Dev Blog

Spent the morning replaying internal matches. The Husk still arrives a beat too cleanly when the trio leaves the Crucibles alone. Treating it less as a balance lever and more as a readability problem — the Scions need a clearer "now or never" cue before things tip. Sketching a tell I can iterate on this week.

Other notes to self: the Luminor still rewards passive play more than I want, and the Arcanist's cleanup loop could use one more interesting decision per cast. Nothing settled. Just where my head is today.

Hardest part of 3v1 is not balance, it is attention. The Summoner sees the whole map from the Veil; the Scions don't. Put two real objectives on the map at once and the trio actually has to talk — who peels back, who keeps pressure on the Crucible, who eats the next wave of Ichor-kin.

Tried it with three concurrent objectives. Felt clever on paper, miserable in voice chat. Two is the sweet spot for now.

Re-tuned a pile of telegraphs over the holidays. The rule I keep landing on: a Scion who dies should be able to say what killed them in one sentence without pausing the game. If the answer is "I dunno, something gold and red," I broke it.

Killed two abilities outright that looked great in slow-mo and turned into particle soup in actual matches. They will come back eventually, in quieter clothes.

Spent November on the prep flow — picking a Scion, setting your stance, getting into a match without fighting menus. Boring work. But I am allergic to making people redo their setup every time I move a number around. Whatever you configured last week should still feel familiar today.

Same goes for the Summoner's deck. Rather move slowly on UI than make people relearn it.

Most of October was the part of the game nobody screenshots: cold start, finding a session, sliding into the actual match. The goal is that you never feel a hand-off. No "okay, now we're online." No mystery lobby room. Just press play.

Doing this early was uncomfortable. Doing it later would have been worse.

Asymmetric authoritative simulation sounded romantic until I had to make four people share a clock. Spent September quietly cleaning up ownership, spawn order, and what happens when somebody Alt+Tabs at the worst moment. None of it ships as a feature. All of it ships as "matches stop falling apart for no reason."

Lessons learned the hard way: trust the server, distrust optimism, and assume two clients will disagree about the world unless you make them agree on purpose.

Drew some hard architectural lines this month. Scion-side systems cannot reach into Summoner-side systems. Experimental features live in their own sandbox. Core flows do not import anything that is not stable. Boring rules, but they let me try wild stuff later without breaking the spine.

Future-me thanks past-me roughly once a week.

Resisted the urge to slap together a vertical slice. Instead, every system gets its own ugly little test room — a stage with one job, one acceptance check, no scenery. Combat math, UI smoke, onboarding hooks. They earn their way into a real scene.

Slower to show off. Much faster to debug.

A lot of early Blood & Ichor was not code. It was writing. Pillars for the asymmetric fantasy, milestone shapes, short specs that said what each system should do and how I would know it was working.

If a sentence could not be falsified by a playtest, it did not earn a place in the doc. That rule has saved me from more rabbit holes than any framework choice.

Before the first scene held anything but primitives, the game existed as documents. The roles. The emotional promise of asymmetry. The non-negotiables — trio coordination on one side, cerebral pressure on the other. Everything else was provisional.

Those early notes were deliberately shallow on how and strict on what. What should you feel responsible for? What should losing teach you? What should never depend on hidden knowledge? Implementation came later. Intent came first.