I have fond memories of playing flight simulators with my dad when I was pre-teen, taking turns in trying landings at Seatle in Microsoft Flight Simulator 3 and cheering our successes and failures. Or even earlier, in the OG Microsoft flight Simulator that had a minigame where you flew a Sopwith Camel and shot at enemy planes. The planes were just stick crosses!

My dad would feed me tidbits of aeronautics each session meanwhile I listened intensely. That’s what I say to myself, but I’m sure my dad had infinite patience meanwhile I just wanted speed and dares while flying between skyscrapers.

He has always been the mad inventor type, so he built rudder pedals out of big paper hole punchers and photoelectric sensors, and collective levers, and throttles with potentiometers. That’s when we got into helicopters, with Jane’s AH-64D Longbow, and later moved to learn our dogfight chops with Microsoft Combat Flight Sims 1 to 3, and also IL-2. To me, a teen at that moment, the adrenaline and speed of dogfighting trumped over any other kind of sim flying.

Then life got more interesting, and quite the chunk of time passed by. Until Covid, when I found myself flying online on IL-2 1946 and IL-2 BoS on a docked Steamdeck, this time with solid rudder pedals and joysticks of my own. I discovered then the joy of flying in groups of hundreds of players, and I’ve been hooked again since!

The skill ceiling when flying multiplayer is through the roof, as you get to pick when to run away, or engage and bet for the win. There’s real need for teamwork in multiplayer, and little left to luck. You debrief with your teammates the .acmi recordings to learn, and laugh at your failures. For me, there’s no gaming experience like soaring in the air with friends.

Or is there?

I still crave a flight sim with more things to do, more interaction and no clunkiness. The community has been asking for the same since the 00s. We see the progress in the gaming space, and it barely touches us.

Due to my day to day work, I have a hubris. I dumped my ideas into a manifesto. Manifestos must be bold. Here goes mine!

The Open Source Flight Sim Manifesto

Sims are niche. One needs time, knowledge, and devices (joysticks, pedals..) to fully enjoy them.

Sims are a captive market: there’s a small and finite amount of customers. Once you have sold the game to the majority of your potential customers, your revenue dwindles to a trickle. Developers are left needing to “redo” and sell a “new” version of the game, or sell the game again and again in installments, for example by selling new airframes or scenery that take a fraction of the effort to create but sell for as much as the base game. Or similar approaches to obtain revenue.

This means the base game doesn’t grow in horizontal features. No more ways are added for players to interact with each other; nor more mission types, radio comms, radar, SDKs, scenarios, campaigns, mission directors, mision editors, etc. The community feels that the sim could be perfect with more to do besides being a “cockpit museum” and still has the itch, so it implements these with external addons and apps. These provide a clunky UX for newcomers, and the cycle reinforces.

Sims are complex: lots of modelled systems, historical accuracy, multiplayer, performance limits, etc. It gets worse with each sim generation: games cost more to author, which means that less actual interactive features are added. Games have more fidelity and yet less things to do. The game loop gets smaller.

We know. We are sim players. We have been through this cycle. We are part of this small tight-knitted sim community, some of us contributing to these community addons, planes, missions, dynamic campaigns, and more, usually without the help or even against the wishes of the original sim developers. We have seen several iterations of the cycle, with old games and new, and we are tired of it.

It’s not 1995 anymore. There’s another way.

Sims are a captive market: we cannot break it by creating more customers (aeronautics is trending down since the 90s), nor by injecting infinite money into development (like Microsoft funded the old Microsoft Flight Sim family as a PC demonstrator). But we can opt off the market. Sim game studios are not big and usually consist of 2 or 3 teams of developers, each with ~10 people. One can find the same amount of developers by just picking one of the many forums or chat servers the sim community has.

Our rally

A sim game & stack from the community for the community, always.

Complex niche products with strongly defined community and needs are the utmost perfect fit for successful open source projects. For example in 3D suites (Blender), databases (PostgreSQL, MariaDB family), music notation (MuseScore), OS Kernels (Linux), web broswers (Chromium-based, Firefox), game engines (Godot, Bevy), etc.

We firmly believe that open source is the perfect fit for our specific niche project: it allows us to skip the perils of a captive market, it works for a distributed global team and not against it, it scales efficiently with the team, and it amplifies the simbiosis that can be achieved with all the stack pieces across different sims and academia.

The sim community is starving for progress, interactivity and scale. Collaborating asynchronously and openly online is how many successful products are created nowadays.

Our way

Successful open source, developed with open doors, requires a “gift culture” approach in contrast with an “exchange culture” approach. This means ego-less conversations in the open, and social status earned by what you “give away” (knowledge, and work) rather to what “you own” (control over knowledge or projects).

Succesful (open source) projects are those that rally around a lighthouse goal, but are allowed to grow from “my baby” to “our baby”. Like well-raised kids, these projects are allowed to meet new friends that improve them and make someone better, eventually becoming autonomous adults, while keeping their principles intact.

To collaborate and scale the team efficiently, one needs openness. This is a precondition and non-negotiable. We develop in the open, and communicate in the open. We ask our questions in public. We are proud of our mistakes because we strive to grow from them.

We want our stack to be for the community always. We do this by ensuring whatever we create is open source, and the pivotal pieces are licensed under a copyleft license. This ensures modifications must be shared back to the community, forever, even if they are adopted by other actors (possibly competitors) in the space. We put the community first, and we hope this grows community around us.

When authoring a sim game, there’s a lot of byproducts: a Flight Dynamic Model library, accurate 3d models, aircraft specs, game engine, etc. By open sourcing these pieces, we ensure they are valuable not only for the specific niche of sim that we enjoy, but others: the parts are valuable for a combat flight sim, but also commercial flight sims, RC sims, academia, etc. This attracts contributors, community, mindset, and knowledge that is seeded and loops back to us, and is an incredible force multiplier that sims developed behind closed doors simply cannot tap into.

We evaluate and contribute to existing libraries if possible, this strengthens the community.

Our goal

To succeed, we don’t build an identity around what we are against, but focus on what we are for. We state our heading clearly, if not for others, for ourselves.

We want a social combat flight sim. We enjoy the social interactions of flying with teammates and against teammates. We find teamwork and coordination to be exhilarating compared to flying alone. Briefings, formations, taskings, and knowing that you have a wing man, a flight, and a squadron or 2 that have your back meanwhile you cover theirs.

We enjoy Advanced Combat Maneuvers and dogfights, and we train and improve each other, for fun. We want a combat flight sim that enables the community to learn ACM proficiently, not just as “something that happens organically” in the game. We want to close the ACM knowledge gap in the sim community and make the sim community grow.

We also want to play our sim in singleplayer, to practice, develop iteratively, enhance multiplayer with sound game AI, let the game grow organically into critical mass, and welcome newcomers at their own speed to the sim community.

We don’t chase fidelity for its own sake, we only increase fidelity when it increases the player’s ability to make interesting choices, be it avionic system choices or multiplayer choices. The player must be able to explain why something happened and have levers to alter the system: without legibility, the simulation is just hidden math. The system must be performant to allow for hundreds of players. The system must be decopuled to allow to run the sim parts at different speeds for singleplayer or stress tests, and it must be deterministic to allow for competitive games and to easily reproduce bugs with a single seed.

We want a sim game & stack without clunkiness. No 3rd party installers, tools or mods. We want mods that are just first-class plugins to the game. Being open source, we can legally distribute game content as a whole.

We want accurate 3d models, and specs of existing aircraft. Specs don’t change, nor does historical information. Even if this sim stack is archived in the future, its models, missions and logic can legally be reused by the community on new open source games. It also provides a venue for being an actual “plane museum” for the community that can be studied and reused, moreso than closed source games.

We want a next-gen sim game. Sim games need a game engine tailored for them. Usually, sim games develop their own custom game engines. We believe the Bevy Engine is a right fit for simulations such as ours, just like the Bevy authors do.

Bevy’s Entity-Component-System paradigm is next-gen, gives insane parallelization (perfect for our perfomance needs), and matches our extensibility needs. It also makes asynchronous collaboration easier: small, isolated systems mean fewer conflicts and easier parallel work.

Its next-gen features such as meshlets (aka Nanites) rival commercial open source engines such as Unreal. Its physics libraries like Avian are well regarded and fit for a Flight Dynamics Model.

The Bevy engine itself is written in ECS, which means that if one understands our game code, one understands the game engine, it’s all turtles downwards. This also enables us to tap into existing features from the Bevy community. Rust is also a subjectively fun language to program on, which draws people in.

Bevy’s shortcomings at the time of writing, like the lack of game editor or the fast development cycle, only play to our strengths: sims don’t need to create skeletal animations or corridor maps, and the fast development ensures that forks are costlier than working on one shared stack, respectively.

The Bevy community is also a right fit for collaborating in the open, therefore we adopt their social customs and code of conduct, as already listed in this manifesto.

Is it easy? No! Is it doable, and fun? Hell yes! Several mod teams are already working this way on their respective sims.

If this is not for you, it’s ok. If only some parts resonate, I hope we can collaborate for the benefit of both. If this goal and method of working resonates with you, join us!

For now, this is just a rally.

Yet this thought has motivated enough me to publish the avian_fdm crate, a Flight Dynamic Model library for the Bevy Engine and Avian physics crate. If I can tackle this, learn and collaborate, others can too, and together we can have the sim game we dream and talk about.

Write me a line! Or find me for now in the avian_crate thread of the Bevy Discord server.

See you in the skies!