How to make a great automation game

Factorio just released version 2.1, its last major update. Or at least that's what they're saying now. Terraria has been releasing new "final" updates every other year since 2011, so only time will tell.

Factorio started the automation games genre, which is now well established with many great games. This is a great time to review the whole genre all at once, what works and what doesn't. And even Factorio got a few things wrong, as the genre was still new back then.

What makes a great automation game

In automation games, players spend time doing 5 things:

1. figuring out and automating new things
2. identifying bottlenecks
3. dealing with bottlenecks
4. waiting
5. bullshit

The best automation games are those where you do mostly 1, 2, and 3, and keep 4 and 5 to minimum.

Figuring out and automating new things

You want to build some new product you don't have yet, or are only making manually. Or you want to use some new more efficient process to get the same product, like foundry based steel instead of furnaces. Or you're setting up new logistics system like a rail network. Or a different combat system like a wall of flamethrower turrets.

This is the part all automation games get right. There's so much variety. Automation usually starts with belts carrying simple items, and machines doing 1-to-1 and 2-to-1 manufacturing, but then it goes into all kinds of crazy directions.

And whenever I think I've seen it all, I try another automation game or mod, and it turns out there's some completely new way to automate things! Dyson Sphere Program lets you solve electricity production by building a ring of solar panels around the equator, so half of them are always in the sun. On Cerys you produce plutonium by funneling cosmic ray particles with charged rods so they hit your stored uranium as many times as possible. On Rubia strong winds prevent belts and inserters from facing West, so you have to build your entire factory from West to East, throwing any excess into the wind.

A very important thing is that this is a one-way progression. Once something is automated, it stays automated without any further effort. The game should never ever break what you automated.

I can't think of a single game or a mod that got this part wrong, the range goes from fairly good to totally amazing.

Identifying bottlenecks

The second thing is identifying bottlenecks. You make some science packs, but you want to make more. You seem to be making enough green chips, but they're stuck on a slow belt and some of the destinations are not getting enough. Your factory is suffering from power outages, so you need more power. You need more space for your factory. You produce too much waste product, and for now you just dump it into a chest and haul it out manually, but you really need to address it properly.

Automation games have a mixed record here. Some bottlenecks are very easy to identify. Belt based designs make it especially simple - empty belt means you need to produce more, overfilled belt means you need to consume more. Or when producers are idle as belt is full, while consumer is idle as belt is empty, you need to increase belt throughput. That's one of the reasons belts are so popular - it's so easy to figure them out!

But often you aren't really given tools to figure things out. How many items per second or per minute a given building produces? Most games including Factorio 2.x now show you that on a tooltip but back in Factorio 1.x you had to do the math yourself. How many accumulators do you need? There's still nothing in game to tell you, you can either find out some ratios online, or add more when you run out of power.

Unfortunately there's a conflict between these two activities. Making processes more complex makes automating new things more exciting, but it can also make identifying bottlenecks more frustrating, and players should be able to stay in the game and figure things out, not feel forced to look for online tutorials.

For more exotic ways of production, you often see that things aren't working well enough, while having no good idea why. Why is my Gleba factory working sometimes, but then it just randomly dies and turns into a cesspit of spoilage? Nobody can tell. Why is my asteroid reprocessing loop getting stuck, while an identical system elsewhere keeps working perfectly? Who knows really. Why my wall can hold biter attack most times just fine, but then they break through once in a while? The game has no numbers to help you, so you just blindly tweak something and hope it's the right thing.

Dealing with bottlenecks

Once you've identified a bottleneck, you need to deal with it. Sometimes it's easy, just add a few machines, upgrade them to next tier, add more belts, and so on.

Sometimes you see that your current approach just isn't working, and you need to automate something new to really address the bottleneck. Maybe your belts just aren't good enough and you need to get some trains. Maybe there's no way to get enough blue chips on your new planet for all the rocket launches, and it's just easier to setup a space ship bringing them from another one where blue chips are abundant. Maybe your gun turrets aren't holding with highly evolved biters, and you'd rather add some flamethrowers to your walls instead of trying to double down on your current solution.

Sometimes you need to get creative. You identified that you need more space. Do you clear some biter nests and build a new wall? Make a landfill factory, then landfill a lake? Bring cliff explosives from Vulcanus to flatten an area that's unused for now as cliffs are making it too awkward to use? Or use a different design that's more space efficient, like adding speed beacons?

And most of the time, you just need to make a bigger version of what you already build, or copy it over a bunch of times.

This part needs careful balancing, and a lot of games do it quite poorly. If you make it too easy, just copy&paste whatever, and it just works, then you don't really have bottlenecks. But if you know what needs to be done, but it takes hours to actually do, it can take fun out of the game, and make it feel tedious.

This is where I feel Factorio got the balance wrong, making scaling up before construction bots just infuriating. Placing 1000 belts? Just run and drag, and it's done in seconds. Placing 100 assemblers? Run and drag, done in seconds. Placing power poles? Smart placement will make sure nothing is disconnected. But then you have to go through ridiculous and error prone tedium of connecting it all with inserters that takes the fun out of the game.

Then you finally unlock construction bots, and the whole problem is completely solved, but at the same time you also unlock logistics bots, which trivialize the game.

There are other operations you pretty much need construction bots for, like moving parts of your factory a bit because you didn't leave enough space, even though you keep telling yourself to leave more space over and over.

Most post-Factorio games got it right. You get equivalent of construction bots either immediately or very early, but logistics bots very late. And many popular mods like FasterStart give you a small number of personal construction robots either right away, or at least very early.

An interesting if simplistic example of an automation game is Shapez, where resources are infinite, machines are free, but there's just one thing that takes a special resource and that's copy&paste cost, so you can't just make your factory infinitely big with a few clicks.

And these are three things automation games should be about. Automating new things, identifying bottlenecks, dealing with bottlenecks, in a loop until you win. Unfortunately...

Waiting

A very common anti-pattern is forced waiting. All automation games want you to produce big amounts of something to progress, a lot of research packs, or a lot of machines, a lot of rocket launches, and so on.

It's quite easy to find yourself in a situation where you feel stuck - you can just wait a few hours doing nothing and then progression just happens, but there's not much you can even do until then that feels meaningful.

This is not the same as dealing with bottlenecks. For example Dyson Sphere Program has "Mission Accomplished!" tech that "wins" the game (you can of course continue as a sandbox). This requires a huge number of Universe Matrices, which are produced very slowly by your Dyson Sphere. There's no good way to speed it up, and whatever bottlenecks your factory has won't matter after you "win" the game, so the best strategy is to AFK for a few hours. Absolutely terrible. Or at least that was the case when I last played it, it's still an early access game, so they might have changed it by the time you're reading it.

Shapez is notorious for this, as every research requires a different item, many of which aren't used by anything afterwards, so you really don't have any incentive to scale things up. And as for doing the next item while you wait for current research, that research often unlocks a new building you need for the next item.

Compared to most other automation games, Factorio is really good at not doing that - technologies that would hard block you tend to be cheap or even unlock automatically with a trigger. The game is also complex enough that there's always a bunch of things you want to improve about your base while you wait. You could still wait for technologies that make your factory more efficient, but they don't require unique items you'll never need again, and they don't block you from progressing.

A few Factorio mods like Cerys got this wrong, and force you to wait for some item like a huge number of blue chips to progress, and Cerys prevents scaling up until very far down the tech tree. Cerys is overall a great mod, but it could use some numbers being tweaked a bit. At least you can remote view into another planet and fix things there while you wait.

Bullshit

And finally, the last category, bullshit - everything that doesn't make your factory better. Especially pointless grind, but there's a few other types of bullshit too. There are games where grind is acceptable, but automation games are not it.

Factorio was clearly designed before it was understood, and it still has two main bullshit grind activities - fixing resource patch depletion, and dealing with biter nest expansion after you clear them up.

And they know perfectly well that it's bullshit. You can litearlly disable biter expansion on campaign start, it's disabled on some presets, and most new world (Vulcanus, Fulgora, Aquilo, space asteroids) don't have any equivalent of biter expansion.

They also know perfectly well that resource patch depletion is bullshit. Space Age made resource patches deplete at 50% speed with common big mining drills (down to only 8% with legendary), and most of all it made mining productivity techs dramatically cheaper.

In Factorio 1.x, a level of mining productivity costs about 940k in minable resources, and 1300k oil.

In Factorio 2.x Space Age, a level of mining productivity costs, between foundries, electromagnetic plants, biolabs, and cheap mining drills, costs about 18k in minable resources, 33k oil, and a bit of rocket launch capacity to send calcite over (it could be zero if you mine it in space).

These changes kept resource depletion in the game, but by mid game it's about 100x less of a problem than in Factorio 1.x. Because they know perfectly well that it was bullshit, but they couldn't quite make themselves just remove it.

Satisfactory is full of grind early game as well (again, I played it during early access). You're going to spend more time gathering fuel and bringing it to power plants than actually automating for very very long time.

Dyson Sphere Project has notable non-grindy type of bullshit - flying between planets. The whole space flight system is highly unintuitive and you can find yourself randomly stranded in space if you didn't quite hit the target, forcing you to load the save.

Most automation games unfortunately have some kind of bullshit. Cut it all out.

Summary

And that's how you can make a great automation game or mod. Have a lot of fun things to automate, provide good tools to identify bottlenecks, don't make fixing bottlenecks too tedious, cut waiting and bullshit to minimum.

Crusader Kings 3 landless play or Paradox games are not always easy even when OP

Paradox games are really complicated, but once you get through all the complexity and know what are the right options and what to ignore, they're normally very easy. Experienced players sometimes do silly things like "Hisn Kayfa very hard no allies no loans" in EU4 to get some kind of a challenge.

But recently I had a chance to play a total noob game.

I had ridiculous amount of hours in CK2, and I played CK3 a bit, but it never managed to pull me in the way CK2 did. It did some things better, some things worse, UI and map were a huge downgrade, overall I just didn't care, and I have limited gaming time.

Even the long overdue map expansion to China and Japan weren't all that impressive, as there are already CK2 mods that do that, which I played a lot.

But at some point they released ability to play as a landless character, and I had to give it a go. Well, at soon as I found some time, which was about a year and many nerfs later.

Robin Hood

All my attempts were custom characters in the earliest start date.

My first try was going in basically blind. I created a custom Anglo-Saxon Catholic Robin Hood character, way over points limit, as I had no idea what I was doing so I thought I'd make it easy for myself.

As it turns out, going over point limit was absolutely unnecessary, but I'll get to it later.

You can choose your camp purpose, so I decided to be a bandit, and I was wandering around, doing contracts, earning some money, and it mostly felt like I've been getting absolutely nowhere. Years passed, and I was still having my tiny bandit band with shitty characters, modest amounts of gold, and very little prestige to show for it.

The only interesting thing I managed to do in all this time was getting a good wife. At first I wanted to use character finder to search for some unmarried woman with good traits and romance her. But CK3 character finder is just shit compared to CK2's. You can't filter by age range other than "adult vs not adult", so most of the matches were some automatically generated grandmas. You also can't filter or sort by distance. There's "diplomatic range" filter, but it seems to have nothing to do with romance scheme range. The game wouldn't let me start a scheme unless I was basically in the same county, and I'd need to stay close to it for a year.

I tried to do this anyway with all my landless characters - find some lowborn woman with good traits, move my camp nearby, and start a romance scheme. But this really isn't the way to go. Game doesn't generate that many unmarried women with good traits early, you probably don't have skills for romance scheme early, and it ties you to the same place for a year or two. It sort of works, but it just isn't great. And I had multiple targets just die on me to a random event bufore my schmeme completed. There's probably some events in the game to cull random landless AI characters to keep game performance acceptable, and my potential wives kept dying to it.

The easiest way to get decent wife as landless character is to just get "rescue a fair subject" contract (justiciar category), which is very easy, and then you can just ask that rescued woman (or presumably a man) to stay with you. This is what I ended up doing mostly.

You can also just visit all taverns, and they have decent chance of getting some kind of "seductress" character you can bribe with some booze to join your camp. The downside here is that they are often quite old (but you can still get them to join for traits).

You can absolutely do the romance scheme, or even elope with rulers' wives, after you get your skills and lifestyle perks a lot higher, which isn't that hard, it's just going to take time.

Anyway, as Robin Hood, I met some good traits wild woman by a random event, did a prowess check, and we ended up getting married.

Contract Types

There's a lot of contracts you can get but they belong in mainly there categories:

• scheme contract - you and 5 of your agents start a scheme, which needs some time before you can roll dice if you win
• transport contract - go to another province, get reward
• skill check contract - go somewhere, do a skill check, if you win, you get a reward

There are so many problems with what I was doing:

First, most contracts are schemes contracts, so I was mainly taking them, and they are absolutely terrible this early. They take extremely long time, so I could only do a couple per year, and they can fail anyway.

Second, contracts have tiers, 1 (offered by counts and below), 2 (offered by dukes), and 3 (offered by kings and above). Rewards for tier 1 contracts are relatively poor, and you only get access to higher tiers if you have enough prestige. How to get prestige? By doing contracts. So if you're stuck doing tier 1 scheme contracts, you'll die of old age before you get access to good contracts.

Interestingly, rewards aren't even that bad by standards of a landed ruler - you can easily get higher income than a king by running a camp poorly and doing really bad tier 1 contracts. But costs are completely unreasonable. Recruiting a random mediocre character costs 100-150 gold. You can buy a whole county building for less. It's baffling that they didn't scale both income and expenses by 5x or 10x less to get them more in line with the rest of the game.

What you want is to do as many contracts as you can, and who cares if you fail. Skill check contracts are best for it, as they are super fast. You probably don't have great traits at this point, and you don't have great agents, but failing a few contracts doesn't really matter.

The next best are transport contracts. You just need to go to some different county, and free money and prestige. The main downside is that you don't know where you're supposed to travel until after you accept the contract, and reward has no relation with distance, so sometimes it's too far and not worth it. But a few times I had travel contract to "travel" to the very place I already was at (but contract giver wasn't). Free gold for 0 days.

I'll get to correct meta later, but at this point I had no idea.

Benjamin Shapiro and the Medieval healthcare system

I ended the Robin Hood playthrough early, and I checked some youtube videos and reddit guides. There are problems with checking what people say, and I'll get to it. But I was getting nowhere, and being a bandit had big downsides if you want to become landed later, so I thought I might as well restart and do things properly.

I wanted to try another meme character, so I made an Ashkenazi Rabbinic Genius character, starting in Venice, and tried scholars camp purpose. I thought I'd be spreading some Judaism maybe.

Anyway, Ben got Consumption before even reaching his first contract. As an established ruler it doesn't matter too much except for some lower health, but as a landless adventurer, it prevents you from visiting any settlements, gathering provisions, and doing basically anything else to make progress.

I was waiting for it to pass or something, but it didn't for years, so I eventually just gave up. I guess God didn't want me to play Ben Shapiro just yet.

This is a pretty shit game design. There's probably some ways to deal with consumption and other diseases, but if it happens so early, you might just as well restart.

Also Ben Shapiro losing to bad healthcare system, the irony.

Marco Polo

OK, so another start. Catholic Italian Marco Polo in Venice, the goal is to travel to China and come back.

At this point I was actually getting the meta a bit. I was avoiding slow scheme contracts, taking skill check contracts, and transport contracts, and I was slowly moving towards China.

I got annoyingy cancelled by the Abbasid Caliph for no reason, which meant I wouldn't be able to get any contracts there. I'm not sure if there are any downsides of just passing through, but either way I turned North towards Central Asia.

I was slowly meandering towards China, and I took a skill check contract to kill some bandit. I had quite high 26 prowess at this point, so I just went for the fastest option to duel him, and then I suddenly died, and got a game over. Well, that wasn't expected at all. My opponent also had a fairly high 24 prowess, and somehow the RNG favored him.

Duels are one of the CK3 features I really dislike. They depend on prowess and traits of both characters and some RNG, which is fine, but the game won't tell you "you have % chance to win / lose / die" like it does with almost all other events. You need to alt tab to google sheets to get that information. If consequences of not knowing was some minor prestige reward then whatever, but this is literally life or death situation for the campaign, and it should not be obscured this way.

Of course in the future I'd know to only select this option if I had way higher prowess, and otherwise take one of the other options.

Contract meta

Anyway, here's the landless contract meta as I eventually figured it out:

The most important thing is to stack all travel speed modifiers - from lifestyle traits, camp upgrades, and good camp officers. Then get some travel safety modifiers. They're not as important as speed, as you can avoid dangerous places. There's customize route option, and you should absolutely take it - it can make your travel faster (default planner really hates taking sea ways even when they're a lot faster), and safer (by going around dangerous areas). Eventually you'll need custom routes to hit "points of interest" too, I'll get to that.

Then as you're upgrading travel speed, just take all the travel contracts you can. If they're really inconvenient, you can keep them open for very long time, they don't seem to expire for years, and they don't cost much when cancelled.

With all the speed upgrades you'll be moving really quickly and collecting rewards all the time.

Whenever there's no good contract available, first ask town crier for more contracts. Especially transport contracts are great. If town crier doesn't have transport contract as an option, you can either pick some other category, or go to another town and reroll (there's a timer until you can ask again, so don't ask if you don't want to take any contracts). I'm not completely sure which other contract types for town crier to pick. Justiciar and stewardship tend to have some quick skill checks, diplomatic and intrigue tend to be slow schemes.

Criminal contracts are a bit of an all or nothing thing. You can do them and get a lot of very good rewards, but you lock yourself out of many decisions like buying land, at least until you clear your reputation. So either go all in, or don't do them at all. Doing them only sometimes is probably the worst option.

Stacking travel modifiers is useful for all other contract types as well, not just travel contracts. You'll get more high tire contracts of all kinds, you'll get more points of interest, and you'll be able to just get where you want to be a lot faster, and with a lot fewer bad events on the way.

And if you don't see contracts you like, just keep on moving. Between town crier and just moving on, you will find many good high tier contracts eventually (if your prestige is high enough for higher tiers).

Once you get enough prestige and unlock tier 3 contracts, you should go somewhere where there's a lot of king tier rulers close together like India. Many parts of the map (like Europe and the Middle East) have only very few big king/emperor tier rulers, and a lot of vassal and independent counts and dukes, who are somewhat lot less profitable. But that's not something to worry about early game, and even tier 2 contracts are mostly fine.

For skill check contracts, just pay to hire any good characters. You can use Scout for Talent travel option to get more events for them cheaply, and you can hire some for settlements (but they're more expensive). Don't worry if you fail (other than literally dying or getting wounded) if all it costs you is some prestige and opinion. Just keep traveling super fast and doing more contracts. If someone's bad at everything, don't bother hiring them.

For scheme contracts, they're not normally worth it as they take too long. However, there are multiple upgrades (lifestyle traits and camp upgrades) that let you do them faster. Unfortunately each such upgrade only applies to a specific kind of a scheme. Once you gathered a lot of good characters, and some faster scheme upgrades, these contracts get pretty good. In any case, go for as fast as possible option, don't spend twice the time to go from 80% to 100%.

Benjamin Shapiro the second

I could have reloaded Marco Polo autosave, but now that I finally figured out how to play, I wanted to give Ben Shapiro another go.

This was a start under 400 point limit, and it really didn't matter. If you visit every tavern and every training grounds on the way, you'll get insane skills in no time. I was the highest skilled character in the game less than 10 years in, and that includes Alfred the Great of Wessex, which the game sets up with insane starting skills.

It's especially easy to get ridiculous amount of prowess, by buying weapons and armor (I got +8 and +5, but there are better) at any random town market, prowess training at any castle training grounds (+6 before the option stops appearing), and blademaster training (+12 total, it starts showing up after flat prowess bonus ends). I started with very low prowess intellectual character, and had 55 prowess in less than twenty years, probably higher than anyone in the world, without even really trying to focus on that. Other stats increased a lot too, but somehow not quick as fast as prowess.

Somehow I managed to get over -100% stress gain, so all the events where I'd be getting stress were giving me 0 stress. I guess facts didn't care about my feelings.

You can farm prestige by doing a lot of contracts quickly, basically via travel speed.

You can farm piety by visiting every church on the way and doing tiny donation, at stupidly good exchange rates.

Sometime during this playthrough I discovered that you can visit "points of interest" to get rewards, especially lifestyle points. The UI for it is atrociously poor, but this lets you unlock all perks pretty quickly. This isn't infinitely farmable, as you only get a reward the first time for each point of interest, and it really encourages you to travel all over the map instead of staying in one area.

You can get some more traits and skills from university visits and pilgrimages too - similar to what a ruler would have access to, but a lot cheaper.

I suspect there are ways to improve other traits even further, as I've seen some screenshots with well over 300 skill sum total characters, my Ben Shapiro has "only" 175 total and 23 lifestyle perks), unless that got nerfed somehow.

But since this was already going so great I decided to look for options to get some land.

Getting landed and the nerfs

There are many ways to get landed. The boring one is to inherit something, but that only applies if you're a former ruler who temporarily lose their title, and managed to murder scheme his way back to the throne. Or if you managed to marry some royal in line of succession. It doesn't really apply here if we want to play a custom character.

You can buy land. Becoming a count is cheap, and it's not too hard to convince a king or a duke somewhere to sell you some. There are also events where you can get a county as a reward for joining someone's war. But who wants to be just a mere count?

Buying a duchy is a lot more expensive (2600 gold instead of 400 from what I've seen), and kings take a lot more convincing, but it's nothing too excessive.

You can convert your adventurer camp into a nomad camp, and seize some pastureland county somewhere, then continue horde gameplay.

There are also some ways to buy estates in administrative state, but I have no idea how that works exactly.

But what I really wanted is to conquer something. I've seen some video guide of people getting ridiculous armies as adventurers super easily - which cost nothing except provisions to run.

So I tried to hire some men at arms, and join some wars. But wait a moment, why am I not getting results anything close to what those youtubers were getting? My soldier count is tiny, they can barely fights, can't siege, and even if I win, rewards are miserably low?

Let's check 1.13.2 patch notes:

Reduced knight limit from Roaring Campfire significantly.

Removed the MAA max size modifiers from excellent aptitude camp officers.

Reduced adventurer government knight limit by 8 (10 -> 2)

Reduced the “Max MAA regiment limit” that you could get from your adventurer camp by 4 (across multiple buildings).

Reduced the “Max MAA regiment size” that you could get from your adventurer camp by 5 (across multiple buildings).

Reduced the individual MAA max size modifiers from the Baggage Train camp building by 2 (3 -> 1).

Added an upper limit to gold reward from the “Stand with Us” contract of 1 yearly income of the employer and their current gold.

All together that's like a 80%-ish nerf to amount of soldiers you can have, and much lower rewards on top of it.

This really surprised me. They didn't really nerf travel speed, skill gain, lifestyle points gain, but they applied heavy nerf hammer to transitioning out of landless adventurer back into landed.

Actually they also reduced gold, prestige, and provisions rewards a bit:

General contract gold rewards decreased by about ~30%

Travel contract gold rewards decreased by about ~25%

Crime contract gold rewards decreased by about ~16%

etc.

But this isn't anywhere close to the massive nerf hammer MAAs got, as you can just do more contracts by travel speed.

DLC Nerf Cycle

This isn't the first time it happened. Paradox often releases a DLC, it's stupidly OP, everyone has some fun, and then the next patch they nerf it into the ground. Usually it's some some specific feature getting nerfs, not whole way of playing.

The closest similar case I can think of were CK2 Republics, which were completely broken OP on release, then two months later they nerfed them with trade post limit, demense size nerfs, removing best plots etc.

I'm not saying nerfs are wrong, but they nerfed a very specific part of adventurer gameplay (being a mercenary and getting land), and not really other things (fast travel, max skills, easy provisions, gold, prestige, piety, and lifestyle points), and they made all the old guides break.

Guides and meta

And it brings me to the last point. I'm still playing this basically blind. All the guides and videos were created back when the feature was released. As far as I can tell, nobody bothered to release new ones post-nerfs, and the game doesn't really tell you anything. Paradox games never do.

What I'm saying here isn't even any meta, I only played it a few times, figured out what still works and what doesn't anymore, but I'm sure I missed a lot too.

Men at arms cost provisions per province moved, and since the meta build is all about stacking up travel speed, this amounts to very high provision cost per year. Apparently there are ways to stack provision cost to -100%, then it doesn't apply, but there still doesn't seem to be much point in having a small army as landless. They're just a cost center without any benefit. Go big or skip the army.

I suspect the right call is to save a lot of money (like 10k+), then only hire man at arms last minute before going landed, after you have all your camp upgrades for a bigger army already built. If rewards from joining wars got nerfed so hard, there's no point joining wars while adventuring.

One thing that seem especially nerfed is getting the Conqueror trait, to move from landless to conquering spree. It requires having a huge army and winning a lot of wars, which used to be pretty easy pre-nerfs. After nerfs, it probably needs alt tabbing to google sheets.

Using AI for French Reading Comprehension Practice

For family reasons, it would be helpful for me to learn some French. There are all the boring ways like textbooks, audiobooks, flashcards, and apps (Duolingo and its clones got really enshittified and focus more on pushing ads than language learning these days, another reason not to use them).

But we live in the LLM era, and language is one thing LLMs excel at, so it would be silly not to try that.

Written French

Languages differ quite a lot, and French is a bit of an outlier. For most "normal" languages like German, Spanish, or Polish, the written and spoken forms are very similar - the correspondence is never quite 1-to-1, but it's close enough you don't need to think about it much.

French is absolutely not like that. Written French is closer to the language people spoke back in the 1100s than to the French they speak now. In many ways, written French sits halfway between Vulgar Latin and Modern French.

The most egregious examples are verb conjugations: written French preserves all the 1100s endings, but in speech they've merged into just a few forms. So in writing there's "chante", "chantes", "chantent" (all pronounced /ʃɑ̃t/), "chanter", "chanté", "chantez" (all pronounced /ʃɑ̃ˈte/), and so on.

Fortunately this spelling isn't arbitrary way, it mostly just reflects a much older form of the language, so you can mostly figure out how a word is pronounced based on its spelling. It doesn't work the other way, a single spoken word can correspond to a lot of different written forms.

English suffers from a similar issue, but it's only about half as bad in terms of time passed and degree of sound change.

How to even learn French

Normally when spoken and written languages are far apart, it's easier to learn the spoken language. This is not the case at all here. If you know English, or any Romance language, you'll have fairly easy time recognizing a lot of words in written French, but when spoken you might just as well be hearing Korean.

And there's input hypothesis - the idea that the only meaningful language learning is by consuming a lot of input just above your current level, and output is pretty much a distraction.

Regardless of how accurate this is, reading comprehension is the easiest thing to practice with AI.

The Prompt

After a lot of experimentation, here's the prompt I settled on, followed by detailed explanations.

Do the following:

• write a short text in French of about 500 words total
• the text will be split into 10 paragraphs of about 50 words each
• the story needs to be at B1 level, only using grammar and vocabulary appropriate for that level
• write the first paragraph, then I translate it to English, then you let me know if it's correct or if I made any mistakes
• then continue with next paragraph until the story ends. Do not ask for confirmation before continuing.
• then print the scorecard
• do not nitpick English translation by suggesting a more natural one, as long as it translates French text correctly
• number each paragraphs
• use sidebar for French text

Topic for today: ...

Prompt explanations

Do the following:

• write a short text in French of about 500 words total
• the text will be split into 10 paragraphs of about 50 words each

First, we tell AI what we'll be doing. Translating a short 500 word French text to English is a good test that you understood the input.

A short text gives a lot more natural context than translating disconnected sentences, and is closer to natural input.

Dividing it into pieces makes it much easier to translate, and avoids having to scroll back and forth you'd need to with a single block of text. It has an additional advantage that if you encounter a difficult word early in the text that's repeated later, you'll get feedback on that word early.

• the story needs to be at B1 level, only using grammar and vocabulary appropriate for that level

It's good to give the AI some idea what level you're aiming for, but this is going to be very vibes based. The AIs don't really understand what's appropriate for each level, and you will see rare words - sometimes because they make sense for given topic, but often AI will throw a rare word even if a much more common word with similar meaning would work just as well.

I tried many variants of this instruction, and at least it generally sticks to appropriate grammar if you request a low level, without using rare tenses and such.

• write the first paragraph, then I translate it to English, then you let me know if it's correct or if I made any mistakes

This is the core idea. Technically "comprehensible input" doesn't require you to translate it, but translation ensures that you actually understand it, and gives instant feedback on any words or phrases you got wrong.

Another thing I tried was reading comprehension multiple choice test, but that's somehow more time consuming that translation (mostly due to scrolling required), and you can often answer multiple choice questions correctly even if you don't quite understand the text, so it wasn't all that useful.

• then continue with next paragraph until the story ends. Do not ask for confirmation before continuing.

Without this, AIs would sometimes proceed to the next paragraph, but sometimes they'd force you to say "continue" 10 times per story, which got really annoying. This solves the problem.

• then print the scorecard

The scorecard is not essential, and every time it will be in a different format anyway. If it was possible to do fine-grained adjustments to difficulty level, it would make sense to develop consistent scoring system, and then tweaks difficulty by +5% or -5% whenever it gets too easy or too hard, but you will get pretty wide range of difficulites with the same prompt, so there's not much point in doing this.

• do not nitpick English translation by suggesting a more natural one, as long as it translates French text correctly

Here's another big problem with current AIs. This is meant to be a reading comprehension practice, not translation practice. So ideally we'd like AI to pick up every error in our understanding, while not nitpicking about trying the most natural sounding translation.

Trying to fine tune the English text to sound great takes longer than just getting the meaning across, and it's a wasted time.

I tried many variants of this instruction, and this one causes mistakes both ways. Sometimes it ignores errors in understanding if I picked a vaguelly similar meaning (like translating "broderies" as "decorations"), and sometimes it nitpicks (nitpicking example: “small city” → better as “small town” (more natural for petite ville)), but it's balanced enough.

AIs also really love to say "good job translating this sentence: ...", which is basically noise, and perhaps it would be worth tweaking this prompt to reduce the noise.

• number each paragraphs

Most AIs do it anyway, but sometimes they don't. It's a bit more convenient to have paragraph numbers.

• use sidebar for French text

This is helpful to include for AIs like Claude that have sidebar interface. Don't include this instruction for other AIs, they'll either ignore it, or just make up some "sidebar" tags like <mui:sidebar> that don't actually do anything.

Topic for today: ...

And finally a really essential part - you need to give it a new topic every day, otherwise you'll see very similar stories about how a random textbook French person spent their day over and over. The topic can be one word like "cats", a phrase like "Bridezilla's wedding", or a short sentence like "competition to make the most ridiculous pizza", and it doesn't really matter what you put there, as long as it's different from the previous ones.

You could also ask AI to pick an interesting topic as well, which increases your odds of getting a non-textbooky text.

Which AIs to use

I tried it with a lot of AIs, free tiers only.

• ChatGPT - consistently best, but quite prone to going into censorship mode if it doesn't like your topic (especially on anything related to current events), something I never saw with any other AI with the topics I tried
• Claude - consistently very good, and sidebar interface for French text is very convenient
• Mistral Chat - somehow really good at it, biased towards slightly higher difficulty
• Grok - very good if you have access to it, but fairly slow message limit in free tier
• Qwen AI - I only tried it a few times, seems to be fine, but I'll pass judgement for now.
• Gemini - I have no idea why, but Gemini consistently disappoints me no matter what I try, not really suitable for this
• Meta AI - completely unsuitable for this
• Deepseek AI - completely unsuitable for this

All these AIs get regular upgardes, so your experience might be different from mine.

If you do this repeatedly (I aim at 2 stories per day), you can reuse existing chat and just say "next topic: ..." or even "pick a new topic". Some AIs have tendency to make the new text overly similar to the one it just did, so starting fresh conversation be better.

I don't think it's a good idea to use any kind of automated "memory" feature, as that will just cross-contaminate conversations.

What else didn't work

The prompt here is a result of a lot of experimenting. One notable thing all AIs are absolutely terrible at is IPA. Asking for IPA you'll get total vibe garbage - like IPA of completely unrelated words.

How you should treat the results

This is a reading comprehension exercise, not translation exercise, so don't worry if your score isn't perfect. It's not even supposed to be.

• typos - try to get better at typing, but you'll likely make a few typos here and there. AI will point them out, but they don't matter, just ignore them (if they were legitimate typos not mistakes).
• English text not sounding natural - it would be waste of time to try to polish it, so don't worry too much about it.
• comprehension errors - if you miss some words or gramatical structures, and they feel rare, that's totally fine. In real life you'll encounter small percentage of more difficult words too. If you miss common words, you're just getting very effective spaced repetition here, in proper context.

5 comprehension mistakes per 500 word text (so excluding typos and nitpicking) is a 99% understandig, and that's what you should be aiming at. If you make a lot more mistakes, adjust the vibe level. If standard CEFR levels are not enough, you can make up your own levels like "A2+", "B1.5", or whatever you want. Just remember the level you give the AI is not really CEFR level, and some AIs (notably Mistral Chat) have a notably different idea what each level means.

How to configure OSX 26.0 Tahoe for software development

With every new Macbook, I'm updating the guide, previous version is here.

When you start installation, one of the questions is about FileVault, so just enable it before we even start. Software updates moved to a new location in Settings > General > Software Update, I recommend just applying all updates before we get started.

During the installation you'll be spammed by stupid popups like "Do you want to allow <app> to <do something>" or "Turn on Accessibility for <app>". You just need to accept them all. Apple tries to apply phone security model to laptops, and it's doesn't match how things are used at all.

Basics

• Install some sensible browser like Chrome, Firefox, or Brave. Chrome used to be the universal default, but with all the anti-adblocker drama, other browsers are worth considering too. You'll get some popups asking to make whichever one you choose your default browser, do it.
• Afterwards either sign up into your account on which you hopefully have your ad blocker setup, or install some. For Chrome best you can do now is uBlock Origin Lite, which is less powerful than manifest v2 uBlock Origin used to be, but for now it seems to work well enough. If anti-adblock technology manages to break adblocking, just switch to another browser.
• Install a sensible terminal emulator. There is only one sensible choices - iTerm2. ghostty tries to be an alternative with some extra features, but lack of ⌘-F style functionality makes it simply unusable. If you start iTerm2 for the first time, it will also prompt you to install XCode Command Line Tools, which you'll definitely need, so just do it now.
• Install whichever cloud sync service you're using like Dropbox etc. And start syncing your stuff. Unfortunately even the most expensive Macbook laptops still have tiny disks (starting at 512GB, in 2025!!!), so you might have to do selective sync only
• Clean up all crap from dock. Other than Apps (previously Launchpad) and System Settings, everything else should be gone. Add iTerm2, your browser, and your text editor, and any application you wish to install there instead of stock Apple crap. Apps you use only occasionally shouldn't be there.
• Remove crap from your desktop, they recently started adding widgets .

Editor

Install some sensible text editor. These days most people use Visual Studio Code. If you do, go to Options, search "Telemetry" and disable it all.

If that's your choice, run it, open Command Palette, and choose: "Shell Command: Install 'code' command in PATH".

Settings

Like every other operating system, OSX has a lot of bad default settings. Here are some obvious fixes:

• If you have multiple monitor setup, go to Settings > Display > Arrangement and drag and drop them into correct arrangement so mouse can move between them correctly. To make Spaces work correctly, you'll also need to set your external monitor to be the main one by drag and dropping menu bar to it.
• You might need to do it twice - with laptop screen open, and with laptop screen closed.
• Also set up which will be your main monitor by dragging that white bar on top of the display icon to it. This looks like Menu placement, but it really mostly controls Dock placement.
• Settings > Appearance > Dark. If you're setting up a new laptop, this will be asked during installation.
• Settings > Keyboard > Key Repeat > Fast (max is correct)
• Settings > Keyboard > Delay Until Repeat > Short (max is correct)
• Settings > Keyboard > Disable keyboard brightness completely. Defaults (slow keyboard, highlight keys) are meant for people who are bad at typing. If this somehow applies to you, get some typing lessons, you can save huge amount of time by getting better at typing.
• Settings > Keyboard > Text Input > Edit ... > Disable "Add period with double-space" - this one really messes up coding
• Settings > Trackpad > Scroll & Zoom > Disable "Natural scrolling". This will also apply to the mouse, restoring correct direction.
• Settings > Sound output > Disable "Play sound on startup"
• Settings > Sound output > Disable "Play user interface sound effects"
• Settings > Sound output > Alert volume > 0% (for Terminal ping)
• Settings > Desktop & Dock > enable "Automatically hide and show the Dock"
• Settings > Desktop & Dock > disable "Automatically rearrange Spaces based on most recent use"
• Settings > Displays > Max out brightness
• Settings > Displays > Turn off "Automatically adjust brightness"
• Settings > Displays > Turn off "True Tone"
• Settings > Menu Bar > Battery > Enable "Show Percentage"
• Settings > Menu Bar > Clock > Clock Options > Use a 24-hour clock. This might be already on based on your regional choices during installation.
• Settings > Mouse > increase scrolling speed and tracking speed a bit
• Settings > Lock Screen > Turn display off on power adapter when inactive > Never. You should generally be doing it yourself, and you often need to leave the laptop running upgrades or unit tests or such, and you want to be able to see the status without constantly poking it.
• Settings > Lock Screen > Require password after screen saver begins or display is turned off > After 5 minutes. You can move it down to 1 minute if you use your laptop in public. I don't recommend Immediately, as that causes endless annoyance with connecting and disconnecting external monitors and just moving laptop around requiring new password.
• Settings > Sharing > Remote Login > Turn on
• Settings > Sharing > Remote Login > (i icon) > All Users
• Settings > Sharing > Remote Login > (i icon) > Allow full disk access for remote users (this seems to be default now)
• iTerm > Preferences... > Profiles > Terminal > Unlimited Scrollback

Press Ctrl-Up arrow, add a few desktops (or "spaces" as they were used to know), then go to Settings > Keyboard > Keyboard Shortcuts > Mission Control - and enable their keyboard shortcuts Ctrl-1 to Ctrl-6 or however many you have there. Sometimes I've seen shortcuts for extra desktops not being there, and in such case I just restarted, and the problem fixed itself.

Open Screenshot app, choose options, then:

• disable "Show Floating Thumbnail"
• Save to > Other Location... choose "Downloads" folder

Drivers

OSX already includes drivers for laptop itself, but you might need some for peripheral hardware.

In particular, external PC keyboard need a tweak to work properly, as left and left Windows keys are in reverse order from Mac keyboard.

Go to Settings, Keyboard, Modifier Keys..., choose the right keyboard from the dropdown (strangely I had ordinary wireless mouse selected by default), and swap positions of Option and Command keys. Feel free to change functionality of Caps Lock key as well, it's a huge easily accessible key with no useful function people love to remap, usually to extra Control.

New Macbooks now come with Fn/Globe key. You probably don't need i to change your keyboard layout so feel free to use it for emoji keyboard or something like that.

If you need any special keyboard layouts, get them too.

Another thing - when you plug in external keyboard, you'll get choose keyboard type dialog. It will likely choose the wrong type. Just pick ANSI, whatever it claims to detect. Otherwise the backtick key will be wrong.

Home/End keys on OSX are also broken. Use this as a fix. You'll need to log out and log back in for it to take effect.

Development tools

You'll need a package manager, and the only one anyone uses is homebrew, MacPorts and the rest died long time ago. You need to tell homebrew to not spy on you with brew analytics off command.

You'll need Xcode. Fortunately iTerm does it for you automatically, and if not homebrew will. If you need to do it manually for some reason, you can install Xcode manually by running xcode-select --install from command line.

Deal with stupid access popups

Since Big Sur, the first time you access some folders from terminal, you get a stupid popup asking you to confirm that you're indeed fine with terminal accessing various folders. So run:

  find .

and confirm all those stupid popups to be done with it once and for all. Well, except you'll still have them when accessing USB drives and such. 

And just to be extra sure, go to Settings > Privacy & Security > Full Disk Access, and add iTerm there.

Create new SSH key pair

Before you do that, name your computer something memorable with sudo scutil --set HostName your_host_name command. You probably go through a lot of laptops, so names like "Name's Macbook" are completely useless to you. Just pick a theme like cats or dinosaurs or whatnot, and give every computer a distinct name.

Open Terminal and run ssh-keygen to create ~/.ssh/id_ed25519 and ~/.ssh/id_ed25519.pub then upload the generated public key to any place that needs to know about it like github, bitbucket, or whatever else you use. Don't upload your private key by accident. These used to be called id_rsa a while ago.

Alternatively you could copy your keys from your old laptop, but it's generally more secure to have separate fresh keys for each machine.

Checkout your dotfiles

Hopefully you're storing your dotfiles somewhere. If it's a git repository, or your Dropbox account, get them now and symlink them all properly.

If there are any other repositories you might need, checkout them too.

Standard paths

OSX renames a lot of directories. The most annoying of those is that instead of /home it has /Users. It used to be very easy to add a symlink, but this kept getting more and more complicated, so I stopped doing this.

Install homebrew packages

Your list might vary. Here's a few obvious suggestions:

brew install rbenv ruby-build mc wget p7zip trash git htop bash zsh yt-dlp jq imagemagick coreutils bash-completion zsh-completion nodenv pcre

Then enable all services you installed, unless you want to start them manually:

  ln -sfv /usr/local/opt/*/*.plist ~/Library/LaunchAgents/

And install non-system ruby, so you can install gems without sudo. Currently I use 3.3 as quite a bit of software did not get updated to 3.4 yet:

  rbenv install 3.3.9
  rbenv global 3.3.9

To make that actually work, you need to make sure ~/.rbenv/shims is in your $PATH. If you type rbenv init, it will tell you what to do.

There's also asdf that offers this kind of service for all languages, if you want to use it instead of rubenv/nodenv/etc. I don't recommend rvm, I've seen it cause too many issues in the past.

Due to OSX limitations you'll need to run sudo htop if you want to use htop.

Install gems

Again, your list my vary. These days most of the software will have its own Gemfile so long list of gems are generally unnecessary. But some global utilities are still useful:

gem install bundler rak pry pry-rescue

Different Shell

OSX switched from ridiculously outdated bash to up to date zsh, so it's no longer a mandatory step.

If you want to use system zsh, it's fine.

If you want to install something else, like proper bash (or brew version of zsh; or something else), first brew install bash.

You'll need to edit /etc/shells as admin and add the following lines at the end of it to enable your newly installed shell:
/usr/local/bin/bash
/usr/local/bin/zsh

Then set it as your shell, with whichever one you prefer:
  chsh -s /usr/local/bin/bash $USER
  chsh -s /usr/local/bin/zsh $USER

Hushlogin

For some reason OSX prints worthless annoying messages on every open terminal tab. Run to touch ~/.hushlogin to prevent that.

Coreutils

This is optional step. OSX coreutils are generally a lot worse than GNU versions you might be used to from Linux. However switching means occasional minor incompatibilities, so it's up to you if you want to do it or not.

If you want to do so, brew install coreutils, then add GNU coreutils to your PATH:

  export PATH="/usr/local/opt/coreutils/libexec/gnubin:$PATH"
  export MANPATH="/usr/local/opt/coreutils/libexec/gnuman:$MANPATH"

OSX coreutils are not as bad as they used to (for example cp -a now works), so this step can be skipped.

Better window manager controls

Sadly OSX window manager is extremely dubious for keyboard use. Fortunately programs to make it usable exist. Unfortunately there's a lot of churn among those programs, and every couple of years the ones I use become unmaintained and need to be replaced by something else.

Currently I recommend:

• Rectangle - for moving windows around on big screens - I don't really like the default keybindings, so I delete them all, then setup Cmd-Control-Option with 1,2,3,4 for corners, with arrows for halves, and with M for maximize. Also set Repeated commands to "cycle through displays". You can install it with brew install --cask rectangle
• AltTab - for switching between windows - it's baffling that OSX completely lacks this function - and Cmd-Tab to switch between applications is absolutely inadequate for any application that has more than one window, which is most of them (browsers, editors, terminals etc.) if you're developing software. Turn off "Show apps with no open widows" in settings, as it really messes up things. You can install it with brew install --cask alt-tab

You'll need to give them necessary access. To do so:

• Settings > Security & Privacy > Privacy > Allow the apps below to control your computer > enable AltTab and Rectangle

Rectangle also doesn't start at login by default, so turn that on in settings.

Open files limit

For some insane reason OSX has default open file limit of only 256, and that breaks a lot of software like databases.

You can do it for processes in terminal by putting ulimit -n 100000 in your .zshrc, which might be adequate, but not every process runs from the terminal. 

Enabling it globally gets more and more complicated with every OSX version. Instructions for Ventura are here.

Lower security settings

Unix used to have very simple model where root user could do anything, and that was great for development. OSX keeps adding more and more security restrictions. They are absolutely detrimental to developing software, and of questionable value to regular users - primarily they're Apple's way of slowly turning computer world into something more like iOS world where they can decide who can run what and take 30% tax on everything.

You'll need to disable some of them. Most important such setting is this:

sudo spctl --master-disable

It will tell you you need to confirm it in settings, so go to System Settings > Privacy & Security > Allow applications from > Anywhere. This has nothing to do with security, it's just Apple attempt at turning OSX into iOS step by step.

If you want to use dtrace (OSX equivalent of strace), you'll need to follow these more complicated steps to disable SIP first.

WiFi settings

This setting wasn't there before. If you want your laptop to have fixed IP on your local network, you need to go to WiFi settings for your local network, and disable "Limit IP address tracking". This confusingly named setting is MAC randomization, and with it enabled, your router has no way to assign fixed IP to your devices.

You can leave it on for public networks. Or just don't touch it if you don't need fixed local IPs.

Android File Transfer

It's honestly embarrassing to both OSX and Android that there's no out-of-the-box way to move files between them either over WiFi or USB cable.

There's official Android File Transfer program but it's just awful. OpenMTP is somewhat less awful, but still not great. If you know of any program that's actually good, definitely let me know.

On Linux and Windows it's possible to mount MTP devices, which is very slow, but still beats OpenMTP. I don't know if it's possible on OSX.

All other software

There's a lot of other software you might want. The most obvious one is the VLC media player.

You might also want some kind of Git UI program, like GitUp (brew install homebrew/cask/gitup).

If you want to use SSHFS, the one in homebrew (macfuse and sshfs packages) don't seem to work, so you might want to try downloadable versions. First time you try to use it, OSX will block it, so you'll need to go to Settings > Privacy & Security, allow it there, and restart (you'll get popup for that).

Docker

There's a lot of software that just plain won't run on Apple Silicon, so you'll need Docker. You can either use Docker Desktop, or if you have licensing issues with that Rancher Desktop. If you use Rancher, set it to Docker compatibility mode.

Enjoy

Once you go through this list, and successfully get everything going, I'd recommend modifying it to your liking and reposting your version on your blog. Everybody's needs are different, so guide like this is just a starting point.

Empire Total War Campaign difficulty levels explained

This isn't explained anywhere else online, so here goes.

The data is based on data files from the latest patch. All of it comes from campaign_difficulty_handicap_effects_tables db table, so mods can adjust it.

Player Bonuses and Penalties

Difficulty levels rely mostly on buffing AI, so player penalties on hard and very hard are fairly small, and bonuses on easy aren't that big either.

Bonus	Easy	Normal	Hard	Very Hard
clamour for reform all classes				-1
cost - buildings - all	-10%
cost - recruitment - land	-30%	-25%	-10%
cost - town watch	-10%		+15%	+30%
cost - upkeep - land	-20%	-15%
cost - upkeep - naval	-30%	-15%
counterspying	+2
max population	+10%
max repression - government type	+1
personal security	+2
research rate	+15%		-5%	-12%

AI Bonuses and Penalties

AI bonuses on the other hand are very extensive.

Bonus	Easy	Normal	Hard	Very Hard
agent cap - eastern scholar		+1	+2	+2
agent cap - gentleman		+1	+2	+2
agent cap - missionary		+1	+2	+2
cost - buildings - farms		-8%	-20%	-30%
cost - buildings - happiness		-8%	-20%	-30%
cost - buildings - industry		-8%	-20%	-30%
cost - buildings - mines		-8%	-20%	-30%
cost - buildings - plantations		-8%	-20%	-30%
cost - recruitment - land		-10%	-20%	-30%
cost - recruitment - naval		-8%	-15%	-23%
cost - town watch		-20%	-50%	-90%
cost - upkeep - land		-30%	-40%	-50%
cost - upkeep - naval		-25%	-40%	-50%
counterspying			+1	+2
GDP		+10%	+20%	+40%
max population				+10%
max repression - characters		+1	+2	+3
movement points		+10	+15	+20
personal security	-1		+1	+2
recruitment cap - land			+1	+3
recruitment cap - naval			+1	+4
research rate		+10%	+35%	+50%

Diplomacy

Diplomacy is definitely affected by campaign difficulty, but the logic is hardcoded, and not exposed anywhere. On harder difficulties AI has a massive and very obvious anti-player bias.

What is not affected by difficulty

And that's all. A lot of things difficulty affects in other games (including other Total War games) are not affected by difficulty level. This means all of these are the same regardless of difficulty levels:

• starting armies
• starting buildings
• starting research
• starting diplomacy
• starting income
• fixed income per turn

How to overcome difficulty bonuses

Campaign difficulty modifies just a few areas, and in each of them, there are very strong counters available.

Research rate - AI on very hard has +50% while the player gets -12%. However AI typically only has 1 school, while you should build 4-6. If you have enough patience, you can also trade techs, something AI doesn't seem to do much. You will outtech all AIs with ease. AI's extra diplomats and scholars also come nowhere near matching the advantage you have by just having a few more schools.

For diplomacy bias, giving away military access to everyone on the map (at least everyone you border, plus countries like Britain and France who love to drop surprise naval invasions on you) really tones AI aggressiveness down to more reasonable level.

AI gets a lot of different bonuses which are basically money cheats. The best way around it is by being better at economy game. If you can, spam trade ships. And never have a big expensive army just sit there doing nothing, something that AIs routinely do. Your main army should go on an offensive. If you need long term garrison, just recruit some cheap crap instead, or units with garrison bonus.

You can also use AI bonus cheat against itself. Especially all the one province minors have nothing to spend their money on, so they upgrade all their buildings as soon as possible. And then you can conquer them. A maximally cheesy version of this strategy is to sell AI minors farm and industry techs they need to unlock the upgrades, and maybe even sell them your worst provinces that need a lot of money to turn around. Then conquer them once they're fully upgraded, and all you'll need to pay is the repair bill.

AI gets some happiness bonuses you don't, but it doesn't really affect you, as you're not trying to out-happiness the AI. You should definitely get better at happiness game, which for absolute monarchies means stacking your cabinet with ministers with as many lower class happiness bonuses as you can. But you should do so regardless of how happy AI provinces are.

AI's extra movement points can take you by surprise - sometimes enemy army is just barely out of your reach, so you think they can't reach you either, then you get surprise attacked when you press end turn. As long as you're aware of it, it isn't usually a big deal. And AI movement cheats have nothing on daisy chaining ships to transport your stacks from one end of the theater to the other in one turn.

And this just leaves AI's huge +3 recruitment cap bonus on very hard (+1 on hard it's no big deal). It can outspam your units with ease. The trick is that while recruiting new units is capped, replenishment is unlimited and just costs money and two turns. So if you have a lot of badly battered units after a big battle, as long as none of them got wiped out, you can get your whole stack back really quickly - something that wasn't possible in earlier Total War games.

Another useful trick is just recruiting a lot of new units in far away lands, and using ship daisy chaining to transport them to where you need them to be.

Naval recruitment cap is technically even bigger, but AI doesn't really spam ships anyway.

Autoresolve

Empire's autoresolve is a black box. There are some difficulty-related settings in campaign_variables_tables, but what they do exactly is anyone's guess:

• autoresolve_easy_campaign_AI_percent_reduction - 0.05
• autoresolve_easy_difficulty_human_advantage - 0.1
• (no settings for normal)
• autoresolve_hard_campaign_AI_percent_increase - 0.2
• autoresolve_hard_difficulty_AI_advantage - 0.08
• autoresolve_very_hard_campaign_AI_percent_increase - 0.35
• autoresolve_very_hard_difficulty_AI_advantage - 0.2

As far as I can tell, autoresolve bonuses are controlled by battle difficulty, not campaign difficulty.

You can mod these to -0.99 if you want free autoresolve wins for some testing. Or if you really hate naval combat so much, and plan to only use it for naval combat.

Battle difficulty

I wanted to provide similar data for battle difficulty, but I couldn't find any game files controlling it. This likely means that the battle difficulty is implemented as part of game logic, and possibly in a way more complicated than just simple stat boosts.

We have some idea what is affected the most. AI troops on hard and very hard get a huge morale bonus, which is a huge nerf to player's melee units, especially melee cavalry, which normally relies on breaking enemy units' morale, and has poor staying power in prolonged battle. Bonuses to accuracy and reload speed are more modest, and for artillery difficulty levels seem to matter the least.

However, it's difficult to put any numbers on these qualitative statements.

And that's all we know about Empire Total War difficulty. If you find out any additional numbers (by reversing executable file?), let me know.

How to build most overpowered custom nations in Europa Universalis IV 1.37

Custom nations have been in the game since El Dorado DLC. When setting up a custom nation, everything costs points, and there are some targets like 200 for a "normal" start, 400 for "easy", and 800 "very easy".

This was always very poorly balanced, and initially custom nations bonuses were weak. But EU4 went through a lot of power creep, and some of the recent custom nations bonuses are really overpowered.

To make OP custom nation, the best way is to make geographically smallest country that's viable, and put all the point in ideas. Viable might mean a single province on most of the map, but for some particularly hostile areas, you might want to make it a bit bigger.

You don't really get 200 points. There's no reason not to take always duchy rank (-10 cost), 0/0/0 age 0 heir (-24 points, will get disinherited immediately), and 0/0/0 consort with irrelevant negative trait like Loose Lips (only -4 points, but might as well take it). You should also take bad trait for your ruler that doesn't matter like Loose Lips (-20% foreign spy detection) or Malevolent (+5% Liberty desire in subjects), but that only gives you -2 points. So your starting budget is 240.

Unfortunately we can't just take 240 points worth of ideas from the menu. There are 10 bonus slots, but slot costs is multiplied by 2.0, 2.0 (two traditions), 2.0, 1.8, 1.6, 1.4, 1.2, 1.0, 1.0 (7 ideas), and 1.0 (ambition). So if we spread idea points evenly, average cost is 1.5x, so we have 16 points to spend per slot to fit our 240 point budget (a bit less as you need at least some land). If you frontload cheap ideas, and end with expensive ideas, you can get a bit better value, but it will obviously make your early game harder.

There's also huge penalty for taking too many levels from the same category. Try to avoid it. Sometimes getting extra levels of ideas makes total cost cheaper! The most broken ideas are in admin or diplo category, so it might be challenging.

You can also increase your budget by taking a negative idea for negative cost:

• Can NOT declare war for -100
• Can NOT build buildings for -40
• Can NOT send merchants for -40
• Can NOT build over force limit for -25
• Can NOT establish colonies for -10
• Can NOT send missionaries for -10

You should always stick these in your 2.0x slot. The alternative of sticking them in your ambition and never taking 3 full idea groups is honestly not worth it. Of these, the only one I'd really consider is the force limit ban. Colonies/missionaries are not worth many points, and the rest are too harmful to you. Sticking a ban on building over force limit as your first tradition gives you 50 more points, and it won't even apply very early game when you're most likely to do this.

Custom nations can form other nations. This is the only way to get new mission trees, some with a lot of overpowered bonuses as mission rewards. But even if you form a new nation, you can't change ideas you originally selected.

Anyway, let's get to the completely broken ideas.

Overpowered Economic Ideas

For 60 points you get flat +2 goods produced. That's equivalent to 10 dev per province, including gold mines. None of the other economic modifier are even in the same ballpark. If your average province is 3/3/3, it will get +333% production income, +333% trade value (it goes into trade node, so you still need to capture it), I have no idea what they've been thinking. For comparison with some pre power creep modifiers, 18 points gets you 20% production efficiency, and 30 points gets you 20% trade efficiency. If you don't have points budget for it, you can get +0.5, +1, or +1.5 for proportionally less points.

Overpowered Military Ideas

Military bonuses are mostly either crap or really overcosted. The game wants you to pay 18 points for crap like -20% recruitment speed. Or for something more ridiculous, +2 leader fire or shock costs ridiculous 140 points, not even remotely worth the cost.

There are just two military ideas worth looking into. The less broken one is Triple manpower increase in religious wars for 50 points, which is basically triple manpower always, as you can always find some infidel OPM to "fight".

The second one that's even better are +1 Infantry Fire and +1 Infantry Shock for 25 points each. These are completely unrelated to unit pip stuff. It's just ridiculous early game. Normally fire and shock of military units depends on tech, and at tech 3, infantry has total of 0.85. If you got both bonuses in your traditions, your infantry will be doing 2.85x damage instead, or +225% more damage. It just melts enemies.

This is somewhat balanced by two things. As game progresses, base shock and fire values for all units increase, so by the end of tech tree, infantry has 5.25 total, or 7.25 with both bonuses, which is respectable but less ground breaking +38% increase. A much bigger factor is that your army composition changes from all infantry to half infantry half artillery, and artillery doesn't get any bonus. By end of the game backrow artillery deals a bit less damage than front row infantry (fire + shock is 8.95, but only half of that counts for backrow units; disregarding pip differences), so this bonus is worth about +20% more damage by 1821. Not game-chaging by then, but still respectable.

Due to the way combat math works with fire and shock phases switching back and forth every five days, both bonuses are almost equivalent, except fire phase happens first, so Infantry Fire +1 is a bit superior to Infantry Shock +1. This is disregarding pip distribution and such, but they don't really change much. Obviously if you use this build, delete all your cavalry.

The obvious question is about doing it for cavalry and artillery. It's mostly just worse, but worse version of ridiculously overpowered is still pretty good.

It's a flat bonus per unit, so for cavalry you'd get same extra damage for 2.5x as much unit upkeep. This only makes sense if you build all cavalry armies (and eventually cavalry + artillery armies). If you mix cavalry with infantry you're watering down your bonuses even more. And penalties for too much cavalry are so high, that there's no way taking cavalry fire and shock without 100%+ cavalry to infantry ratio is a good idea. Base is 50% and you can take cavalry to infantry ratio bonuses in custom ideas (+60% for 18 points), and you and get some from religion (+25% pure Tengri, +10% Sunni) and various government reforms like Steppe Horde. It's mostly a weaker version of infantry bonuses, but it's still strong, going from +200% at tech 3 to +33% at 1821 all-cavalry army, or +19% for 1821 cavalry/artillery army.

Artillery bonuses are noticeably worse. Because artillery is normally in the backrow, you'll only be getting half the bonus, so it gets +0% for early game armies, by midgame if you switch to half inf/half art armies at tech 16 it peaks at +28% (infantry bonus would be +56% at this point), and only +10% by 1821.

There's one exception to this. If you build your armies with just artillery, with no infantry, and no cavalry, the bonus will be +190% damage at tech 7, gradually falling to +22% in 1821. But your artillery will also take double damage. It's a bit of a meme build, but these ideas make it sort of viable.

And yes, for both cavalry and artillery, +1 fire and +1 shock are pretty much the same thing.

Overpowered Province Warscore Cost

Province Warscore Cost is a modifier that's fairly mid when you have a bit of it, but stupidly overpowered when you stack it near -90% cap. It turns out custom nations let you do this stacking really easily. 25 points will get you -25% province warscore cost, 18 points will get you -25% province warscore cost against other religions, and then you can take Diplomatic (very decent idea group) for -20% more, having -50% discount in 1444, and -70% discount as early as 1460s. At this point you really want to keep stacking it, and Age of Reformation gives you −25% more against other religion, already taking you over the cap. Other good sources are Malta monuments that go to -15%, and Mecca monuments that go to -10% but only for Muslims. You could also get −10% for military hegemony.

AIs will be giving you land practically for free, but you still need to get your AE and OE management game. This build relies on your religion being different, so picking Catholic or Sunni is notable counter-production. There's no real reason to pick Sunni anyway, Shia and Ibadi are mechanically almost identical to Sunni, and picking either gets you access to Mecca monument, so it's a great build.

Switching from Catholic to another Christian religion is more costly, as Catholic is by far the strongest Christian denomination since Emperor DLC (usually whoever got the most recent DLC is the strongest). Christian Europe is also a relatively poor way for going ham on province warscore cost, as you'd be facing coalition of the whole continent in no time. But even with these issues, it's still not a bad build. You'll be getting up to -90% against other religions (Council of Trent provides the last −10%, if they vote right) and -45% against other Catholics (-25% custom ideas, -20% from diplomatic ideas).

Overpowered Monarch Point Savings

All Power Cost. You can get literally -10% all power costs for 60 points. By midgame you'll be making about 13 points per month of each kind (3 base, 1 power projection, 5 advisor, and thanks to disinheritance and abdication monarchs average 11-12 points so that's about 4 of each type), so that's about 45 monarch points saved every year.

Administrative Efficiency. 10% costs 60 points. It's best if you stack it, unfortunately you'll need to wait for Age of Absolutism to start really stacking it. Until then you can get 5% more from Alhambra monument, and some more from mission trees.

Administrative Honorable Mentions

-20% coring cost reduction for 30 points. This would have been a top tier idea, but power creep went so far it's merely an honorable mention now. CCR is obviously good, and it's even better if you stack it, so take administrative ideas for −25%, and then you can squeeze the last percentages from your religions (Hindu can get the most -10% for Shiva plus another -10% for Kashi Vishwanath Temple monument) and mission rewards.

+1 colonist for 30 points. The first colonist is far more valuable than additional colonists, and you can get one for really cheap. This used to be somewhat limited, as you had to wait for maps, but nowadays if you have a colonist, you can get free explorer from your estates and do exploring, so that problem is completely gone. It depends on your location, but you can get far ahead on colonization if you put this in your traditions. On the other hand starting too early you might struggle with colonial range.

+20% goods produced for 30 points. This is decent economic modifier, and it becomes a lot more valuable once you get manufactories going everywhere. It's nowhere near as good as flat goods produced - if average province is 3/3/3 (0.6 goods produced), +20% bonus gives you +0.12 early game, and +0.32 after manufactories, while flat +1 for the same price gives you +1. You can stack percentile bonus and also flat bonus, and you'll run out of things to spend money on halfway through the game.

-20% minimum autonomy in territories for 18 points. This also applies to trade companies, but you can stack it with -20% from economic hegemon, -10% from expansion ideas, typically -10% from government reforms, for trade company autonomy cap of just 30%, or 7x the manpower and tax you'd be getting normally. You could get it even lower with Hindu or Confucian monuments.

+40% Governing capacity modifier for 30 points. You can normally keep up with governing capacity with technologies, estates, and buildings even when expanding rapidly, but bonus you can get here is surprisingly large. The most any normal nation gets is +15%, and even that is rare.

Diplomatic Honorable Mentions

-20% Aggressive expansion impact for 30 points. Not quite as good as −25% that Ryukyu gets in base game. This modifier is best when you stack it, so definitely take Espionage ideas for -20%. It's mediocre normally, but it's really useful for this build. Age of Discovery has ability for -10%. Maxed out prestige is another -10%. The Grand Palace of Bangkok monument is -10%. Many religions can top it off. Orthodox can get -10% from icon, Muslims can get -10% from Hanbali scholar, Hindu can ge -5% from Shiva deity and -10% more from Kashi Vishwanath Temple monument. Catholics can get -10% from a bull, and -20% from curia control, but you can't count on consistently having either.

-20% diplomatic annexation cost for 30 points. This is best if you stack it. You'll need to take Influence ideas for −25%. Estate privilege Nobility Integration Policy gives you -5%. Influence-Administrative policy gives another -15%, and you're probably taking Administrative anyway, so that's already -65%. Pushing it further is harder. Influence-Quality policy is extra -10% but Quality is a weak idea group you probably wouldn't want normally. Curia controller is −10%, but that's not reliable. Parliament can have an issue for -15% if you have two or more annexable subjects. Quite a few countries have a discount as mission reward, so if you want to form some nations on the way, that's your best way to reach the -90% cap.

+30 Vassalization Acceptance for 5 points. This is equivalent to 10 whole diplomatic reputation, for this one interaction. This is way better if you stack it, so you should definitely take Espionage for +15 more bonus. You'll probably want Influence for 2 diplomatic reputation (+6 more bonus) and cheaper annexation cost, the game has a lot of other diplomatic reputation bonuses to pick up. Going all into vassalization game leads to a very different gameplay than normal. If you want to try this in Europe, you should consider destroying HRE and getting the pieces as vassals. The main limitation of this is diplomatic relations limit, there's some DLC countries with subjects that don't count towards the limit, there's probably some potential for even more extreme build there.

Military Honorable Mentions

+20% siege ability costs 30 points. Your armies typically spend more time sieging than fighting, and for this reasonable price, you can make them 20% better at it.

+0.5% yearly army professionalism for 10 points. You can get more up to +2% for 40 points, but this is one of the modifiers where stacking is actually not recommended, as once you hit 100%, it won't do anything (except recover faster from slackening). The best thing about max professionalism is +20% siege ability, which is very nice mid-game, but you don't need to rush it. Taking a level or at most two is good use of points.

Not Recommended

These seem really powerful, but the cost is just too high.

Siberian Frontier for 200 points. This is a great ability, but it's priced very high, and you want to take it early (in 2x cost slot) so it's really 400 point of your 800 very easy budget. To make it worth the cost you'd need to either start in Americas, or move your capital to Americas. And if you're playing natives, many recently got some speed colonization mechanics anyway.

Immortal ruler trait for 800 point. Immortal 6/6/6 is very tempting, but the cost is too high, so leave it for roleplaying. For 800 points you can get such insane custom ideas, you really won't miss it.

Free Ideas

If you want to fit limited budget of 200 points, you'll need some filler ideas, as you can't take 10 overpowered ideas.

You'll also need to balance your ideas between categories. Weirdly the game checks how many levels you took, not how many points you spent, so level 2 for 3 points and level 2 for 140 points count for the same.

Here are some 0 point ideas suggestion that provide decent value, by category. You can double most of them for 3 or 5 points if you have a few to spare. None of these are game changing, but since it's a free filler, might as well fill it with something useful.

Administrative:

• +10% governing capacity
• -5% core creation cost
• +5% goods produced
• -5% administrative tech cost
• -5% minimum autonomy in territories

Diplomatic:

• -5% aggressive expansion impact
• -5% diplomatic annexation cost
• +5% trade efficiency
• +5% improve relations
• +15 vassalization acceptance
• -5% diplomatic tech cost

Military:

• +5% siege ability
• +0.5 army tradition
• +5% infantry combat ability
• -5% regiment cost
• -5% military tech cost