Seedfolk Sim Tuning: Making the First Loop Testable

A Seedfolk build note on using simulation guardrails, RC1 tuning, and playtest judgment to make the first mystery-seed loop clearer without turning a cozy game into a spreadsheet.

By Isaac Johnson · Published July 7, 2026

The current Seedfolk tuning pass is about a small question with a lot of consequences:

Does the first loop make the player want to plant one more seed?

That is the real target. Not a perfect economy. Not a spreadsheet that looks balanced from across the room. Not a first build that tries to carry every future genetics, breeding, mutation, and collection idea at once.

The first loop is quieter than that.

Mystery Seed
-> species reveal
-> Field Journal
-> Village Note
-> reward
-> another planting choice

If that loop feels good, the whole project has somewhere to grow. If it feels muddy, every later system inherits the confusion.

So this RC1 pass has been less about making Seedfolk bigger and more about making the first promise testable.

The tuning pass is not trying to automate taste. It is trying to give taste better evidence.

The soft loop has hard edges

Seedfolk is a cozy farming game, so the player-facing experience should feel gentle:

Plant something unknown.
Watch it become identifiable.
Learn that the discovery mattered.
Use that crop to complete a request.
Receive something that makes the next planting decision feel possible.

That sounds simple. Underneath it, the economy can drift quickly.

Too many Mystery Seeds and discovery stops feeling special. Too few and the game can stall. Too many Village Notes and the player starts reading the board as noise. Too much currency before there is a real sink and the reward becomes decorative. Too much waiting before the first meaningful reveal and the core fantasy does not get a clean introduction.

Those are not only numerical problems. They are emotional problems wearing numerical clothes.

The RC1 tuning pass gives me a way to look at both sides at once. I can ask whether the player has enough to do, whether the first reveal lands quickly enough, whether the request board arrives at the right moment, and whether rewards are helping the player continue rather than simply decorating the inventory.

What RC1 is trying to protect

The current baseline starts from a deliberately narrow route.

The player gets three plot kits, three Compost Blends, and three starting Mystery Seeds. The first starter identities are scripted as carrot, raspberry, and potato so the opening route can be tested repeatably. The first Mystery Seed reveal is shortened enough to teach the discovery moment quickly, while the full harvest timing still matters afterward.

That distinction is important. A reveal is not the same thing as a harvest.

The reveal teaches:

This unknown thing has become knowable.

The harvest teaches:

This known thing can now be used.

If those moments collapse into one another too early, the player may understand that a timer finished, but miss the more interesting idea: every seed begins as a small botanical question.

The RC1 route also stages Village Notes one visible note at a time. That may sound like a tiny UX decision, but it changes the feeling of the board. A single visible note says, “Here is the next useful thing.” A pile of early notes says, “Please sort the designer’s thoughts for them.”

For the first route, I want the game to offer direction before it asks for strategy.

Plot kits are capacity, not churn

One of the clearer decisions in this pass is that Act 1 plots do not expire.

That means plot kits are durable capacity, not a repeat consumable sink. Placing a plot spends a kit, but removing that plot should return the kit. In the first playable route, plot removal stays out of the player’s hands because it is too easy for a cold player to delete the little bit of farm capacity they need before the mental model has formed.

That is a good example of tuning as product design, not only balance math.

The question is not simply:

How many plots can the player have?

It is:

When should the player understand plots as expandable capacity?
When should they be allowed to reshape that capacity?
What can go wrong if they do it before the game has taught the loop?

The Supply Shed carries that progression. Extra Garden Plot Kits unlock after enough Species Folio progress. Compost becomes repeatably available later. Mystery Seeds do not move into the shop, because buying mystery directly would turn the core discovery fuel into a vending-machine currency. For now, Mystery Seeds belong in the request loop.

That keeps the loop readable:

Requests help you keep discovering.
Discovery helps you satisfy better requests.

Simulation is a guardrail, not a designer

Alongside this tuning work, I have been reading GEEvo: Game Economy Generation and Balancing with Evolutionary Algorithms, a 2024 paper by Florian Rupp and Kai Eckert. The paper describes an abstract framework for graph-based game economies, simulation runs, and objective-driven balancing with evolutionary algorithms.

I am not applying that framework wholesale to Seedfolk.

Seedfolk does not need an optimizer to decide what a carrot should mean. It does not need a machine to tell me whether a reveal feels magical, whether a request feels generous, or whether a player smiles at the right moment.

But I like the discipline behind the paper:

Define the target.
Run the system.
Inspect the pressure, bursts, stalls, and dead ends.
Bring the result back to design judgment.

That is the lane Seedfolk’s balance lab is meant to occupy. It is not the game engine. It is a design instrument. It lets me run progression and economy assumptions before they become player-facing promises.

For RC1, the useful questions are practical:

How long does it take to reach the first reveal?
How long until three species are discovered?
When do requests become available?
Does the first request completion happen before the player loses the thread?
Is there Mystery Seed starvation?
Can the player reach a dead-end state?
Are plot kits creating useful pressure or accidental friction?
Do generated Farmers Market requests arrive as future action instead of already-completed freebies?

None of those answers can prove the game is good.

They can, however, show me where the first loop is lying to itself.

Waiting can be active or empty

One of the trickier parts of tuning a cozy game is that waiting is not automatically bad.

Growth time is part of the fantasy. A seed should not feel like an instant transaction. A crop becoming ready should have a little tension, a little anticipation, a little “I want to see what happens next.”

But there is a difference between engaged waiting and dead air.

Engaged waiting sounds like:

My crop is growing, and I know what I am hoping to do with it.

Dead air sounds like:

Something is happening somewhere, but I do not know what the game wants from me.

The RC1 timers are trying to create the first kind. The opening reveal comes quickly because the game needs to teach the premise. Later starter crops take longer because the player should begin to feel the rhythm of tending, waiting, harvesting, and choosing the next action.

The current targets are intentionally modest: a very short first reveal, a first harvest around a few minutes in, a first request claim around the early-session mark, and an Act 1 handoff that can support a longer one-hour arc.

Those are not sacred numbers. They are testable promises.

Recovery should be invisible until needed

Another quiet part of this pass is deadlock recovery.

Seedfolk needs to protect the player from true no-path states. If someone has no Mystery Seeds, no planted UID seeds, no known seed stacks, no plot kits, and no claimable requests, the game should offer a recovery path.

But recovery should not become the normal economy.

That is why the recovery Field Pack is hidden until the player is actually stuck. It can grant enough material to restart the loop, but it should not become a faucet that players route through on purpose.

This is one of those places where the design goal is almost invisible when it works. A player should not think, “Ah, the recovery system has saved me.” They should simply feel that Seedfolk did not abandon them.

Good guardrails usually feel like mercy, not machinery.

The human test still wins

The sim can tell me a lot. It can show request density, seed pressure, idle time, plot capacity, unlock timing, and deadlock risk. It can point toward contradictions that are hard to see when I am playing one hand-authored route in the browser.

But the human questions still matter most:

Did the Mystery Seed feel like a mystery?
Did the reveal feel clear and rewarding?
Did the Field Journal make the discovery feel permanent?
Did the Village Note give purpose without crowding the player?
Did the reward make the next planting choice feel inviting?

That is the heart of this pass.

Seedfolk is not trying to become a balanced economy in search of a game. It is trying to become a little world where curiosity turns into knowledge, knowledge turns into care, and care turns into one more small decision.

The sim helps me tune the rails.

The playtest tells me whether the train has a reason to move.

And for this first loop, the question stays beautifully small:

Do you want to plant one more seed?