Michael Kellett – Mym111

There’s a very specific kind of lie that gets told about “educational games,” and it goes like this: play this, have fun, and absorb a real, transferable skill as a side effect, with none of the friction of actually learning it. Usually that’s marketing copy wearing a lab coat. But every so often, a game comes along that doesn’t cheat the premise – one where the code you write to solve a level is the same code, same syntax, same debugging headache, that you’d write at a desk job. I went looking for the ones with the receipts: strong reviews, and more importantly, comment sections full of actual developers leaving praise. Here’s what held up.

Screeps – JavaScript, PC (Steam)

Screeps is the one I keep coming back to when people ask if a game can really teach programming. It’s an MMO where you don’t control units directly – you write JavaScript that runs continuously, even when you’re not playing, to manage a colony that mines, builds, and defends itself against other real players doing the same thing. There’s no hand-holding difficulty curve here. You’re writing actual state machines and resource-allocation logic because if you don’t, your colony gets steamrolled by someone who did. It’s sat around an 89% positive rating on Steam for years, which for something this unapologetically nerdy is a genuinely rare feat. If you already know a little JS and want to see what “production code under pressure” actually feels like, this is it.

The Farmer Was Replaced – Python-like language, PC (Steam)

This one’s a farming sim with a twist: your drone doesn’t move, harvest, or plant anything unless you write the logic that tells it to, in a language deliberately built to look and feel like Python. It’s beloved specifically because it doesn’t let you fake your way through – there’s no drag-and-drop escape hatch. You’ll hit real walls around loops, functions, and optimization the same way you would writing an actual automation script, and the game is honest about that friction being the point. A lot of the praise for it comes from people who say it’s the first time a “coding game” made them debug something out of genuine frustration rather than being led to the answer.

SHENZHEN I/O & TIS-100 – Assembly-adjacent, PC (Steam)

I’m grouping these two because they’re from the same studio (Zachtronics) and scratch the same itch. Neither one teaches you a language you’ll drop into a job application, but ask any working engineer who’s played them and they’ll tell you these games taught them how to think like a programmer – how to break a problem into the smallest possible instructions, how to work within brutal constraints, how to read documentation (SHENZHEN I/O literally comes with an in-game manual you have to reference like real hardware datasheets). TIS-100 has you programming imaginary assembly-like processors to solve puzzles that get genuinely nasty. These are cult favorites among actual software engineers precisely because the skill they build – decomposing a problem before you touch a keyboard – is the part coding bootcamps often skip.

Turing Complete – Logic gates up to a full CPU, PC (Steam)

Full disclosure: this one is not fun, exactly, and I say that with real affection. Turing Complete starts you with a single logic gate and makes you build your way up – gate by gate, component by component – into an actual working CPU with its own instruction set, at which point you start writing assembly for the processor you just built. It’s less “game” and more “the most patient, well-designed hardware course you’ll ever sit through,” and that’s precisely why it works. Where CodeCombat teaches you to write code and SHENZHEN I/O teaches you to think within constraints, Turing Complete teaches you why any of it works in the first place. If you’ve ever wanted the “oh, THAT’S what’s actually happening under the hood” moment, this is where you go get it. Just don’t expect it to hold your hand, or really to smile at you at any point.

Opus Magnum – No code at all, PC (Steam)

I’m breaking my own rule here, because Opus Magnum doesn’t ask you to write a single line of code – it’s a puzzle game about building alchemical machines out of arms, tracks, and repeating instructions to transmute raw materials into finished goods. But I’m including it because the thinking it builds is the same thinking that makes someone good at code, just with the syntax stripped away entirely. Every solution is a small machine made of reusable, composable parts that all have to run in a synchronized loop, and optimizing one means thinking in exactly the terms a programmer thinks in: sequence, timing, dependency, and how the pieces interact rather than just what each piece does alone. It’s also, not coincidentally, one of the most beautiful games on this list – brass and glass and looping arms clicking through their motions like a machine you’d actually want on your desk. If you want to build the instinct for componentized, systems-level thinking before you ever touch a keyboard, this is a wonderful, quiet way to do it.

Satisfactory & Factorio – Systems design and data flow, PC (Steam; Satisfactory also PS5/Xbox)

I’m putting these two together because they’re solving the same problem from two different angles, and if you’ve played one you basically know the shape of the other. Both are factory-building games where your job is to take raw resources in one end and turn them into increasingly complex finished products on the other, by designing conveyor and pipe networks that have to balance throughput, avoid bottlenecks, and scale without collapsing under their own complexity. Neither one has you writing code, but what they teach maps almost embarrassingly well onto real systems and data engineering work. You’re constantly thinking about throughput versus capacity, where a pipeline backs up, how to refactor a section of your factory without breaking everything downstream of it, and when to stop patching and just rebuild the whole line. If you’ve ever built a data pipeline and had to explain why one slow step upstream tanked everything downstream of it, congratulations, you already understand Factorio’s entire thesis. Factorio is the tighter, more mechanically focused of the two; Satisfactory trades some of that precision for scale, verticality, and a first-person view of your own sprawling mess, which I happen to love.

Bitburner – JavaScript, PC (Steam, also browser/free)

Bitburner is a hacking-themed incremental game where you write real JavaScript (or TypeScript, if you want) to automate scripts that run on in-game servers, earn currency, and eventually take over the world’s financial system, as one does. It’s free to start in-browser, which makes it an easy first stop, and it uses an actual code editor with syntax highlighting and error messages that mirror what you’d see in VS Code. What people highlight in reviews is how naturally it teaches automation thinking – writing a script once and having it scale – which is a genuinely useful mental model to carry into any real job that touches scripting or ops work.

RoboCo – Python, PC (Steam)

RoboCo has you building and coding custom robots, and it goes out of its way to mimic a real dev environment: it supports external IDEs like VS Code, includes a debug console with actual error tracking, and uses real Python rather than a simplified dialect. It’s more approachable than Screeps or the Zachtronics games, which makes it a solid on-ramp if you want your first steps in a real language to happen somewhere a little more forgiving before you go pick a fight with something harder.

CodeCombat – Python, JavaScript, and others, PC/Browser (also mobile-friendly)

CodeCombat is the most accessible entry here, and I mean that as a compliment, not a knock. It’s an RPG where you type real code to move your character through dungeons, and it ramps from “your first for-loop” to genuinely non-trivial problem-solving without ever switching you over to drag-and-drop blocks. It supports several real languages, so you can pick the one that’s actually relevant to what you’re trying to learn, and it’s cheap enough (with a solid free tier) that it’s an easy recommendation if you want proof you can stick with this before spending real money elsewhere.

The honest caveat

None of these will hand you a job-ready portfolio on their own – none of them are, or claim to be, a replacement for building a real project with real stakes and no built-in win condition. What they’re good for is the part that traditional tutorials are bad at: making you write code under a little bit of pressure, with a goal you actually care about, often enough that syntax stops being the obstacle and starts being the tool. That’s a real head start. Where you take it from there is still on you.