Game Development is Top-Down
Critical decisions to make between craft and innovation
As we stand on the brink of a new era in gaming, marked by an explosion of user-generated content, AI and virtual realities—it is the time-honored, top-down approach to game development that continues to be the backbone of innovation.
This article is about the future of games: it elaborates on what it means to do top-down development; how choices in the course of making a game impact velocity and the likelihood of success—and how emerging platforms like UEFN, technologies like AI and live services platforms show us the future landscape of game-making.
Metavert Meditations is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.
In top-down game development, you start with a concept, and then you iteratively apply the craft of game-making to realize your idea. This involves inventing playable components, creating artwork to convey a sense of the design and setting; and testing game mechanics that are playable-enough to glean the experience.
Along the way, you’ll bounce between the more conceptual layer (game design docs and vision, concept art, whiteboard sessions, prototypes) and the concrete (3D models, environments, code, playable systems). The more rapidly you can iterate between these layers, the more quickly you’ll reveal problems in the design and the faster you’ll get to a fun game.
When the emphasis is on the technology, you aren’t building a game. You’re building a computer-science project or a tech demo.
That’s not to say that technology isn’t integral to crafting amazing game experiences. It’s instructive to go all the way back to 1993 and look at the development of Doom; when John Carmack programmed it, he had to create whole new ways of rendering game graphics. Indeed, sometimes game concepts require technological innovation that hasn’t been attempted before.
Because the team that created Doom consisted of 3 core developers, the loop between artistic vision and technological innovation was extremely short—and Carmack had a very good handle on what was and wasn’t really possible for their game. Furthermore, Carmack had learned the discipline of shipping from his time at Softdisk: he shipped games on a regular cadence numbering in weeks—and if the tech was getting in the way, it was always going to take a back seat to getting a playable game out the door.
One can compare this to a more recent game like Valheim—made with a similarly-sized team—which used Unity for their graphics—allowing them to shift their focus to world building and perfecting their crafting and building systems. Their reward was a highly profitable game in an extremely competitive market.
Craft and Innovation
Every game has some mix of craft and innovation present.
By craft, I mean working within existing forms to synthesize the right mechanics, features, stories, art and progression you can. There’s usually a set of traditions to learn, game-makers with practical experience, and great examples to draw upon.
In the case of innovation, it is about moving the frontiers of what’s possible by investing in new technologies, new business models, new platforms and novel game mechanics.
Something that is 0%/100% craft/innovation is usually referred to as a “tech demo,” although the effort may produce some interesting discoveries that could be applied towards a real product later. To be clear, there’s nothing at all wrong with making tech demos—and often a great deal that can be discovered with them—as long as you’re clear on what you’re making.
A number of games have achieved incredible success by focusing on one or two key technological innovations. For example, Angry Birds really learned how to make use of a touch screen in a novel way; and Beat Saber figured out how to make use of the unique experience of VR. Others have innovated by combining features together in a unique way (such as the shrinking-battlefield, loot system and Battle Royale format inside PUBG).
Other games have innovated more applying existing technologies towards unique game mechanics, such as the message-leaving in the Dark Souls series, or the unique multiplayer design in Journey.
Doom, as mentioned above, utilized a number of technological and design features that hadn’t previously been used in a video game.1
Something that is much higher on the craft side of the craft-innovation continuum will feel very familiar to the gaming audience, and can potentially deliver a finely-tuned, perfected experience. Most commercially-successful games fit this category, including many of the best you’ve played.
For example, Baldur’s Gate 3 is perhaps one of the finest RPGs ever made. Their storytelling, artistry, acting, world building, level design, and D&D implementation are at the pinnacle of the craft.
Larian chose to build on their own game engine (Divinity 4.0) for BG3, perhaps because their studio has a long legacy of creating their own engine for bird’s eye view RPGs; their engine is good at dealing with large maps while allowing for a high level of detail. I imagine they did the math and found that sticking by the Divinity engine would serve the narrative and level-design goals of the game than attempting to fit these features into Unity or Unreal. This technology investment was driven by the requirements laid out by the game’s vision.
Could you afford to build a new engine for your game? Unless, like Larian, you’ve got extensive background in it and highly specialized requirements—and can justify the cost relative to all the other investments you might make—it is likely to be a high-risk option. Even companies like CD Projekt Red, famous for Witcher and Cyberpunk 2077—recently opted to move to Unreal for future games, given the high cost of technical debt and the high investment needed to stay on the leading edge of 3D graphics technology. This trend is continuing across the entire industry: not only for 3D graphics, but for all aspects of the game development tech stack.
Games are Now Live Entertainment
At about the same time Carmack was working on Doom, I was working on one of the first games on the Internet: Legends of Future Past. We also had an ethos of shipping: every week, we’d post a new update and there was a live story to participate in. I carried this approach forward into the products like Game of Thrones Ascent and Star Trek Timelines, where building a robust authoring pipeline that could handle regular updates was essential.
It turned out we were ahead of our time. Games aren’t just products you buy, play and dispose of—they become part of your life. People want to be part of living worlds. Today, most of the large games you see are have some aspect of this.
Shipping is the key.
The Creator Economy for Games
“Professional” game developers use 3D engines—typically Unity or Unreal—and build a wide variety of games. These platforms allow for a high degree of emergent creativity: if you can imagine it, you can build it—given enough content-creation and technical resources. This is the ecosystem that generated over 14,000 games on Steam in 2023 (and many more on mobile).
But these platforms do not capture the totality of game development. There are 3 other platforms you’ve absolutely got to pay attention to: Minecraft, Roblox and Unreal Engine for Fortnite (UEFN). I previously wrote about the growth of these creator-driven worlds. Here’s how many people are on these platforms, and the percentage of total gamer marketshare they occupy (given 3.3B gamers in the worlds):
Minecraft has two core experiences: survival-mode — plus the pure joy building environments and worlds. Because the game is moddable (albeit not exactly easy to do, requiring skills with Java programming) it has spawned a vast ecosystem of multiplayer servers with unique games and experiences. People make servers and worlds in Minecraft because doing so is a top-down experience; the loop between an idea and the implementation is relatively short, typically something within the grasp of an individual person with the willingness to learn a bit about Java.
Fortnite is a game that’s fun on its own, providing a core experience that people enjoy endlessly. Without that core experience, I suspect UEFN (Unreal Engine for Fortnite) would be a tough sell: today, it’s a very slimmed-down version of Unreal, without even the ability to integrate into third-party SDKs and server tech—yet the creator mode is attracting developers because there’s a core experience (along with a huge number of players) to interconnect with. Again, there’s a tight loop between an idea and the ability to execute.
In the case of Roblox, there’s no true “core experience” in Roblox—a path that took them around 15 years to bring to fruition, proving how hard gaming platforms can be when they lack a core experience. It’s a development platform that combines product distribution along with a social graph of players; it’s a top-down platform that focuses on enabling creativity, while they deliver the distribution, toolset, multiplayer features and a social graph of players.
A Tale of Two Technologies
Two different technologies demonstrate what happens when the bridge is too far between idea and execution: let’s compare generative AI to blockchain.
Generative AI has rapidly found use cases in everything from conversation AI like ChatGPT, answer-driven search engines (like Perplexity) to art platforms like Midjourney. The reason for this is that all of these are top-down driven products. Yes, the computer science behind LLMs and diffusion models took years and years and lots of effort—but it was productization that brought it to mass-market acceptance.
In game development, people use ChatGPT to iterate ideas, platforms like Midjourney to develop concepts, and are even starting to create in-game assets using products like Scenario. The common theme here is that all these technologies accelerate the process of ideation and execution—rather than introducing new processes that slow you down. Furthermore, they tend to latch-on to parts of the existing workflow, rather than build too many interdependent parts at once.2
Compare this to blockchain, which is a lower-level technology. There are interesting use cases for blockchain, but like the 15-year journey of Roblox3, they’re likely to take a while before they deliver on the promise. One of the challenges with blockchain games is that it often drives teams to pursue development that is not top-down. Teams frequently get caught up working at the level of smart contracts, RPC servers and APIs—rather than getting playable prototypes onto the screen. Imagine if you had to design a datastore methodology and security framework for each game you wanted to build in Roblox before you could get started… Unfortunately, blockchain has been a bridge-too-far for many game teams, leading to millions upon millions of squandered capital.
Some have tried to solve these with APIs and SDKs that add abstraction layers over blockchain, making some of the interfaces seemingly more familiar to people who are building virtual worlds. And while this isn’t a terrible idea, it misses most of the point: game development is a top-down process where iteration speed is paramount.
This is why online games are simultaneously hard and easy to make. They’re hard—extraordinarily hard—if you want to build up technology from the bottom. But if you build inside a platform where everything is immediately accessible to you (like Roblox) it is within reach of nearly anyone. This simplicity has driven Roblox to hundreds of millions of users per month, with some individual developers making millions.
The core concept of decentralization is that no single entity should maintain control over a technology—which would otherwise establish gatekeepers, rule-makers and tax-takers—each of which is a stranglehold on innovation. Decentralization is the premise of the internet itself, the domain name system, blockchain, open-source software, and open-source AI models.
This is essentially what we’re trying to do at Beamable: by utilizing a top-down, workflow-centered approach to multiplayer game development, our vision is to bring the simplicity of Roblox game development into the world of Unity and Unreal. This means creating an integrated set of tools which work from the editor (IDE) environments, interoperability between components, simplified scripting to control server-authoritative behavior, and an extensible marketplace of add-ons. Developers truly own their products and decide how to distribute and monetize them—while tapping into an interoperable ecosystem that draws from the best capabilities in the market.
Key Questions In Development
When making any decision to use a technology, my experience is that you need to consider several important questions beyond the initial feasibility of the technology; this is relevant whether you are going to take on a technology project yourself, or whether you’re going to use off-the-shelf technology (like a 3D engine, or a live services platform):
Is the technology satisfying a core requirement of the product’s vision? If not, it is probably a distraction—at least until more of the game concept is proven-out.
If you do adopt the technology, how will the business model impact your present and future risks? Business models with costly up-front capital will deplete your runway; excessive revenue-shares will impair your margin (and consequently the capital you can reinvest ingrowth); and centralized platforms might introduce unacceptable risks around security, data ownership and business continuity.
Even if the technology works, how does it impact the velocity of your team? The cost of incorporating a technology is not only its initial time and money: it is the acquired tech debt, support costs, workflow and deployment impact.
Will it scale? Consider what happens when you have millions of players, high concurrency, lots of data, lots of inventory, lots of interactions, maps sizes are growing, etc. (the variables of scalability will change according to the type of game you’re creating)
In my experience, many teams are fairly sensitive to the first two questions—but either overlook or gloss-over the risks associated with the second two, leading to mid-to-late crises as velocity degrades and the end of your runway looms.
By maintaining a top-down approach to game development that optimizes for speed during all phases of the development, you’ll dramatically increase your chances of success.
The key to unlocking future success lies in balancing technological choices with the mastery of craft. By embracing a top-down approach that values rapid iteration and creative vision, developers are not just creating games—they're shaping immersive experiences that define the evolving landscape of gaming. As we venture forward, the challenge for game developers remains: how to not only succeed but to innovate in ways that leave an enduring mark on the world of gaming.
Creator-driven worlds explores the phenomena of Roblox, Minecraft and UEFN.
Evolution of the Creator Economy discusses how creative tools tend to evolve over time.
The Direct from Imagination Era Has Begun dreams of a day when we can speak whole worlds into existence.
The fine details of Doom’s innovation are a bit beyond the scope of this, but some of them included using binary space positioning (a method for avoiding the computation of the occluded parts of a 3D scene), adapted from earlier non-game applications in computer graphics; highly optimized texture-mapping techniques; and ambient lighting. And it also introduced the game-design innovation of the Deathmatch mode.
This is part of why 2D art is already finding a foothold in game development. 3D is not only harder, but it has many other parts. I posted an article on building the holodeck in 2023, and the parts that were hard then are still hard, while the stuff like 2D and audio that fill much more discrete niches have improved substantially. The 3D pipeline will get there eventually but there’s enormous complexity involved, in addition to fitting it to established workflows.
Roblox shows what the potential for blockchain gaming might be: interoperable assets, particularly the avatar-customizing ones, that span multiple games made by tiny teams. This already happens in the non-chain universe of Roblox. This seems like the low-hanging fruit in the blockchain gaming market, but to take off we’ll need to see a handful of hit games before it proves the use case. I wrote about this in Game Economics Part 2, NFTs and Digital Collectibles, back in 2021.