Idle Game Design Principles

The article came in late this week. I was crunching over the weekend to help a friend design the economy for their idle game. Despite having “economy design” in my previous job title, I had never designed an economy from scratch before. Idle games are perhaps the purest expression of economy design, as there’s practically zero other gameplay. It was a fun exercise!

The game itself was very similar to Cookie Clicker, with a few minor twists. First, there are multiple currencies. Second, all rewards are doled out via loot boxes rather than direct currency payouts. In working on the design, I gravitated towards a few principles that I based my design around.

Reengagement Frequency

The basic loop of an idle game is:

1. Generate resources

2. Invest those resources to speed up generation

3. (Repeat)

The game is “idle” because players can pause the loop, go AFK, and come back to a fat stack of resources to reinvest. While progress is fastest with active reinvestment (similar to continuously compounding interest), you don’t lose all that much from those pauses.

Our game contains several “clocks” that encourage reengagement after a certain time. For example, the Creamery produces Cheese every 30 minutes, and has a maximum stash of 10. If Cheese remains uncollected for 5 hours, the stash fills up and the Creamery stops producing. To optimize Cheese production, players are encouraged to return at least every 5 hours to collect their Cheese.

Initially, new players are highly engaged and actively reinvesting in the game economy. In their first play session, they might be actively playing for 15 to 60 minutes. As their interest wanes, the frequency and duration of their play sessions drops. Players might check in once an hour, then a couple times a day, and eventually once a week or so. Ideal idle game pacing follows that decaying engagement.

To address this, the various clocks have exponentially longer optimal wait times. Milk-producing Cows cap every 20 minutes. Cheese-producing Creameries cap every 5 hours. Engine-producing Shipyards cap every 2 days. This allows players to feel good about optimizing certain processes even if they fail at others. A player who is idle for 4 hours might feel bad about failing to handle their Milk production, but they can feel good about optimizing their Cheese production.

Differentiation and Playstyles

Unlike Cookie Clicker, our game contains multiple currencies. While in early drafts, these functioned simply as larger denominations (ex: copper, silver, and gold in WoW), I realized this created an opportunity for playstyle differentiation.

Most games create strategic resource allocation by allowing players to specialize. Consider the talent tree in your favorite RPG. Players specialize in their preferred playstyle by growing a particular branch of the tree; sneaky players invest in stealth perks, brawlers invest in combat durability, magic-wielders might ignore defense in favor of more powerful spells, etc. Real-time strategy games often offer a similar tradeoff with a “basic” resource and a “tech” resource (ex: gold and wood in Warcraft3, minerals and gas in Starcraft). Large armies and expansive bases are mineral-intensive, but upgrades and high-tech units are gas-intensive, and players need to plan their resource income according to their strategic needs.

Idle games don’t offer much in terms of playstyle variety. But one simple differentiator is the aforementioned reengagement frequency. Some players might play all day with a browser tab open, checking in every 15 minutes. Others might only check their phone while copping a squat or before bedtime. These heterogenous players can optimize their resource production towards their preferred reengagement frequency.

In our game, Cows have short clocks and Creameries have long clocks. So the browser tab player is encouraged to upgrade their cows since they will be able to take full advantage of their short production cycles. But the pooping player focuses on Creameries instead, since their production clock better matches their reengagement clock.

Exponential Growth and Diminishing Returns

Progression is at the heart of idle games. There’s really not much going on besides your farm growing and your numbers going up. Thankfully, numbers are easy to change, which is part of why idle games are so cheap to create. But the thing about numbers going up is that, once you’re acclimated to a higher baseline, the same growth doesn’t hit as hard. Like a junkie, you’re fiending for a bigger hit just to feel something, to feel anything.

The drug reference isn’t just a joke – there’s some psychophysics behind this. When human brains perceive sensory input, be it light, sound, or smells, they are usually on exponential scales. Psychologists measure this with a just-noticable difference. Experimenters adjust the intensity of some sensory input to see when humans can notice differences, and they are usually exponentially scaled. For example, people will notice a difference between a pile of 5 and 6 beans, and they will not notice a difference between a pile of 100 and 101 beans, despite each having the same +1 bean difference. But people will notice a difference between 100 and 120 beans, which has the same x1.2 difference as the 5 and 6 piles. This same phenomenon is observed with light intensity, spiciness, psychoactive drugs, etc. Audio scales even account for this; decibels are actually an exponential scale against energy, not linear.

This same exponential growth works in idle games. To ensure that growth rates are always noticeable against a new baseline, everything increases exponentially. For every Creamery level-up, their Cheese production rate rises by x1.1. However, exponential growth gets out of hand quickly, and we generally want to slow down the rate of progression as players progress. Once again, we can solve this with exponential growth: the cost to level up the Creamery rises by x1.15 per level.

Another nice attribute of exponential growth is that they ease tuning of initial values. If the initial Creamery cost is set stupidly low, early progress will be incredibly quick. But eventually the exponential cost growth will reach a point where level-ups slow down to a reasonable rate.