What game development actually is — and the path I’d take if I were starting in 2026

People tell juniors that game development is “making games.” That sentence hides the job. Most of the hours I have watched — and a chunk I have worked, on tools and a couple of shipped small titles — are not trailer moments. They are import pipelines, a physics quirk on one platform, a producer asking why the build is 900MB, and a designer changing a number that invalidates a day of tuning. If I were starting in 2026 I would train for that texture of work, not for a cinematic reel.

I am not a twenty-year AAA director. I have shipped two small commercial games with a tiny team, contributed tools on a mid-size Unity project, and spent enough time adjacent to engine people to know which stories are real. This is the path I would take if I were twenty-two again and allergic to mythology.

What the work is, when you strip the keynote

A game is software with a loop, assets, and a feeling. The loop is code: input, simulation, render, maybe a net tick. The assets are someone else’s week: a mesh, a clip, a shader graph, a voiceline. The feeling is the part that does not show up in a Jira workflow until it is missing. Game development is the job of making those three stay in the same build without lying to each other.

That means you will spend more time in Unity, Unreal, Godot, or a custom editor than in a blog about ECS. You will spend time in Perforce or Git LFS because a 4K texture is not a source file you rebase for fun. You will spend time in a spreadsheet of drop rates. You will spend time waiting for a cook or a shader compile. If that sounds boring, the industry will still hire you and then surprise you. Better to be unsurprised.

It is also several jobs wearing one Steam page. Programming, design, art, audio, QA, production, live ops, community. A solo itch.io game lets you wear all of the hats badly. A 200-person title lets you wear one hat and never see the trailer until launch. Both are “gamedev.” The skills transfer less than Twitter implies. A gameplay programmer is not a technical artist is not a narrative designer. I would pick a hat early enough to get good, late enough to know I liked the hat.

The path I would take in 2026, in order

Year zero: finish things that other people can play. Not a design doc. A build. Godot 4 or Unity, not a from-scratch engine, unless the engine is the hobby. I would ship a tiny game every month for six months: a score attack, a local two-player, a narrative toy in Twine or Ink if I wanted writing. The muscle is closing. The industry is full of people with a folder of prototypes and no itch.io page.

I would learn the editor’s ugly corners: addressables or resources, import settings, a basic CI that produces a build (GitHub Actions for Godot; Unity’s own cloud if someone else pays). I would learn Git well enough not to wreck a teammate, and I would learn why studios still use Perforce when the art folder is huge.

I would not start with Unreal “because Fortnite” unless I already knew I wanted that look and that pipeline. Unreal is a career. It is also a way to spend a year in the material editor and never ship. Godot is the honest 2026 starter for a person who wants to finish. Unity is still the job-listing gravity in a lot of mobile and mid-core. I would touch both enough to be literate. I would go deep on one.

The first job: tools, gameplay, or QA that leads to either — not “generalist at a dream studio.” I would take a junior gameplay or tools role at a small studio, or a QA role that lets me sit next to programmers, before I held out for a narrative job at a famous name. The famous name will hire a specialist. The small studio will let me see a ship. Seeing a ship is the education.

If I were more programmer than designer, I would learn C# well (Unity / Godot), or C++ if I was aiming at Unreal systems, plus a scripting language the designers already use. I would write gameplay that is data-driven enough that a designer can change a curve without a rebuild. That is the job. Clever architecture that a designer cannot touch is a resume piece, not a game.

If I were more designer than programmer, I would still learn enough Blueprints or GDScript to prototype without a ticket. Designers who cannot prototype wait in line. I would build levels, write a GDD that is a page, and playtest with strangers. I would not collect GDC talks as a substitute for a build.

After the first ship: pick a lane that still ships. Live ops and economy design are huge in 2026 because games are services. I would only go there if I like spreadsheets and ethics arguments. Graphics and engine are a long apprenticeship. Tools programming is the underrated path: you make the other people faster, you still write real systems, and you are employable when the gameplay team is full. I would seriously consider tools if I liked editors more than I liked death animations.

What I would not do, even if the thread said to

I would not write an engine in C++ as my first public work. I would not spend two years on a multiplayer architecture before I have a fun single-player minute. I would not treat Unreal Marketplace packs as a portfolio — art directors can smell it. I would not ignore math. Linear algebra shows up. I would learn enough to debug a transform, not enough to stall on a graphics PhD unless that is the lane.

I would not assume AI image and codegen replace a pipeline. They will write a shader comment and a placeholder texture. They will not sit in a playtest and notice the camera makes people sick. I use Copilot in tools code. I do not use it as a design director.

I would not move to a AAA city before I had a shipped something and a reason. Remote contract work and small remote teams exist. Relocation is a bet. Make it after you know you like the hours. The hours are real. Crunch is not every studio and it is also not a myth. I would ask, in the interview, what the last six months before ship looked like, and I would believe the people who flinch.

Schools, bootcamps, and the portfolio that actually gets a reply

A games program can help if it ships team projects and if the teachers have shipped. It can also be an expensive LAN party. I would look at graduate credits: not “we used Unreal,” but “here is the build, here is my role, here is a postmortem.” A CS degree plus a portfolio of finished small games is still a fine path. A bootcamp that is really a Unity tutorial with a logo is not.

The portfolio I would hire from in 2026: three playable builds, one of them with a teammate, one of them with a sentence about what I cut. A GitHub of engine fragments does not beat a five-minute game that feels good. A vertical slice that never became a game is weaker than a complete tiny game. Complete is the signal. We are an industry of abandoned slices.

I would put the role I want in the first line of the readme. “Gameplay programmer — I owned player movement and camera in this Godot title.” Recruiters do not decode a generalist vibe. I would include a short video because nobody will install your engine version on a Tuesday.

Money, stability, and the second career people do not mention

Games pay worse than product SaaS for the same programming skill in a lot of markets. I would go in knowing that. I would keep a boring backend skill warm — I have watched friends move to product companies when a studio folded, and the ones who had shipped web services ate. That is not a betrayal of the dream. That is a runway.

Indie as a living is a studio with a marketing plan, not a Steam key and hope. I would treat the first commercial indie as a product: wishlists, a trailer date, a scope I can finish. I would not quit a job for a passion project that is “almost done” at 40%.

Mobile and live service work is most of the employment. If that content is not what I want to play, I can still do the job if I like the craft of systems. If I cannot stand the craft of systems without loving the fantasy, I would stay small and poor on purpose or I would leave. Self-knowledge is cheaper in year one than in year eight.

The first twelve months, if I were starting Monday

1. Install Godot 4. Finish a tiny game this month. Put it on itch. Ask three strangers to play.
2. Do it again, with one teammate, even if the teammate is a musician from the internet.
3. Learn C# or GDScript deeply enough to debug without Discord. Read the docs for physics and animation player. Avoid a custom engine.
4. Apply to junior / associate / QA-at-a-dev-shop roles with the two builds. Write the role in the subject line.
5. Keep a notes file of every ship I see: what slipped, what the tool did badly, what the designer changed too late. That file is the start of taste.

Game development is making a feeling survive contact with a build machine, a store page, and a person who does not owe you attention. I would start there. The path is finish, pick a hat, see a ship, then specialize. The mythology is a studio tour and a rendering talk. I like the talks. I would not let them replace a download link.