerable) near by. My home world carries the game for me until I start conquering worlds.

This is not correct. Merit raises will be 2bc per turn with the Philanthropic trait, 12bc per turn normal, and 24 bc with Greedy. This is on top of their normal pay raises. While Philanthropic is not a problem, I carefully consider the 12bc, and will almost always deny the 24.

I've seen them target mine a lot, but keep in mind my settings mean a game averages 600+ turns. You can see quite a lot during that span. I don't target theirs because I don't have to. As their worlds fall and their fleets dissolve, their leaders leave and become available anyway.

I wait on the ship leaders till before they leave as well, though situational awareness plays a role. For example, Before selecting a free survey ship, I'll hire a ship leader that is waiting. This means that if there are no leaders with an explorer trait in the pool, my leader will gain the trait automatically (It's part of the reason I got Seth up to expert explorer in this session). If Pike is hanging around and I am about to start researching a weapon, I hire him for his bonus.
I don't terraform as it would be suicide on my settings, therefore no moving colony leaders whose desires have been met. Accompanying a survey ship, heck I wrote that years ago and you are correct in bringing it up.

Well I'll get into details and touch upon that subject as well, but a qualifier first. I see the AI use lots of support ships, I use none. But that isn't because they aren't handy, I just don't think it would be fair.

I've had as many as eight (8) ship leaders and they weren't enough. Basically the number of leaders I have directly determines how many fleets I will deploy. (Note, fleets, not temporary or permanent task forces, those get no leaders)

P.S.

I truly like to read about how you play, it is so different from my own games that I find it fascinating and enlightening.

 

25% research, 75% production always. The current session also has
few monsters
Normal events
2X distance
Average planets
Average resources.
My race has
+25 ship defence
+25 ship attack
+1 research
+1 production
+1 money
Suspicious
-50% population growth
Command carrier Titan
Resilient settlers and the field.
But, there are some house rules.
Cannot research colonization tech, that means you can only colonize with your ability (no colony spamming that way)
No support ships, ever.
No terraforming, restricted to eco level enhancement only
No building worlds
You must be set at emperor to ensure only garbage worlds are nearby.

The AI gets
Standard stock races minus the humans who are replaced with my super efficient (but few in number) Terrans
Most favorable start possible, giving them extra tech, extra buildings, a larger starting pop, extra infrastructure, plus ideal and rich/ultra-rich worlds nearby.
Also, a 20% advantage in both research and production in relation to me. (I may up this to 30%)

This creates a situation where the initial advantages held by the AI gain greater impact as they are more difficult to overcome. Combine this with the 50% population penalty, harsh surroundings, and the self imposed restrictions, and the result is you get a very long game while never really having the ability to outproduce or out-research your opponents. It creates an epic game where you are the underdog.

I think this would be very difficult for a one colony, (or system) approach, even on Sol.

 

The idea with the race is that you have fewer of everything, but at least it isn't garbage (quality over quantity). One of the things I dislike is how ineffective early weapons are in ISG. Most players wont notice this as much as they will quickly research tech and algorithms to offset it. I, with my settings, don't get to research anything quick, the AI with its settings will research quicker on top of their initial advantage. In the end it just keeps me able to compete, though I could spend it elsewhere and not much would change (provided there are no very early conflicts). In the end, 25% to ship attack and defense is not very powerful.

Yes there are, but combined with the rest of my settings/restrictions they aren't really available to me. So I have not only a slower growth rate, but I cap at a lower number as well seeing as I am on crappy worlds. Hold on, I will show you something, I'll be right back.
Edit
Take a look at the new game, this one may get an AAR
I have just set my second colony down, it's turn 72. My total population is 10, soon to be 12. That's pitiful at this stage. The small tolerable world has caves so I got lucky, the pop there will go to 10 normal, 12 with my vaults, but that will be a long time from now. The medium hostile world was actually the next best choice, it will cap at 8 (6+2), but again, a very long time from now. Do you see who is next to me? They have no such constraints and I had to accept a tribute treaty as I can't fight them this early. That population penalty combined with the rest of the constraints I am facing can only be overcome one way. Conquering some Alien worlds. Like you, I tend to destroy most of them, but I do assimilate quite a few.


Double Edit
Here it is 28 turns later, total pop 13. Can you see how the settings are keeping my growth restricted?

 

I neglected to mention that I set AI aggressiveness to high, (one level before maximum), that may play a role. Usually I see many tribute requests, I'm surprised you don't considering your lack of a credible fleet. I find whenever I am weak, they ask, when my fleet grows, they tend to get more polite.
Good luck on the stellaris game, I have one on hold, and it's fun, but the two games are very different.

 

Part 2 Infrastructure


Situational awareness leads to situational decisions, this is a key component to successful game play in many facets of ISG. Today’s topic, Infrastructure, is no different.

When first colonizing a new world the player will be presented with the screen below and prompted to make a choice as to the first infrastructure pick. It should go without saying that all picks are not created equal and are dependent on numerous factors. Before getting into that however, let’s briefly recap the attributes of the five choices available in Infrastructure.




Planetary Enginnering

Unique among the options available, planetary Engineering allows Empire wide bonuses to be gained by completing three of three picks instead of six of six. More than that, on the surface it seems to bestow many more benefits than any other option. By selecting PE infrastructure the player will be rewarded with faster terraforming on the planet in question, faster ecological development (if pursued), bonuses to Planetary engineering research, and bonuses to the effectiveness of any support ships as to their effect on PE. Wow, it’s almost as if you are being pushed into taking this route, and when one considers the length of time that terraforming needs (under some conditions) I can see the appeal.

However, what happens once a world is terraformed and the maximum ecological level is reached? What happens is that most of the bonuses gained by these picks are lost, and unless you have the racial trait that allows you to rearrange your picks, they are to a degree, wasted. My suggestion then would be this, if you take this option, do so on a world that can truelly benefit from it and follow it through. Select all three picks to gain the Empire wide effects and then use it sparingly if at all. While the AI generally does well in allocating Infrastructure picks, I do find that it favors PE too much. Look at the next shot.




This is a Kaek colony that I conquered. The Kaek had only recently settled this world, it was already ideal to them before they cololonized. By the time I conquered it, the infrastructure level on the planet was 2, one level was lost as I targeted Infrastructure before invading. (I like a clean slate). I set the colony to 100% infrastructure and was shocked to see (on the level up), that the Kaek had spent a point on PE on an ideal world with an ecological level of two.

Civil Engineering

This is the next grouping available to the player and it is sub-divided into two sub-categories. A player needs all six of the picks here to gain Empire wide bonuses. Let’s have a closer look at what these paths offer.

Civil developers

The picks available in this path will offer bonuses to construction research and will speed up the construction of any colony buildings. Great… except that there are a limited number of buildings per colony and once the cap is reached, you lose the main bonus affiliated with this pick. On smaller worlds, this pick is almost a waste compared to other options. On larger worlds, (or those facing production challenges) a judicious pick here and there may come in handy. For a player seeking to gain the Empire wide bonuses from Civil Engineering, I suggest you focus on a single colony for doing so.

Civil planers

The other subdivision in Civil Engineering is Civil Planners. Picks allocated here grant a bonus to Economic Research, increase the number of buildings you can have on a colony, and permanently reduce the maintenance cost of any buildings present. This is a must have on smaller and Hostile worlds. In addition, the bonuses granted by these picks are permanent.

Aerospace Engineers

This is the next grouping available to the player and it is sub-divided into two sub-categories. Similar to civil Engineers, a faction needs all six of the picks here to gain Empire wide bonuses. Let’s have a closer look at what these paths offer.

Ship craftsmen

The picks available in this path will offer bonuses to weapons research and will speed up the construction of ships at the colony. This is a great pick for “shipyard” worlds (the ones you designate as such, not the planet special), and the research bonus aids one of the larger tech trees in the game. That said, unless you are going to be building ships at the colony, there may be better options. Don’t forget, the main bonus here is only beneficial if you are actually producing ships.

Ship Support

At first glance this may seem like a poor choice compared to other options available. You gain a single SSP (ship support points) per pick, a bonus to Defence research, and more experienced starting crews on ships constructed at a location where this pick has been made. Don’t be fooled, depending on your play style you may quickly find out that SSPs are all too few. The bonuses from picks here are all permanent, do not undervalue their significance.

At some point in the game you may find a perfect world for gaining all six Aerospace engineer picks, go for it if you possess a large ideal colony with a good resource rating.

General advice

Before I close this part, I want to offer some general advice that should apply to numerous play styles.

Specialize a few key worlds when it comes to picks, the effects are dramatic.

Consider the size of the colony and what you will be doing there, as a general rule I follow this;

Tiny Worlds

I hardly ever colonize these without a compelling reason, but If I do, most (if not all) picks go to Civil Planners so I can get additional building slots.

Small Worlds

Usually a mix of Civil Planners and Ship support for these worlds works well. If the planet is mineral rich, I may use a single pick on Ship Craftsmen (dependent on weather I have good ship building worlds or not)

Medium Worlds

At least one pick on Civil planners (possibly 2 if hostile to reduce penalty costs). Mix of Ship Support and Ship craftsmen with mineral richness being the deciding factor as to ratio. Possibly a single pick on Civil developers (if I need aid constructing buildings), or Planetary engineering if you want to improve the ecology quicker or plan to Terraform

Large Worlds

This is highly dependent on mineral richness, race compatibility and gravity. Suffice it to say that these are the type of colonies that I seek to establish Empire wide bonuses on. With 9 total picks available, you can also make some luxury selections on these worlds and spend a pick or two in an area that isn’t your primary focus.

Huge Worlds

Similar to large worlds, this is highly dependent on mineral richness, race compatibility and gravity. These type of colonies are excellent candidates to establish Empire wide bonuses on. With 11 total picks available, you can also make quite a few luxury selections on these worlds and spend multiple picks in an area that isn’t your primary focus.

AI use of picks

The AI for its part handles these picks credibly but cannot maximize their effectiveness to the same degree as a veteran and determined player. That said, depending on your play-style, you may find that the AI factions are very competent at Infrastructure development, with perhaps only their over-reliance on PE picks being a drawback.

Next part, All about ships

 

Yes there is a separate setting now, when I get home I'll send you a Screenshot as I can't picture the location right now,

 

My races tend to be extremely vulnerable early on and the AI seems a bit more predatory than in the past. That said, you running around with starting Frigates were a prime target, I'm surprised they didn't get pesky with you.

 

Part three All about ships

(Subsection alpha, military ships overview)

While there are many ways to play and win the game, my particular focus is to win in space. As such, I tend to create sessions where the AI factions never relinquish their lead in production and technology. This in turn leads to games where I must beat the AIs in the field, often with technologically inferior ships. The key to being able to do this is to understand the military ships available, designing them well, and producing enough of them to at least attain localized parity. Regardless of how a player approaches the game, understanding the ships available and the roles they are best suited for is beneficial.

Let’s quickly revisit the five hull types available to all factions first.

Frigate

A tiny vessel with very high evasion, the Frigate is the easiest ship to build in the early game, and the maintenance cost of this hull type is only a single SSP. This ship is very susceptible to missiles and torpedoes, and has limited value do to its lack of space. In a game with normal (or accelerated) tech and production settings, this class can be discarded entirely. While I do build these early on, I eventually scrap them all, (by retiring the class entirely), and relegate their role to the Destroyer class. This keeps my large fleets a bit more manageable, but the main reason I retire Frigates is their lack of space in relation to high tech weapons.

Primary role, raider, escort for larger ships, backwater garrison force.

Destroyer

The Destroyer is the largest hull that can be built without an orbital base. Their evasion stats are not as good as the Frigates, but they have over 100% increased hull size. They are twice as expensive, (and difficult to construct), in comparison to a Frigate, but a determined effort could see you fielding one relatively quickly. Upgrade them over time as new tech becomes available, (I never scrap these), and they will serve you well throughout the game. The Destroyer has a maintenance cost of 2SSP and is perfect in the role of an escort or to use in garrisoning vulnerable but less important systems such as outposts.

Primary role, raider, escort for larger ships, garrison duty.

Cruiser

This is your first capital ship in more ways than one. First, you need a star-base to construct one, so until you research star-base tech and actually build one, you can’t start production. Then we have a hull size double the size of a Destroyer, this means your options on what components to install increase a lot. This is also the smallest hull type that can incorporate bomb racks, but if you elect to install them on your cruiser… it can only have the offensive capability of a Destroyer. True it will be better shielded, (due to larger shields), but that will be negated somewhat by lower evasion inherent in the Cruiser hull. Usually I install bomb racks on my Cruisers during the early game (when Battleships are harder to produce), but I remove those racks during upgrades when other heavier hulls are more prevalent, and devote the extra space to weapons. Cruisers require quite an effort to construct in the early game, but at only 4SSP maintenance they are nicely balanced.

Primary role, small task force leader, fire support, garrison duty, colony bombardment.

Battleships

The Battleship is a true capital ship with poor evasion that can inflict a lot of damage, (even if carrying bomb racks). At twice the hull size of a Cruiser, it takes quite an effort from a well developed colony to construct one. It is however, susceptible to all weapon types, even direct fire Kinetics. While the Battleship can be used as a single ship force, I prefer to only do that against targets that are not heavily defended, otherwise it is a good idea to escort Battleships with numerous smaller ships. The maintenance cost of this hull type is 7SSP, which means that while it is a bargain, having a lot of them will eat up your SSPs quickly.

Primary role, Fleet/task force leader, colony bombardment, fire support, heavy garrison duty.

Titans

Titans are OP in my opinion. Not only are they allowed a unique Titan special, but they get an extra 50% structure point bonus (not actual space). While the special is cool, I don’t care for the bonus to structure. They are prohibitively difficult and expensive to construct, but at twice the size of a Battleship, they offer the greatest freedom on what components you can place on them. They have no evasion to speak of. The maintenance cost of Titans is only 12SSP, which is a bargain for what you get. Indeed, the player is actually motivated to build these exclusively once it becomes feasible (and why I dislike the structure bonus, the Titan special and discounted SSP are enough).

However be warned, playing a gigantic map may mean you need numbers, and an all Titan fleet does not allow that. You must also guard against sending these out alone as some weapons can stun the crew. It’s very frustrating when your only ship is unable to act while the enemy pounds away at it.

Primary role, Fleet leader, colony bombardment, concentrated firepower.

Next, subsection Beta, Ship design

 

Part three All about ships

(Subsection beta, military ships design)

Like all things ISG, ship design is not as straightforward as it seems at first glance. In order for a player (or the AI), to create an optimal design, many nuances and synergies need to be considered. The need to do so however is largely determined by play style. For example, if your play style leads you to overtake the AIs in tech and production, ship design is not so important as you’ve basically won already. My own playstyle though allows the AI to hold the lead in tech and production throughout the entire game. I must therefore maximize my designs so they can defeat technologically superior ships consistently. Today I’ll offer a glimpse of how I do that.

Before even considering a design I pay the proper respect to maintenance cost (SSP). What this means is that a Destroyer with battle pods and triple hull will have the same exact SSP costs as one without. Why would I ever design one then that lacked these features? Indeed, on the harder levels you can expect the AIs to design their own ships with triple hull and battle pods, but they won’t do so consistently. By ensuring that all my ships incorporate these two excellent features, I get ships that have increased defensive capabilities and more space to add components. This in turn gives me the first edge over AI ships as mine are more robust in aggregate.

The next consideration that occupies my thoughts is subtle yet vital. I think ahead of time of the tactics I want to use in battle and the fighting style I want to employ. For me this is to engage at long range, put out as much effective damage as possible as quickly as possible, and still have the ability to counter any tactics used against me by an AI. Easier said than done… but doable nevertheless.

Let’s take a look then at my early designs. While I usually do build some temporary ships along the way, this is the first true class of ships that I will produce in mass. (Note, with normal or accelerated tech settings, I would skip this design and go for something better, but I play with a 25% research rate and that isn’t on option. In either case, this is an “early game” ship)




Doesn’t look like much does it? I even recall Adam calling something like this under-powered. Perhaps it is… by itself, but the thing is, it is never by itself. This Destroyer will be accompanied by at least two Frigates. Those Frigates will have the same exact layout minus the heavy lasers. As soon as possible, a Cruiser is added. The Cruiser has the same layout as the Destroyer but with a third heavy laser and bomb racks. (In the early game when battleships are hard to come by, I install bomb racks on Cruisers so I can bombard colonies if the need arises). What happens then is that I am perfectly poised to engage early AI designs, even if they have better tech. My missiles can be concentrated if need be, and because they are fast, they will usually hit a target before any of my own ships are hit. The heavy lasers aren’t much, but they too can snipe away at long distance, something most AI designs cannot do. As the enemy moves closer (while I hold my ground), they run into my missile barrage even quicker and the damage output of my heavy lasers increases due to the reduced range.

At this early stage of the game I don’t score many kills as the AIs will withdraw their superior ships once it becomes obvious they are about to lose some. The main weakness of these ships is the lack of credible PD weapons, but by keeping them close enough to support each other, I helps somewhat. The funny thing is that these ships are most vulnerable to AI ships that are configured in a similar manner. No matter, these designs are meant to buy me time and allow me to have a ready pool of ships to upgrade…to this



This is the second class of ships I design, what do you notice? Yes it has better shields and armor (as it should), but the philosophy behind the design is exactly the same as the first iteration. The two heavy lasers are gone, replaced now by a single heavy particle accelerator. At first glance it may seem that I only have a marginal increase in damage output from the main guns, but remember when I said effective damage? The particle gun is shield piercing, does not suffer reduced damage at range, and is completely immune to Neutronium armor coating. True, it isn’t as accurate as a beam weapon, but by this point I also focus on ever improving targeting algorithms throughout the game.

Now see the missiles. The Nukes are gone and replaced by Polaris, but the configuration stays the same. Ammunition stays at 5 (you don’t really need more than that with my tactics), they are fast, incorporate ECCM, and the warheads are MIRV. The damage output from the missiles is now double compared to my first design and they are more robust in general. The PD weapons have also gotten a boost and this design can cause serious trouble for AI ships. If the AI stands off, I hold an advantage, if he closes, I hold an advantage, and if he uses combat jump… I rip him apart with point blank salvos that are almost guaranteed to hit. In this part of the game the Frigates lack the accelerator and Cruisers have double the weapons total, but no longer incorporate bomb racks. The racks are now placed on Battleships which have double the firepower of a Cruiser. Underpowered or not, when working in tandem, this class of ships can handle quite a lot.

Further designs see better kinetics, Flak cannon as PD, better missiles (perhaps even Photons if the game goes really long), but the underlying design philosophy does not change.

Before concluding it is important for the reader to understand that I am not advocating these specific designs, but rather the overall philosophy behind them.

Make your ships as robust as possible (always use triple hull and battle pods)

Make your ships complimentary to one another, ( for example, it doesn’t help if some of your ships have combat jump but others don’t, or if some have long range weapons and others don’t),

Anticipate the kind of ships that the AI will throw at you and create balanced designs that won’t be at a disadvantage against any particular tactic or ship design.

In the end, this is an area where the AI can always be outclassed by the player, and that is inherent to the game. It would be boring if every AI put out the same optimum design, I understand that. But it does leave those factions fielding an assortment of ships (some of which may be quite good), that are not complimentary to one another in fleet actions.

Next up... there is no next up. I'm going to take a break for a while and prepare a new generic early game walkthrough for new players. I had posted one two years ago but the game has changed enough since that time where I feel it is outdated.