Balance
Lower the difficulty after deaths and raise it after wins, without a see-saw
A game that reacts to every loss feels like it is helping, until it flips back the next fight, or until players learn to die on purpose for an easy ride. This entry shows how to build the adjustment with a pause and limits, and how to check it with a run of outcomes.
Armor looks weak at high values, but is it really? Judge it by health, not by percent
Players say the tenth point of armor is worth less than the first, so you cap it or nerf it. The percent on the screen misleads here. This entry shows which number tells you what armor is really worth and how strong enemies erode it.
After which try do players give up on a boss, and is it the same try for everyone?
The number of tries until a win hides the people who never won. This entry shows how to follow the players who are still trying at each attempt, how many of them leave right then and when a boss has started to wear people out instead of teaching them.
A hero who joins late falls behind the party: how much catch-up bonus is fair?
A bonus for the weak member helps until it makes the whole party wait for them, or until players bench the leader to farm the bonus. This entry shows how to simulate how long the catch-up takes and how to stop the bonus from becoming the best way to play.
The wiki says three percent but players report one: is the drop rate really off?
A few hundred kills are enough to doubt a drop rate, but not enough to prove a bug, and reports from players are biased towards the lucky and the furious. This entry shows how to check your stated rate against logs and how many kills you need to see a given error.
Where would one more checkpoint save players the most replay?
A death near a checkpoint costs seconds and a death far from one costs minutes, and the average hides which stretches are the cruel ones. This entry shows how to measure the repeated distance per death and pick the best spot for the next checkpoint from real death positions.
Which companion do players leave at the camp, once you account for who they could take?
A companion who joins in chapter four is taken in fewer fights than one from the start, yet that proves nothing about how good they are. This entry shows how to compare companions fairly, given when they were available and how many seats the party has.
Which crafting material holds players back, and how long does one full set of gear take?
One rare ingredient can decide how long a whole progression takes, no matter how fast the common ones come. This entry shows how to add up the hours per place where materials are found, name the bottleneck and see what changing one drop does.
Can players make endless gold by crafting and selling? Check every recipe, not just one
A recipe that turns cheap parts into something that sells for more is fine once, and a money machine when the parts can be bought cheaply and the result sold again. This entry shows how to price every recipe through its whole chain and find the loops.
Players pour everything into critical hits: where is the best split, and where is the cap?
Two stats feed the same idea, how often a hit is big and how big it is. Past a certain point, one of them stops paying and players do not notice. This entry shows how to find the point where your numbers break and what to do about it.
Tiny attacks do zero damage after armor, and the order of bonuses changes the result
At low levels a fight can freeze because neither side can hurt the other, and two players with the same bonuses deal different damage. This entry shows what to do about rounding and flat reductions so that every number the player sees adds up.
Which skills does nobody pick, and which one does everybody pick?
A skill nobody takes is wasted work, and one that everybody takes makes the choice fake. But a deep skill is picked less simply because fewer players got that far. This entry shows how to compare skills fairly so you only fix the ones that are really off.