Bitcoin's issuance is determined entirely by rules every node enforces, with no discretion at any point. Understanding why it is difficult to change explains more than the schedule itself.
Issuance happens only in the coinbase transaction
Each block begins with a special transaction that creates new coins and pays them to the miner, and it is the only way coins come into existence.
Every node checks that this transaction creates no more than the rules currently allow. A block claiming more is rejected as invalid.
The reward amount is a function of block height, so any node can compute the correct value independently and no coordination is required.
The halving is arithmetic on block height
At fixed intervals of blocks, the reward is cut in half, and the sequence of halvings produces a total that converges on a fixed ceiling.
Because it is measured in blocks rather than time, the actual dates drift with how quickly blocks are found, which is why halvings do not fall on calendar anniversaries.
The reward eventually becomes too small to represent in the smallest unit, at which point issuance stops entirely and miner revenue comes from fees alone.
Changing it requires almost everyone to agree
Nodes enforce the rule independently, so a block with a larger reward is rejected by any node running unmodified software.
Increasing issuance would therefore require operators, exchanges, custodians and users to adopt new software voluntarily, and the change would dilute exactly those holders.
The difficulty is social rather than technical. Editing the number is trivial; persuading a network of independent participants to accept the edit is not.
Scarcity is a property of enforcement
The cap means nothing on its own. What gives it force is that a large number of independently operated nodes reject blocks that violate it.
This is why node distribution is treated as important beyond redundancy, since verification is what turns a rule into a constraint.
A network where most users trusted a small number of operators would have the same written rule and a much weaker guarantee behind it.
The long-term question is fee revenue
As the reward shrinks, the security budget increasingly depends on what users pay to transact rather than on newly created coins.
Whether fees will be sufficient is a genuinely open question, and it depends on transaction demand at the base layer as activity moves to systems built above it.
The design settles the supply question completely while leaving this one to be answered by how the network is used, which is the trade the schedule makes.