Ashes Of The Singularity Interview: “It’s Total Annihilation Meets Company of Heroes Meets Kohan”

Ashes of the Singularity [official site] can let you build, control, and display thousands of units in the course of a single battle, but that enormous scale is in some ways an attempt to make manifest the technical progress that Oxide and Stardock’s 64-bit, DirectX 12 RTS represents. There are a lot of things happening under the hood that Stardock’s Brad Wardell is happy to tell you about, things that will make Ashes a unique achievement compared to all of its predecessors, but they are things that programmers and developers would appreciate the most. For the rest of us, there are these vast armies clashing across miles and miles of terrain, a graphical feat that shows us progress we can appreciate.

For all that Ashes is an attempt to usher RTS games into the future of programming and gaming hardware, however, its design is rooted in some of the most important and promising moments in the genre’s past.

The first thing to understand about Ashes of the Singularity is that it is not, despite all appearances, a remake of Supreme Commander. It may be about building shockingly large armies of armored units, and managing an income stream from metal harvesters, but that’s about where the similarities end and a far less conventional, more innovative RTS begins.

“I like to think it’s Total Annihilation meets Company of Heroes meets Kohan,” says Brad Wardell, the head of Stardock, who are publishing the game. Wardell is acting as a producer and evangelist for the game, but his first challenge is explaining what makes Ashes special. “But unfortunately, because so few people know [Kohan], that reference doesn’t work for most people.”

[You can watch Stardock CEO Brad Wardell play the game and talk to Rob about it in the below video –Ed]

Yet to RTS fans who were fortunate enough to play Kohan, Wardell’s reference is music to the ear. Kohan remains a fascinating RTS experiment, an evolutionary branch of the genre that was tragically ignored in favor of more lavishly-produced, conventional fare. One of its key innovations, and one that Ashes borrows, was to make armies, not individual units or control groups, the unit scale that players controlled.

So while Ashes might look like a game about commanding absolutely bonkers numbers of units, appearances are deceiving. While you always have the option to micro-manage small clusters of units yourself, most of the game is spent commanding a handful of armies comprised of dozens or hundreds of units.

It’s an important difference, Wardell insists, because a control group is a different beast from an army, in terms of AI. “See, the problem with a control group is that you still end up with the case where someone at the edge of your control group is getting killed and the rest of your guys are just standing there.

“But when it’s an actual meta-unit, if you’re nailing someone on the end of the line, someone else is coming over. If someone needs some health, he’s going to get healed because there’s a healing unit as part of the overall group. As opposed to it saying, ‘Sorry, you’re five pixels outside of my zone, so you’re just going to die.'”

Because an army kind of behaves a single collective entity, it will fight differently depending on its composition in order to maximize its efficiency. To help them do that, they receive subtle buffs based on their composition to help them play a specialized role if necessary.

“So if I have a bunch of scouts in there, my overall army will move a little bit faster. If I have a bunch of Artemises [an artillery unit] in there, my firing range will be a bit further,” Wardell explains. The system might still be a little too subtle for its own good, though. “We haven’t played around with it too much because what we have found, unfortunately, is that most players don’t even notice it. You have four or five battles going on at once, so it doesn’t end up being communicated very well to the player.”

Since you’re not as worried about micro-managing armies, Ashes frees players to focus on the big-picture strategic dynamic, which is heavily influenced by Company of Heroes’ point-control system. You must build resource extractors on patches across the map, where they generate a steady stream of income, but you can’t do that until you’ve captured the node at the center of the territory, and can trace a route from that node back to your base.

So while controlling armies and supplying them with fresh units is obviously a key responsibility, their role is chiefly to protect friendly supply lines and jeopardize the enemy’s. And on Ashes’ sprawling maps, which can accommodate thousand and thousands of units, there are a lot of ways to cut someone’s economy to shreds without beating them head-on. More than once, I’ve turned a game by end-running a fast scout army through lightly defended enemy territory deep behind enemy lines. Even if you’ve only captured a quarter of enemy-held territory, if you’ve cut the important links between the front and their home base, you’ve cut them down to just a quarter of their income. It’s an element that makes for huge mid-game swings.

These are neat touches, but do they make enough of a difference? While Ashes boasts impressive armies and user-friendly design, it’s still working in the vein of traditional RTS games, and that genre has become a harder-sell in recent years. They’ve been eclipsed by their progeny, the MOBAs, and most of the major RTS franchises are gone. After StarCraft 2: Legacy of the Void, Blizzard’s own RTS future is hard to predict.

Wardell isn’t blind to the challenges. He knows that RTS games have been taken down a few pegs since their zenith in the late 90s.

“There have been three generations of RTS games. The ‘Gen 2’ RTS games: Age of Empires, StarCraft, Total Annihilation… those games had gigantic budgets. It’s tough to do that today. Especially now. People don’t want to take the risks that they used to. And there’s also the issue of new IP. New IP is getting really scary. I can create a game that’s new IP, put my heart and soul into it, get great reviews and it can bomb. Because 40 other games were released on Steam that day.”

But Wardell also thinks that RTS games fell into a pattern of giving players more of what they already had.

“I don’t think people are turned off by the genre. The difference is that, unlike other genres of games, they haven’t been able to innovate. So let’s look at all the recent RTS games you just mentioned. They’re all 32-bit, DirectX 9 games. And DX9 came out years ago. And that translates to how the games look, and how complex they can be.”

Wardell knows those limitations well, because Stardock and developer Ironclad encountered them with the Sins of a Solar Empire series. By the end of Sins, they had maxed-out what the Sins engine was capable of handling with its 32-bit engine. Not just in terms of graphics, Wardell is quick to stress, but in terms of how sophisticated it could be. That has imposed major trade-offs on every developer working on RTS games. By operating with a 64-bit engine and requiring a minimum of four CPU cores, Ashes transcends those limitations.

“In the near-term that’s going to bite us in the butt, because a lot of people are still running dual cores,” Wardell admits. “But it lets us have an AI that people haven’t seen before. People don’t even know what a good RTS AI really is. Not because the developers aren’t good, because there are a lot of great AI programmers out there, but there’s only so much you can do if you have to calculate your AI between frames.”

As Wardell explains it, most games have to operate under the assumption that they will be running on a single core. Each frame has to be calculated from end-to-end on a single CPU core. A game runs from CPU bottleneck to CPU bottleneck.

Ashes doesn’t work that way because Oxide have been able to design for a quad-core minimum-spec. Those bottlenecks go away.

“It’s not just doing twice as much. You’re talking about orders of magnitude more CPU power than could be thrown at the AI previously. Earlier AI is either running a script or reacting as fast as it can. But when you play this game, you’ll see the AI using combined arms and using groups, not just sending a death-train at you.”

I haven’t played enough of Ashes to validate these claims, especially since Ashes is far from being optimized and throttles my PC even during smaller battles. But I’m not sure 64-bit, multi-core design is the panacea that Wardell makes it out to be. This makes Ashes a throwback to the golden age in another way: it is made with a conviction that better technology and more system resources can produce something revolutionary rather than an incremental improvement on what came before. When Wardell is telling me everything that Oxide’s new engine means for Ashes, I start to share his enthusiasm.

But when I actually play the game, I wonder if it goes far enough. I love the influences that have gone into Ashes of the Singularity, and the ways it streamlines player-control rather than frustrates it. These are small, but meaningful improvements. They are twists on a classic recipe, but, at this early stage, they are not quite the leap forward that Wardell describes unfolding in the code.

Ashes of the Singularity is out now in early access.

24 Comments

  1. Premium User Badge

    AceJohnny says:

    Oh darling, you had me at “Kohan” <3

    (see the retrospective from 4 days ago here)

    • EhexT says:

      It’s a shame that so many Kohan inspired games miss a core point of Kohans combat system – you can’t fight forever without resupply and getting out of combat is HARD and has a cost. Just having groups of units and calling it Kohan isn’t really getting it.

  2. EhexT says:

    What’s really impressive is how they managed to make a game that looks worse than SupCom 1 (seriously, compare the header image to an image of SupCom 1 on max settings at equal zoom distance) run so much worse, despite all the fancy dx12 and improved ground-up multi-threading.

    And how they’re still so stuck in the old RTS paradigms when they clearly are aiming for something different – such as having to place individual extractors, or even having to manually place them at all (you’re hardlimited by extractor locations, the capture of the central node and the presence of an engineer anyway – why isn’t the engineer doing it automatically if all the conditions are met?).

  3. onionman says:

    I have to say, I basically didn’t care about this game at all until reading this article. Now I’m psyched.

    • Hedgeclipper says:

      Yeah but its Stardock, they always talk a big game with all sorts of interesting sounding ideas and then they release and its all a bit underwhelming.

      • pepperfez says:

        C’mon, when has a Stardock strategy game ever failed to live up to expectations?

        • Gnoupi says:

          “Elemental: War of Magic” rings any bells?

          • gnodab says:

            It still ended up great in the end. Fallen Enchantress and especially Legendary Heroes are pretty amazing (just really, really ugly).

          • Shadow says:

            Elemental ended up passable at best. Fallen Enchantress and Legendary Heroes were better, but to call any of them “great” is quite a stretch.

  4. BlackeyeVuk says:

    Ooooh, when I yell’d like a raging buffoon about ye ol’ Kohan nobody give cookie about it. Now its all the bandwagen you can get.

    Well , to all new fancy composed kohan fans….

    Welcome. May you thrive.

  5. Risingson says:

    The Ancient Art Of Opportunistic Namedropping.

  6. ScubaMonster says:

    Never utter the words “Kohan” unless it’s followed by a “3”.

  7. Ur-Quan says:

    Well I played the early access version and… sorry but a lot of this sounds like pure marketing bullshit talk. I couldn’t see any difference between armies and traditional controlgroups at all, the game runs really terrible on my quadcore processor and it’s full of rather archaic RTS leftovers.

    I really hope they get their stuff together, but to be honest at the moment I think this will be a very average bland RTS at best.

    • jeeger says:

      Yeah, I watched the recent gameplay trailer, and it just looked so boring. Maybe it’s because of the alpha status, and RTSs have a tendency to look boring unless you’re quite familiar with the game, but really, there wasn’t anything where I’d say “This looks interesting”.

      They chose a very small map, there’s just about 6 kinds of units, the effects of grouping units are invisible, and there’s not enough automation to take care of the “boring” stuff like building mexes or shuffling units.

      Saying “Current technology will help us make a better game” is all fine and dandy, but so far, I haven’t gotten anything like that out of the game.

    • Unclepauly says:

      You wouldn’t happen to be running an AMD quad core would you?… o.o

  8. briangw says:

    Did you read the rather beefy system requirements before trying it? In a previous post by them, they mentioned you needed a pretty hefty system now until they can optimize it.

  9. LintMan says:

    The Total Annihaltion/Kohan mix sounds awesome. but the “fast scout army disrupting your supply lines” worries me.

    I’m envisioning the AI constantly harassing your supply lines on those huge maps, forcing you to spend most of your time/focus on protecting them instead of focusing on your strategic goals.

    • K33L3R says:

      Kinda realistic then if the focus is on disrupting supply lines, supply lines make or break armies in the real world
      Understand how that might not translate well into a game/take the fun out
      Can you set up patrols in this game? That would help solve the needed micromanagement of protecting supply lines properly

  10. AllixX says:

    Looking at how they made the maps with capture nodes it worries me. The map seemed to be full of choke points and no ways to really flank an opponent.. There is air and maybe you will have transports but it seemed a bit dull. Maybe de orbitals will spice the battle out hopefully..

    Also the graphics seem to lack a bit. I am hopeful for this game to succeed but what I see now will only make me play Company of heroes 2 more..

    • Unclepauly says:

      In the vid he clearly put all the settings on low so the game would run smoother.

  11. Premium User Badge

    BlueTemplar says:

    I have a hard time to believe that

    “People don’t even know what a good RTS AI really is. Not because the developers aren’t good, because there are a lot of great AI programmers out there, but there’s only so much you can do if you have to calculate your AI between frames.”

    because from what I’ve learned, it’s not so much that there’s not enough processing power or talented AI programmers out there, but that spending a lot on AI doesn’t make sense when most players won’t care (or hate that the AI is beating them), and those that do care will still prefer playing against a real player.

    So it’s much easier for a developer of a strategy game (real-time or turn-based) to make an “average” AI that only a newbie will have a hard time against, and to make harder versions of the same AI by giving it economic bonuses (or flat out make it cheat by for instance being able to see the map).

    To support this claim I present the “Berkely Overmind” article :
    link to arstechnica.com
    They were able to make an AI able to compete with a professional level player, but the incentive was to win an AI competition (not to fool your average player), and a whole class of computer science students focused for almost a year on that goal.

    Now, this also doesn’t really apply to Stardock, as they’re well known for making GalCiv2 around its AI (and with results : the difficulty level where the AI is at a level playing field compared to the player is significantly higher in GalCiv2 than in other empire games), but maybe with the caveat that they don’t spend as much effort on other aspects of the game, so GalCiv2 is described by a fraction of the playerbase as “boring”…

    So that Wardell’s statement might well apply to Ashes of Singularity in the sense that they’ve again spent a lot of efforts on making a better AI… and where to do that better than on a “tech demo” game like this one?

    • Unclepauly says:

      I would much rather play an A.I. that uses different tactics than an A.I. that just has a beefed up economy or even cheats.

    • rmsgrey says:

      The real selling point for better AI isn’t an enemy that uses smarter tactics to beat you rather than having unfair advantages; it’s having assistance on your side so you don’t need to micromanage everything – for example, having nearby units co-ordinate and focus-fire on the same enemy without needing constant orders to do so.

      I’ve long felt that what grand strategy games really need is a good supply of sergeants – people who take the strategic orders given by officers and break them down into sensible orders to individual soldiers, rather than almost all the intelligence being concentrated at the top, with a little at the individual soldier’s level (who might have enough sense to react to being fired upon)