KickStarter Jon Shafer's At the Gates

[Infinitron]

https://www.eurogamer.net/articles/...iew-a-4x-experiment-more-fascinating-than-fun

Jon Shafer's At the Gates review - a 4X experiment more fascinating than fun
Explore, expand, exploit, existential crisis.

The former Civ 5 director's long-running passion project is filled with nice ideas, but they never threaten to pull together.

One of the first tutorial pop-ups you'll get in At the Gates - and there aren't many - warns you that this is a "hard, slow game". It's not the warmest of welcomes but that's not what At the Gates is about. This is a game about surviving gruelling winters and, slowly but surely, expanding your economy and influence against the odds. There's nothing wrong with that; hard games are often the most rewarding, and slow ones can be soothing. The problem, however, is that it's less hard than it is exasperating, and less slow than it is straight-up catatonic.

The setup in At the Gates is pretty standard 4X stuff: choosing a faction - with the only one available at the start being the vanilla Goths (more are unlocked by conquering or allying with them in-game, which is a nice touch) - and then spawning, with very little, in a random position. The structure, though, is surprisingly different: you get one Settlement for the entire game, and instead of buildings within it, all of your economy is built around people, or Clans.

At the Gates has fantastic tooltips, to its credit. I do love a good tooltip.

At the surface this is a neat idea - you're playing from the humble beginnings of a hunter-gatherer tribe, not a grand, imperial civilization - and so bringing the human element of early civilization to the fore makes sense. The problem is that in practice, as with so many things in At the Gates, it just doesn't quite click. You start with one Clan - a single unit - and you need to train them in a profession, like Reaper or Gatherer, that falls within one of six disciplines, like Agriculture. You need to unlock the profession first, by studying it from the tech tree, and then once you've studied it you can begin your little unit's training.

The idea, I think, is that this builds into a nice, efficient cycle of studying the tech tree and then training units in what you've unlocked, but in my time so far it's never really threatened to get going. To train a Clan in a profession they need to be housed inside your settlement's base building, where they're not doing anything - so you'll need to pluck core parts of your economy out of the economy itself to actually progress them.

Winter is tough, and it's very easy to get caught out - sometimes unfairly, however.
That slows things right down, of course, but more than that it often just leaves you with two equally unsatisfying options: make no progress in your economy (often at the risk of genuinely starving to death), or make no progress in the actual advancement of your society. Too often you have to choose the latter, keeping your extremely narrowly-specialised units out in the wild gathering resources of whatever kind that's available, for you to just about get by, whilst you languish behind the other more-developed societies in the game. Repeat reminders that you're not training anyone - !!! - mean you're made to feel almost guilty that your settlement isn't continuously re-tooling units into something else, but there just isn't the incentive to. There's no reason to change your Reaper to a Fisherman in a land-locked field of wheat, just because you can. Instead, it feels like the game is built for you to stagnate.

The other big headliner for At the Gates is its seasonal changes, a weather system designed to give the game a certain rhythm, which again feels like a small-but-smart shift in 4X design on the surface and, unfortunately, again feels like more of a shackling of the fun than an enhancement of it in practise.

Depending on where you start on the map, you'll feel the winter months to different effect - the in-game years take 24 turns to complete, with about 10 to 12 turns of winter - but you'll always feel it to some extent. As the year progresses from spring through to winter the map, a gorgeous, painterly thing, gradually evolves with it, until it freezes over, the vast majority of it painted a deceptively festive white. The impact for you is decidedly less cheerful. Many resources, particularly food, become ungatherable, so you'll need to stock up in advance. Some tiles become impassable due to blizzards, on top of already impassable or severely hindering terrain like swamps, flooding, or rain. The rest are suddenly deeply hazardous - your units often can't move more than one tile, and if they're outside the small area of effect of your settlement, they'll need supplies to survive, or they'll starve to death. Supplies are gathered simply from the tile they're standing on, and almost all of them have no supplies in the winter, so you'll need to plan well ahead: leave a Clan more than a few tiles outside of your settlement for the cold months and there's a good chance that, without careful micromanagement, they'll die.

Diplomacy is rather unformed in At the Gates. There are few options other than accepting or rejecting the demands of the AI as and when they come up.

It sounds tough, and it is, but where it moves from tough to exasperating is in the little failures of the UI in clarifying the consequences of what you do - no warnings or notifications that a Clan's struggling for instance, either in combat or in starvation, means that once you've acquired more than a handful it's very easy to lose one without even noticing. Multiple times I've had long-serving Clans that had ever so gradually built up experience and level in their discipline go on to die without even realising it. On another occasion, I tried to move one from its position just two tiles out of my settlement area back into safety, and it died on the one tile in between because it wasn't quite clear whether moving it would allow for the usual one-turn grace period where it can survive on low supplies or not. There goes five hours of training with it.

As for the rhythm, what this all means is that you effectively can't do anything for half of every in-game year - in other words, for half of your time playing the game. Every winter, you need to bring your units back to safety from the resources they were foraging, or micromanage them in short-term encampments. You could retrain them, seeing as you've already shuttled them back to the settlement, but then there's little point in that because there are only certain resources nearby, and only these bottom-of-the-tech-tree professions can harvest them. So you spend ten to twelve turns doing little more than clicking 'next turn'. Then maybe four turns just walking units back to where they were through the awkward terrain, and before you know it it's time to start prepping for that winter again - don't forget it takes even longer for them to get back!

Arguably the primary way you interact with At the Gates is through its tech tree, but with the pinnacle of your civilization being... fishmongers, it fails to capture that compulsive spirit of progress.

None of those gripes really get to the core of At the Gates' problem though: that it just doesn't give you a reason to keep playing. It's almost a survival game, asking you to find ways to hold out against the harsh world and worrying foes that surround you, but it just doesn't have the texture to make that compelling. There's not enough of a reason to stick with my silent, often belligerent and precious Clans through the tough times. There's not enough of a tease of what's to come, a game built around a tech tree - which should be a 4X fan's dream - but with the most glorious accomplishments of that enforced and seemingly endless teching being an upgrade of my Bard to a Minstrel.

After a long time grinding out the seasons you'll eventually find an equilibrium, where you just about get enough food and resources to keep ticking over, you've stopped having any kind of trouble with bandits, there's no reason to progress further up the tech tree to simply take your people out of their finally-just-about-working jobs, and the enemy factions don't really have any interest in attacking. That Roman Empire you're supposed to be overthrowing just keeps giving you some food every now and then for playing nice. At that point, naturally, you take a step back: to think about not just what needs doing next, but what point there is in you doing anything whatsoever, and that's where At the Gates hits a pretty impassable wall.

 

[thesheeep]

Hmmm.... looks a bit like he reached for the stars, but couldn't quite get there.
I wonder if, given half a year and some patches, it would become much better.

 

[Alpan]

I like how none of the reviews posted here so far mention Endless Legend, even in passing, when it comes to the winter mechanic. Including RPS, which declared Endless Legend their GOTY in 2014.

 

[oscar]

To me the bizarre thing is representing the Goths, Saxons, Franks etc as literal cavemen who in 400AD still apparently had yet to learn how to eat berries or skin a cow. The Dark Ages weren't that dark Jon..

The AI thing isn't much of an excuse either. I've played plenty of small budget indie strategy games with fantastic AI (much much better and more competitive than the ones John developed with far more time and resources). A self-described 'hard' semi-rougelike that apparently even game journalists easily won on their first playthrough is pretty damming. Seems he failed to learn his lesson from screwing the Civ series with (again) feeble passive AI and bloated convoluted features resulting in an AI that sucks at both fighting - and - producing.

Don't. Develop. Systems. Your. AI. Cannot. Use

Sent from my ANE-LX2J using Tapatalk

 

[Dayyālu]

Don't. Develop. Systems. Your. AI. Cannot. Use

If. You. Do. Make. It. A. Multiplayer. Game.

I'd argue that games likes Conquest of Elysium are guilty of the same sin (after all, the AI of "mortal" enemies in CoE is subpar) but then I remember that a pseudo-roguelike strategy game does not need an opponent working on the same rules as you, exemplified in CoE by Hell invasions and Planar enemies.

Also by deer.

I'm not particularly willing to be merciful to a designer because he was a weakling and trucked himself right into depression after having a shot at high-tier game development, particularly if the game he devoted his ruined life to is a boring mess with some Late Antiquity skin grafted on.

Barbarian Invasions was more interesting, I guess, and it was Hollywood history at its finest.

 

[GrainWetski]

I like how none of the reviews posted here so far mention Endless Legend, even in passing, when it comes to the winter mechanic. Including RPS, which declared Endless Legend their GOTY in 2014.

That game ceased existing in their minds when Civ 6 was released.

 

[astrosmurf]

I've played plenty of small budget indie strategy games with fantastic AI

Do tell!

 

[flyingjohn]

 

[Beowulf]

Any examples?

I haven't actually played it, but Pandora: First Contact (post AiL patches at least) is a commonly cited example.

If the developer was criticized for the AI it's because it was weak, as it is standard in 4x genre, but the moded AI was actually more difficult (and I can't imagine how a developer could catch some flak for that).
And the rumors about the AI opponents being actually somewhat competitive piqued my interest in the game (and I can see it still mentioned from time to time as an example of good AI in a 4x game on various forums).

 

[rezaf]

I didn't want to claim that the Civ 6 AI was (doing these things, because it was) smart.

I still think that most people prefer an AI which "feels" good (by acting like no human player would), not one which is purely competitive.

A agree, and I'm one of "most people".

The thing with a competitive AI is it's speed and brutal efficiency. There are no misclicks, there's no time spent scrolling around, there's no being occupied with a battle and thus not being able to properly handle that other battle or the production chain.
AIs can do everything at once.
Some RTS games have pretty decent skirmish AI, and you really have to be on top of the game to beat AI players even on middling difficulties. Sometimes you can exploit flaws in the AI or in the game systems themselves, but failing that, you have to really get your act together to even match their production speeds. I remember that it took me quite some time to find a sweet spot of skirmish difficulty in AoE3 where the AI didn't ROFLstomp me at least some of the time.
A couple of RTS games have made the mistake of employing this kind of AI in their campaigns, which can lead to some very, very frustrating moments and has occasionally caused ME to ragequit or even ragequituninstall games.

Anyway. In Civ, Jons idea to switch from a (very flawed) simulationist approach, with AIs that were basically glorified barbarians designed not primarily to WIN themselves but instead to prevent you from winning, or rather to represent a roadblock on your path to winning, to an extremely gamey design philosphy, with shallow systems that could be in a boardgame (I think Jon described it like this himself) and AIs that tried to win as any player would try to win in a boardgame.
But even in vanilla Civ5 with relatively few gameplay systems, the AI failed to properly play the game and could only provide a challenge when massively subsidized - it's evident how the high difficulties were reserved for really experienced and invested (minmaxing etc) players up until Civ4 - in Civ5/6 your five year old son or brother and your mom were/are playing on immortal - and winning, too.
Of course, the only reason this is possible is the retarded decision to go 1UPT in a global strategy game. 1UPT is nice by itself, and one doesn't have to point to hex wargames to find a working implementation - it even works in a more abstract, but otherwise quite similar game such as Warlock. Mobility is a key factor, it's annoying, tiresome and "CPU intensive" for a human player to control a carpet of doom where every individual unit can just move a single hex or two, how should poor AI deal with it?
Of course, it doesn't help that the approach to modern AI design is often enough "that'll do" after a bare minimum has been implemented, but surely the AI of SSI games in the early nineties weren't rocket scientists, either...

In the end, much of it comes down to what oscar said.

Don't. Develop. Systems. Your. AI. Cannot. Use

No matter how bad your AI is, you can build a game it's sufficient for. Many a FPS has gotten by with an AI that just beelined to the player on sight and shot whenever possible. Done.
Was it nice to be outflanked, yelled upon or having to pursue fleeing enemies in games like Half-Life or F.E.A.R.? Sure. But can one still have fun in Wolf3D, which does none of these things? You bet.
In a similar and more on topic fashion, playing Civ1 is still fun after all these years. It didn't even have the stack of doom problem (exaggerated as it may be) due to the simplest solution of em all.
Grampa Sid knew best, or so it seems.

 

[vonAchdorf]

It will be interesting to see, if - after a popular middleware is released - developers will lazily implement a machine learning AI, which gets trained by playing a million games against itself.

 

[baud]

In a similar and more on topic fashion, playing Civ1 is still fun after all these years. It didn't even have the stack of doom problem (exaggerated as it may be) due to the simplest solution of em all.

What prevented the stack of doom in civ1 ?

 

[vonAchdorf]

Because you killed all units on a square with a successful attack.

 

[Agame]

Because you killed all units on a square with a successful attack.

Was this a good system? It seems like the worst parts of doom stacks and 1UPT combined.

 

[rezaf]

Was this a good system? It seems like the worst parts of doom stacks and 1UPT combined.

Why?

The worst part of doom stacks is their existance, i.e. the tendency of players to assemble a pile of units in one square that is so big it spells doom, i.e. the "target" won't be able to stop it.
With the Civ1 system, it was very unwise to plop all your units into a single tile, since it could very well mean losing them all to a single successful attack. So you didn't do it.

The worst part of 1UPT is that it clogs down into traffic jams, where getting a unit where you want/need it turns into a giant, interconnected set of sliding puzzles. Civ1's system didn't have that, since while assembling stacks was discouraged by the mechanic described above, they were not prohibited per se, so if you needed to pile those units up to get through a narrow stretch of terrain, nothing stopped you (except the risk of losing all stacked units).
"Mild" stacking, two or three units, was still frequent, for example you could stack a unit with high defense and a unit with low defense but high attack. But it was a tradeoff.

(Note that there were two examples to the "lose everything on a lost defense" rule, Cities and Fortresses, where you only lost the defending unit and all other units were unaffected.)

The downside of Civ1's system was that it didn't come with the quality of life stuff "true" stacks have, mostly moving the entire stack at once ... but then again, 1UPT has this same limitation. And you only really need these things if you're frequently assembling stacks.

IMO, Civ1's system was simple and elegant, and it got the job done. Like I wrote, it's still a joy to use after all these years (I'm sad CivWin doesn't work on x64 systems or I would play much more frequently) - and fast, too.
Playing Civ1 almost feels like playing a casual game these days, everything is lightning fast compared to the late Civs.

 

[vonAchdorf]

I think it's telling that in the later Civ they constantly introduced features to patch either Stack of Doom or 1UPT deficiencies, e.g. artillery damaging more units in a Stack or the ability to stack some units again in the later Civs.

 

[Infinitron]

https://www.pcgamer.com/how-making-...ly-destroyed-civ-5-designer-jon-shafers-life/

How making a new strategy game nearly destroyed Civ 5 designer Jon Shafer's life
"I had nothing left, financially, physically, or mentally," Shafer thought, halfway through the 6 years he spent on At The Gates.

At age 21, Jon Shafer was asked to be the lead designer for Sid Meier’s Civilization 5. It was a dream come true for a young designer who had been creating mods for Civ, just a few years before. Within three years of shipping Civ 5, though, he’d quit lucrative jobs at Firaxis and Stardock and suffer the crushing reality of being an isolated programming prodigy with ADHD, trying to make his dream game. It all came crashing down in 2015. "I had nothing left, financially, physically, or mentally," he wrote. "The last shreds of creativity and productivity finally slipped between my fingers."

Since 2015 Shafer has been slowly building his life back up from ruin. He spent six hard years on his passion project At the Gates, finally finishing and releasing it this January. Here's how he got there.

Strategy savant
Shafer began making his own games at a young age, programming his own Pokémon RPG and Dragon Ball Z games. At the same time he was getting into history, reading his mom’s war books and playing games like Panzer General.

"My mom was a history teacher and a fan of the old King’s Quest and Quest for Glory games, and my dad was a programmer," Shafer says. "He taught me PCD3, which was used by United for Flight Training Software back in the 80s. It was old, weird stuff but it had a graphics editor that was kind of like Paint."

His passion for strategy games and history inspired a high school teacher to introduce a teenage Shafer to Sid Meier’s Civilization 2. "My first introduction to Civilization was with a pirated copy of Civ 2!" says Shafer. "I never did buy Civ 2 but in the end I think I gave something back to the series."

Shafer poured himself into Civilization, becoming a modder and joining online communities to run civilizations as real-world democracies. After Civilization 3 released, Shafer was studying programming and history at Colorado State and began bugging Firaxis for an internship. Eventually they agreed, and Shafer began his video game career in 2005. He was 20 years old.

"My first project was Civilization 4, but honestly I wasn’t doing much," says Shafer. "I wasn’t a professional engineer and nobody knew what to do with me. I wrote a tool in Python so you could take images from a Google search, like a map of Europe, and convert them into a Civ map. I also wrote scenarios for Civilization 4, like a World War I scenario. I got involved with whatever I could get my hands on."

Firaxis had a unique company culture that allowed Shafer’s seemingly boundless energy to flourish, provided he could work on new things that piqued his interest. During the development of the Civilization 4 expansion packs, Shafer was busy prototyping new systems and total conversion mods. Shafer says he developed the entire espionage system that was introduced in Civilization 4: Beyond the Sword ("It sucked, espionage in Civ game are always terrible") as well as The Final Frontier space scenario.

Firaxis then took his work on a Colonization mod and decided to turn it into a standalone expansion, which became Civilization 4: Colonization. Shafer is credited for the original prototype. By the time work started on Colonization, though, Civ 4 lead designer Soren Johnson had left the company, and Firaxis needed someone to head Civilization 5.

Jon Shafer's Civilization
"Civ games are huge undertakings and most people only have one of those in them," says Shafer. "I felt ready. I definitely wasn’t but I felt it. I had an energy I needed to use. With Civ 5 I wanted to do new things. Civ 4 was so amazing but I didn't want to make Civ 4.5. We’re going with hexes, we’re changing combat, and the entire economy."

Firaxis handed the Civ keys to Shafer, granting him, as he puts it, "dictatorial powers" over every aspect of the game. "Firaxis has a tradition of trusting its lead designers completely, going all the way back to Microprose with Sid Meier and Brian Reynolds," says Shafer. "It’s a big job, a hard job, but so many talented people did it in the early days."

Shafer claims that Civilization 5 was (and still is) the biggest evolutionary leap in the franchise, and it’s hard to argue otherwise. The initial reception was divisive among hardcore fans but the game was a critical and commercial success, and would become Firaxis’ best selling Civ game to date, though Civilization 6 is on track to surpass it.

In the end, shipping Civ 5 permanently eroded his passion for the series. "I was very young and didn’t have a ton of experience. I never thought about my career path or where I was going to be in 10 years. I just always had a huge project in front my face," Shafer says. "I worked 239 days in a row to finish Civ 5. I couldn’t look at the game anymore… I was worried I’d be working on Civ forever with Firaxis. I left a few months after it shipped. I made it, I did it, but I had to go somewhere else and work on something new. I still love it but I really can’t play Civ any more; the magic’s not there."

Having Civ 5 on his resume opened up lots of opportunities. Shafer had become a name for 4X fans and studios. He landed a job at Stardock, which was coming off the disappointing release of 4X strategy game Elemental: War of Magic.

Stardock didn’t operate like Firaxis, however. "I had ideas coming off Civ and wanted to explore them in my own way. But Stardock is like me, they want to do everything." Stardock was also a much smaller company that worked on more games, leaving little room for experimentation or autonomy.

Shafer would spend only a year and a half at Stardock, though his fingerprints can be found in Galactic Civilizations 3. "Some of the features like space terrain and adjacency bonuses for buildings on planets were my ideas that I had written up and designed." At Stardock Shafer held out hope that he’d be able to design a new game, but it never happened.

Then Kickstarter exploded.

Spoiler

Trouble At the Gates

Shafer left Stardock and convinced his friends and roommates to help work on his dream 4X game in their spare time. After six months of work, they formed Confier Games and launched a Kickstarter campaign in early 2013, raising just over $100,000 from 3,000 backers.

"Originally At the Gates started as a prototype of ideas," says Shafer. "A lot of 4X games have static maps, so I wanted to try cyclical map evolution, and seasons was the easiest way to do that." Shafer set the time period during the fall of the Roman Empire and founding of medieval Europe, and was inspired by Imperialism, a strategy game published by SSI in 1997.

"One of the big problems in 4X games is you quickly get bogged down when you have 20 workers and 30 cities, and you only really care about about four of them, but you keep building them up," says Shafer. "The game works against itself once you reach a certain point. Imperialism got around that problem by giving you only one city, but that capital city has a lot going on. I wanted to take that concept, along with the Roman theme and cyclical map changes, and started prototyping what would become At the Gates."

For about a year things went well. Shafer was finally free to pursue his dream 4X game. It didn't last. "In 2014 the game had several cool elements, but you weren’t building towards anything," says Shafer. "Part of the fun with Civ is seeing the visual element of your progress through roads and wonders and factories. We didn’t have that."

Inspired by Crusader Kings and King of Dragon Pass, Shafer added characters, clans, and professions to At the Gates. Professions became the entire tech tree. "It fundamentally changed the game and added a bunch of time to development. None of that was in the original game," says Shafer.

Looking back Shafer claims it was the right decision, but the extra work led him to drop life outside work, which would soon prove disastrous. "I had been diagnosed with ADHD around 2011," says Shafer. "I always kind of knew I had more trouble focusing than other people, but I was achieving a lot. I graduated high school in three years. I was the lead on Civ 5 at 21. I was never like, ‘poor me.’"

Shafer’s programming prowess and energy came at a cost; he didn’t know how to pace himself at work or how to separate work from his personal life. His social life began to crumble along with his physical and mental well-being. Around mid-2015 he stopped posting Kickstarter updates altogether.

He's uncharacteristically somber as he talks about this period of his life. "There was a couple years where I couldn’t do much of anything," says Shafer. "I stopped paying bills. It wasn’t just that this game was hard and I was feeling burned out. I was fundamentally destroyed as a person."

Today Shafer has come through a dark period in his life. When he was working on At the Gates, he started abusing his ADHD pills, gained more than 60 pounds, and spent all of his money, which he talked about candidly in a personal blog post this January. "It finally sunk in," he wrote. "It was over. Everything was over. I had destroyed everything. There was nothing left. Of the game. Of my career. Of my life. Anything. It was all gone."

Gradually he put himself on a regimented diet and daily routine that involved eating healthily, exercising, and working regular hours, writing "breaking out of that prison and rebuilding my life has easily been the hardest thing I’ve ever done." Ultimately one of the best decisions he made for his road to recovery was moving to Stockholm, Sweden to pursue yet another big name 4X video game developer, Paradox Interactive.

"With Paradox I wanted to be in a more structured environment, and not just sitting at a desk by myself. Plus, I had no money left."

While at Paradox, Shafer’s contract allowed him to continue working on At the Gates, and he soon felt like he was working two jobs at once. Six months later, Shafer would quit Paradox, largely for the same reason as Stardock. "I had my own games I wanted to make," says Shafer. "The expectations were completely different to how I was used to making games. The Firaxis and Paradox models are very different from each other. I know how to make good games, but I need an environment that gives me creative freedom. That’s what I was used to at Firaxis."

Paradox didn’t work out, but Sweden provided a new life for Shafer. "I made a lot of friends there—I have a Swedish girlfriend!" says Shafer. "I love it there. Living in a city is awesome. I grew up in the suburbs of Denver, Baltimore, and Detroit. Now I’m in downtown Stockholm and there’s so much to do and see."

Shafer is well aware that the tumultuous six years he spent on At the Gates wasn't sustainable or healthy. "With At the Gates I didn’t really have milestones and wasn’t very disciplined with scope," says Shafer. "I’m going to change how I do things from now on. The one guy in a room by himself, I did that to finish At the Gates."

Shafer plans on taking some time off after At the Gates' launch, with Dead Cells as the game he’s most excited to play. Afterwards he already has plans for his next game, ones he’s had for a long time, but he’s not ready to announce it yet.

When asked if he has any advice for young developers, he stresses the work-life balance and the importance of taking care of yourself. "This is the kind of business where folks, especially younger developers, really give it their all," says Shafer. "But it’s incredibly easy to burn out this way, and once you do that passion that drives you will go away. I’m not going to tell people how to live their lives, but look after yourself and establish healthy work and life habits."

 

[Dickie]

According to a thread on Steam, it looks like 1.2 is the version where AI might start to do something.

v1.2

The third major patch, mainly focused on gameplay balance, the user interface, and AI. Target date: Late May.

Then again, 1.3 seems vital, too.

v1.3

The fourth major patch, mainly focused on diplomacy, the late-game, AI, and modding. Target date: Mid-September.

Add New Diplomatic Options
Improve Diplomacy AI
Improve Late-game Strategy and Balance

I hope some people are still interested in September, so I can see how the game shaped up to be before I decide if I want to buy it.

https://steamcommunity.com/app/241000/discussions/0/1743358239834179395/

 

[oscar]

I've played plenty of small budget indie strategy games with fantastic AI

Do tell!

Hannibal: Rome and Carthage in the Second Punic War has a great (and full of personality as an outright game mechanic , it starts out over-competent and naive to stimulate the cocky guys who handed Hannibal his early total victories but gets a lot wiser in the mid and late game to represent competent generals like Fabian and Scipio taking the reigns). A fun and very focused single player experience (I think I won only by the edge of a knife with a single turn remaining too).

As mentioned Pandora was respectably solid and I played that before the expansion packs.

I'd also recommend Field of Glory II as possessing a AI that can give the player a run for their money (though I recommend moving to multiplayer ASAP as humans are a lot more fun and chaotic to fight than the competent but unimaginative AI).

Dominions series AI is a mixed bag (understandably as some factions are a lot more gimmicky and complex to win with than others). But even on unaided difficulty settings they can be a real pain in the ass with factions like Ermor.

Ultimate General: Civil War is another one man band strategy game with an AI that does the job rather well most of the time and will give even a seasoned gamer some nasty fights (though I also inflicted some outright blood baths on it at certain battles I think I was stronger than the game had expected for in the campaign, I was starting to feel bad for the Yankees. But even the first mission is pretty tough for both Union or Confederacy).

Stretching a ways back but I remember Panzer General and some of its clones and modern spiritual successors (Unity of Command I think was the name?) having a mean AI. Though I always found that tended to lend the battles a puzzle-ish tinge. Still would tick the box.

 

[Beastro]

A good / humanlike AI would do a lot of things, which many players dislike. That includes gratuitous backstabbing even when on friendly terms and other stuff. For example many people dislike if a friendly AI turns on them, even if it's the only logical move to prevent the player from completing a winning condition.
The AI in CIV 6 was criticized for being "unreadable" and "random" (e.g. attacking you, even if you have great relations) - but any competitive human player is unreadable and exploits any weakness. But many players don't like it, if their AI "friends" attack them out of nowhere, even if it's a logical move.

As bad as the AI was in games like CivIII and ETW (Have played either series since those iterations), I appreciated that I HAD to keep some armies along the borders of even friendly nations because it would be so advantageous for them to attack me while spread thin elsewhere.

It created a neat situation that tied a lot of my forces downs, but tempted me to denude them as I tried to find the sweet spot where I kept peace, but used as many units as I could in the wars I was fighting.

I liked that kind of thinking in AI when I have to read into their vague behavior. What I don't like is when a good AI is very clearly just a machine and reacts in abrupt, sudden ways to counter the player that even other players wouldn't do as they pursue their own strategies and goals.

I still think that most people prefer an AI which "feels" good (by acting like no human player would), not one which is purely competitive.

I don't get this, but that's because I've always looked on end state AI as trying to be as human as possible.

The failure of the gaming industry to give a damn about proper AI, since it doesn't draw as much money in as grpahics is one of the things I see as a major failure of the past 20 years. That failure is what I see behind the various ways developers have manipulated multiplayer in games as social interaction is cut out trying to make other human players in online into living NPCs to fight while preventing interaction that would cause verbal fighting and insults that results in people contacting customer support over frivolous things.

The thing with a competitive AI is it's speed and brutal efficiency. There are no misclicks, there's no time spent scrolling around, there's no being occupied with a battle and thus not being able to properly handle that other battle or the production chain.
AIs can do everything at once.

This is what I hate.

I don't want a hobbled AI because I don't want a challnge, I want it hobbled in a way that emulates people's actions and reactions so I'm not up against a machine that can tell what rock/paper/scissors I'm going to pick as I'm still in the motion doing so that reactis immediately to win beyond human capability.

To me the bizarre thing is representing the Goths, Saxons, Franks etc as literal cavemen who in 400AD still apparently had yet to learn how to eat berries or skin a cow. The Dark Ages weren't that dark Jon..

The beginning of every Civilization game has nothing in common with how civilization actually began, much less for how it began for every culture contained in each game. That's where the premise and gameplay diverge and one has to accept those bits in orer to get the ball rolling into the good stuff.

Was this a good system? It seems like the worst parts of doom stacks and 1UPT combined.

WTF are you thinking?

It was a neat way of doing things that helped make up for how terrible the AI was at expansion.

It also produced things organically, like how armies would form as a result with high defence units protecting artillery while cavalry rode spread out that helped with their natural movement advantage. It even opened up room for mixes, where you had high defence units protecting the "high attack, 1 defence" cav units leaving the "knight" branch of cav to be self supporting given their little bit of defence at the expense of attack.

The issue about all of that was that the AI didn't field units in a cohesive manner so you set upon one unit path and rode home with it, usually picking the high att cav and accept casualties, though the way circumstances could go you could stumble into situations that would have you using the infantry/artillery armies even if you didn't plan on using them originally.

 

[Hoggypare]

I'd also recommend Field of Glory II as possessing a AI that can give the player a run for their money (though I recommend moving to multiplayer ASAP as humans are a lot more fun and chaotic to fight than the competent but unimaginative AI).

This might be very much true, I played its predecessor pike and shot, and it has very respectable AI. However it is a wargame, very much math based - AI's handle such systems rather well

Dominions series AI is a mixed bag (understandably as some factions are a lot more gimmicky and complex to win with than others). But even on unaided difficulty settings they can be a real pain in the ass with factions like Ermor.

It still can't use vast majority of the game's systems. Game has thousand of spells and items, AI can't handle any competent thug building, buff cycles etc. "Don't. Develop. Systems. Your. AI. Cannot. Use" eh?

Ultimate General: Civil War is another one man band strategy game with an AI that does the job rather well most of the time and will give even a seasoned gamer some nasty fights (though I also inflicted some outright blood baths on it at certain battles I think I was stronger than the game had expected for in the campaign, I was starting to feel bad for the Yankees. But even the first mission is pretty tough for both Union or Confederacy).

Funny... I recently installed a mod for that, which makes a small tweak of limiting charge distance to 500 meters, so AI doesn't exhaust itself by charging over half a map. Top AI, truly. This is an example where I disagree totally - the whole challenge in this game comes from the fact that it covers a timeframe when people shot at each other in open field with increasingly advanced guns, so high casualties are bound to happen on both sides, and the player has limited reserves, while AI doesn't. Now I might agree that its predecessor Ultimate General Gettysburg had a great AI, but it was because it focused on one set scenario, so it was really polished.

Stretching a ways back but I remember Panzer General and some of its clones and modern spiritual successors (Unity of Command I think was the name?) having a mean AI. Though I always found that tended to lend the battles a puzzle-ish tinge. Still would tick the box.

Again, true, but it is a very much probability based wargame with a lot of abstraction in its systems. Order of battle or Panzer corps indeed have good AI, but they are entirely different games from any 4x

So we arrive at my original point. There are good AI's in case of
1. Abstracted systems in pure wargames (FoG II, P&S, PG, OoB, PC)
2. Limiting player's input in some way (like indirect battles of Dominions, or simulation of comms in more complex WW2 games like Graviteam series or even Combat Mission)
3. Preset scenarios that can be honed (UG:G and I guess that Hannibal game You mentioned)

The only 4x outlier here seems to be Pandora (which I admit, makes me interested). But there is no promised land of great AI 4x indie games anywhere - AI in those games is almost universally shit - as was the original consensus

Now, I will also use this opportunity to correct myself from earlier. My oppinion of At the Gates was without of knowledge of its late game. I simply didn't arrive there yet. Later the game has huge problems and is outright broken, just FYI. Ehh... next time Late Antiquity... next time...

 

[oscar]

Dominions series AI is a mixed bag (understandably as some factions are a lot more gimmicky and complex to win with than others). But even on unaided difficulty settings they can be a real pain in the ass with factions like Ermor.

It still can't use vast majority of the game's systems. Game has thousand of spells and items, AI can't handle any competent thug building, buff cycles etc. "Don't. Develop. Systems. Your. AI. Cannot. Use" eh?

Plenty of players cannot handle these either (Steam shows me 21 hours logged and I had barely scratched the most simple of those strategies). We are talking about arguably the most in-depth turn based grand strategy game of all time.

Multiplayer is clearly the crux of the game (with singleplayer not much more than a glorified tutorial) but that it has a semi-okay AI at all with the insane complexity of the game is telling. A single player only game with non-existent AI is a death penalty. You're paying for the intro video and the game manual because otherwise you might as well be bouncing a hackey sack in the air. There's good AI, okay AI, speedbump AI but to release a single player only strategy game with no AI is an insult to your customers. What is the point?

 

[Hoggypare]

Dominions series AI is a mixed bag (understandably as some factions are a lot more gimmicky and complex to win with than others). But even on unaided difficulty settings they can be a real pain in the ass with factions like Ermor.

It still can't use vast majority of the game's systems. Game has thousand of spells and items, AI can't handle any competent thug building, buff cycles etc. "Don't. Develop. Systems. Your. AI. Cannot. Use" eh?

Plenty of players cannot handle these either (Steam shows me 21 hours logged and I had barely scratched the most simple of those strategies). We are talking about arguably the most in-depth turn based grand strategy game of all time.

Multiplayer is clearly the crux of the game (with singleplayer not much more than a glorified tutorial) but that it has a semi-okay AI at all with the insane complexity of the game is telling. A single player only game with non-existent AI is a death penalty. You're paying for the intro video and the game manual because otherwise you might as well be bouncing a hackey sack in the air. There's good AI, okay AI, speedbump AI but to release a single player only strategy game with no AI is an insult to your customers. What is the point?

I would argue Dom AI isn't really good. It is just that player is almost equally limited by it. If both player and computer opponents are limited to a set of simple orders, even if AI does things randomly, You still can't exploit that too efficiently (to a point). Battles in TW games would also seem hard, if You couldn't micromanage Your cavalry to exploit flanking and causing mass routs. It wouldn't mean the AI is any less retarded. Now, don't get me wrong - it is all about the perceived difficulty and I agree limiting player's input is a good way of creating challenge. But it is a workaround nonetheless.

You claimed there are plenty of strategy games with good AI. The counterpoint is that the AI in 4x games is almost universally bad (bar one example of Pandora that I'd have to check). Workarounds or multiplayer focused games might indeed be good, but that is an entirely different point. I was genuinely interested if there were good strategy games with decent AI that I didn't know of, but as of yet, I am not convinced.
Now, At the Gates has a non-existent AI, and that is another problem...

Btw. If You want a nice summary of the capabilities of Dominions AI, look at the gods it generates.

 

[Axioms]

So I got At The Gates recently and what really annoyed me was how after I finally had a game where I had it all together and was getting ready to smash it, my game started crashing all the time. FML.

 

[Johannes]

Dominions AI is terrible. While it can match a total newb even on the level where it doesn't cheat, that's not saying much... I could tell a simple strategy for any nation for a newbie to execute, that'd beat the AI every time. It'd need only very simple instructions to improve it - sensible pretenders, don't have it waste endless troops against independents where it obviously can't win (or against player-built pd) , emphasise good troops over bad, build pre-determined item/spell script load outs for thugs (and possibly mages). These all would be pretty simple to write in pseudo-code, though implementing them in game might be much harder.