Building an MMO alone: the plan

An online RPG is the classic project a solo developer should not attempt. The scope, the order of work and the rules that are meant to make this one finishable.

“Do not try to make an MMO on your own” is some of the oldest advice in game development, and it is good advice. No Kings Left is an online action RPG made by one person. This post is about why that is not as reckless as it sounds, and where it could still go wrong.

Scope: what the game is not

The plan works by leaving things out.

  • No open world. The game is made of towns and short instanced dungeons. A dungeon is a contained piece of work that one person can build, test and finish.
  • Four classes, no more. Four classes with fourteen skills each is already a lot of skills to make feel good.
  • Simple systems. Gear comes in five tiers with no rarity colours. Crafting is a recipe, ingredients and a button. Each system should fit in a paragraph.
  • Desktop browsers only. One platform, one input method, one frame-rate target.
  • No PvP and no guilds in the first version.

Every item on that list is something players reasonably want. Each one is also months of work, and a small game that exists beats a large one that does not.

Order: single-player first, on a multiplayer design

The usual way a solo online game dies is this: build a single-player game, then discover that adding other players means rewriting it.

To avoid that, the game is a client and a server from the first day. Until the multiplayer phase, the server simply runs inside the browser tab. Combat, loot and dungeons are built against a real client and server split, with no machines to run and nothing to pay for. When multiplayer arrives, the same server code moves onto a real server. How a 3D MMO runs in a browser tab has the details.

The roadmap follows from that: foundations, then combat, then the character and its gear, then towns and dungeons, and only then multiplayer, parties and trade, and a first closed playtest.

Rules that keep it honest

A few rules run through the whole project.

  • Decide things once and write them down. Every design decision is recorded with its date. When a new idea contradicts an old decision, that gets noticed and settled, not quietly forgotten.
  • Test before committing. Features that could go either way have to prove themselves first. The combo test is the clearest example.
  • Build tools early. The map builder is being made before the maps, because one person cannot hand-place a world without good tools.
  • No dates that cannot be kept. There is no release date. There is an order of work.

Where it could go wrong

It would be dishonest to stop there, so here are the real risks.

The amount of content. Skills, monsters and dungeons all have to be made, and there are many of them. This is the largest risk in the plan.

Multiplayer arriving late. Even with the design above, the first time two real players fight together will surface problems that no single-player test can. That is why the game is meant to be developed with artificial network delay from the start.

A small player base. Level ranges split players into groups, and a small game can leave someone alone in theirs. Dungeons that work for one player, and classes that do not need each other, are partly an answer to that.

Why do it at all

Because a real online action RPG that starts when you open a link, with short dungeons, gear worth chasing and no shop selling power, is worth making. A plan like the one above is the only way one developer gets to make it.

This devlog will follow the work as it happens, including the parts that take longer than planned.