Frogboy

Join Date 03/2001

+2421

Ashes II Dev Journal: A Preview of Internal Alpha 1

External Link

We’ve been hard at work on our internal Alpha 1 milestone. The focus has been finishing the new faction: the United Earth Forces (UEF). This is a unified military command made up of the United States, China, Russia, the EU, Australia, Japan, Korea, and others, all placed under your control to deal with the so-called “Post-Human Coalition.”

Our story so far

Those familiar with Ashes of the Singularity (2015) may recall the then far-fetched idea that, in the not-so-distant future, a handful of tech-industry leaders would develop artificial general intelligence (AGI) and use it to uplift themselves into a virtual world called the Metaverse. From there, they would interact with the real world through their constructs: drones, robots, and more. Together, they would form the Post-Human Coalition (PHC).

The first game centered on the conflict between the PHC and the Substrate, a sentient AGI called Haalee and her forces. The battles took place away from Earth. In fact, Earth and humanity were barely mentioned beyond the campaign’s comment that there were only around 40 Post-Humans and 10 billion normal humans left.

If the plot of the first game felt esoteric, that was deliberate. It was all setup for what would become Ashes of the Singularity II, which we originally intended to make much sooner.

Ashes of the Singularity II is where things become much clearer and, given the rise of real-world AI, a lot more plausible. The story opens with the PHC converting the surface of the Moon into Turinium, a material that combines energy production and data-center capability. Think of a nanotech substance that handles both computing and power generation.

What’s new in Ashes of the Singularity II?

One thing I like about RTS games is that they don’t “age” the same way other genres do. The challenge for developers is making the case that players should try something new when they can still boot up StarCraft II and it looks great, or Company of Heroes 1, or modern free projects like BAR (Beyond All Reason). Sanctuary: Shattered Sun is also looking impressive. And that’s without touching on Age of Empires, Age of Mythology, and others.

You get the idea though: A new RTS game has to have compelling features to justify my (your) time. So let’s start with what have we done here:

Feature	Ashes of the Singularity II	Ashes of the Singularity	Comments
Factions	3	2	NEW: The humans! Earth!
Scale	Tiny to Massive	Tiny to Massive	The scale is about the same in terms of size. What has changed is that the armies in Ashes II were designed with this understanding. Ashes I had near StarCraft levels of micro at times but Supreme Commander levels of scale. This time, every unit always belongs to an Army. Reinforcing and even choosing what is an army is done at the Army UI instead of finding buildings to build them. Armies will always send orders to your factories and the units will initially deploy from the nearest factory to the army.
Unit Production	Army Based	Building Based	In the first game, players spent a lot of time hunting and pecking through many buildings to find the right factory to give orders to. Now, construction is global. Your army needs a new tank or mech, you put in the order. This lets players focus on their strategy and not how quickly they can find the right factory.
Map Choices	Procedurally Generated	Pre-Made	In the first game, players picked a map from a list. In the sequel, while we will have pre-made maps, the default is to simply choose some map settings and let the map generate it. This means that EXPLORATION is a big part of the game because no two maps are the same. From playing the Alpha I can say this is a huge deal.
Tech Progression	Tech Tree	Building Unlocks	This is a classic “with the benefit of hindsight” design improvement. Having a Tech tree allows us to have units unlocked through interesting choices rather than forcing the player to build a specific building to unlock a unit. The building unlocks a unit design isn’t inherently bad until you remember the scale. A player might not realize that some key building was destroyed because of the scale of the game.
Map Organization	Regions w/ power lines	Regions	Because units being produced are initially deployed in the nearest factory to the army that makes that nearest factory a lot more important than before. If I have a factory near the front lines and behind the front lines there are 10 factories, then that front line factory will be deploying units pretty fast. And if that was the end of it, that would make for a very OP mechanic (imo). However, the map is broken up into regions and if you flank the enemy and cut off a region, then power to that factory is gone and thus that factory is no longer supplying units. This makes the risk/reward of getting in front of your supply lines that much more enticing.
Construction	Region based Engineers	Individual Engineers	In the first game, you would produce construction units (Engineers) and I would queue up orders for them. But this design didn’t scale. Late game you might have dozens of them and you’d be spending your time looking for them. It was often easier to spend precious resources to simply call down a brand new Engineer than to find one. Now, Engineers are automatically provided and tied to their region. You give orders to the region and the Engineers do the work. This is similar to what we do in Sins of a Solar Empire.
Rally Points	Automatic	Manual	The first game was pretty traditional – you select a building and have its units go out to a particular rally point. But in Ashes of the Singularity II, no need because they will automatically join the army that ordered them in the first place.
Ballistics	Physics Based	Abstract	In 2015, the sheer quantity of units made it a challenge to deal with weapons hitting and missing. A given unit might have nearly a dozen different turrets on it. And with no unit cap, we were mindful of the late game performance. But in 2025, where everyone has at least 8 logical cores to work with, you could have a core dedicated just for ballistics (we don’t do this but you get the idea). Thus, weapons fire is no longer abstracted but is instead based on physics. This is something I loved about Total Annihilation (1997) which had this (albeit they didn’t go nuts with the turrets).
Wrecks	Persistent Wrecks	None	While we haven’t decided whether wrecks can be scavenged for resources yet, when units die, they persist on the map, creating a hazard and marking the scene of a great battle.

Now, this is, by no means, remotely complete. If there are specific questions, please comment below. But in the meantime, here are some screenshots that I think will help visualize the differences.

Screenshots

Australia

ss01_AotS2_screenshot_OB-01_20251113a-20251120-173432.png

This battle takes place in the Outback. The map is procedurally generated, with natural passes and pathways emerging from the terrain.

You can see tanks, mechs and soldiers. The units on both sides are UEF, but the blue side is actually mercenaries hired by the PHC. The Leopard 4 tank supplied by Germany is state-of-the-art early in the conflict (in the near future; today the Leopard 2A4 is, I believe, the latest). If you look closely, you can see the UEF beginning to deploy mechs.

Here’s the same battlefield from a recon drone:

ss05_AotS2_screenshot_OB-Wide_20251113a-20251120-173950.png

Russia

This screenshot is from Siberia. In the 2030s, the Russian government leased millions of acres to conglomerates that would later join the Post-Human Coalition, expecting AI tech in return. (They didn’t get it.) The UEF is hunting a PHC Nexus. The cold is a real challenge for humans, less so for the PHC drones.

Zoomed out, you see this:

ss04_AotS2_screenshot_Strat-01_20251113a-20251120-173950.png

Colorado

While it is not yet decided whether Colorado will make it into the campaign or be referred to, green forested mountainous terrain will certainly be an option.

As you zoom out, the units will gradually convert to icons for better readability (before converting to full strategic abstraction). But here you can see a battle in progress.

So what makes Ashes of the Singularity Special?

Replayability matters. Being able to jump into a game with friends matters. Facing off against the AI and having that experience stay interesting matters. Having strategic decisions matter more than click speed matters. Holding critical territory matters.

And honestly, what I want from an RTS has changed. While PvP matchmaking is important, it’ll be there, I find myself preferring 2-vs-AI matches with a friend. I’m not sure why, but those games are simply more engaging. I think many players feel the same. That only works if maps are fresh every time and if the AI is interesting. I’d really like your opinions on this. I used to play Total Annihilation in PGL and now I find myself playing games 2 vs AI. I used to be cool! (actually, I was never cool).

So much of our engineering effort has gone into making replayability the defining feature: intelligent computer opponents, procedural maps, and a design centered on taking and holding territory. Protecting supply lines. Making every match feel like its own story. From those decisions, everything else follows.

We’ll be getting into the details about the United Earth Forces units soon. In the meantime, we’d love to hear from you.

ashes dev journals

Pinned Post +1 Karma

Advanced Search

Watch this post

Do not email me updates for this post Email me updates for this post

Successfully updated karma reason!

Wingflier

Really good write-up. I agreed with most of your points.

As someone who started playing Total Annihilation in 1997 when I was 10 years old, graduated to TA Spring later, then Forged Alliance, Supreme Commander 2, Ashes of the Singularity, Planetary Annihilation, and most recently Beyond All Reason - I can say that you seem to understand some of the aspects that make this particular genre so great.

In particular, I'm glad that Ashes 2 is going even further in the direction of macro-oriented low-micro playstyle - this is the direction that RTS games NEED to be moving in as a whole already. The latest two examples or RTS games that attempted to replicate the micro-heavy Starcraft style of gameplay, Battle Aces and Stormgate, crashed and burned in spectacular fashion despite being multi million dollar projects.

For me that's a message from the RTS community, loud and clear, we want to move in the direction of epic-scale, macro-oriented battles. The heavy micro small scale RTS games had their heyday, and that era is now past. So the recent trend of RTS games like Warno, Broken Arrow, Sins of a Solar Empire 2, and now Ashes 2 means we're finally moving in the right direction.

However, I do want to both emphasize the final point you made and also urge a word of warning:

You are correct in saying that the 2v2 style Player vs AI battles are ABSOLUTELY the most enjoyable and replayable parts about these games that keep people coming back forever. However, ALL of that is contingent on having a good AI.

Just as a prescient example, Sins of a Solar Empire 2 is probably the best RTS made in at least a decade in EVERY CONCEIVABLE WAY that is held back by one profound design flaw - The AI is absolutely atrocious. Even with massive resource cheats and huge advantages over the player, it is so incompetent and irrational in its decisions, it cannot be relied upon to make interesting or close games.

This is a shame because it's like making the highest quality sports car known to man, then forgetting to include the driving wheel, the gear shifter, or some other intractably important component that the game can't survive without. Games like Sins 2 and Ashes 2 will NEVER have a big and consistent enough community to always sustain the large PVP style games that people love, and even if they did, most players just don't enjoy PVP.

There are 3 major ways to handle this:

1. In 2025, the easiest and most sensible solution is to find a way to make machine learning work for you. With all the modern AI options available to developers, there's simply no reason you could not invest in a machine learning solution that would create - without any doubt - the best and most advanced AI ever created for an RTS to date. The upfront cost of developing this system might initially be problematic, but once the system was in place, the algorithm basically teaches itself.

This is the best option not only because it produces the best possible results, but because of the NOVELTY of the approach. You can advertise Ashes 2 as being the first major RTS ever made to use machine learning to create an unbelievably powerful AI that does not need to cheat or be given unfair advantages over the players just to keep up. This creates more hype and excitement about the game as not just another RTS sequel or TA spinoff, but something profoundly unique and interesting in its own right.

2. Make the AI moddable to players, or alternatively, allow players to create an AI from scratch for the game.

This is basically crowd-sourcing the AI to your own community, but nonetheless, the community will be willing to create a better AI in most cases than your team can, and for free.

Again, a big disappointment I have with Sins 2 that I hope you can learn from is that they have one of the most modder-friendly and easily accessible modding architectures of any RTS ever made - except for the one area that actually matters the most - allowing the community to tweak and develop the atrocious AI.

3. The third option is the traditional method of just hiring some very experienced developer(s) to create an AI for you, but I find this to be the weakest and least attractive option of the 3. It doesn't have any novelty, it's what every other RTS is already doing and has always done, and in the end, no matter how talented the AI devs are, once the players understand how to beat the scripted design, it becomes boring and the outcome is pre-ordained.

The value of the first 2 options are NOT ONLY that they can create better AIs and that they will continue to produce better AIs until the end of time, either by machine or community - it's the multiplicity of different AI engines that you are capable of creating by either outsourcing this to machine learning or modders.

So my final thoughts are, learn from Sins 2's tragic mistake. If you're going to make a game designed around PVE, then God damnit, make a game designed around PVE. Do something no dev has ever done before and structure this design axiom into your game from the ground up. If you launch the game with a mediocre and predictable, scripts-based AI, while calling it an epic scale PVE game, don't expect people to review it well on Steam. Expectations are high, and you have to have something that sets you apart from the pack.

Successfully updated karma reason!