Transaction fees on a public blockchain can move from negligible to substantial within hours. The cause is a fixed supply of block space allocated by auction.
Block space is a scarce resource
A blockchain produces blocks at a roughly constant rate, and each block holds a limited amount of data. Throughput is therefore capped by design rather than by capacity constraints.
Transactions waiting to be included sit in a pool. When submissions exceed what blocks can hold, the pool grows and waiting times extend.
The cap is deliberate, since larger blocks would increase the burden on the nodes that store and verify the chain.
Inclusion is decided by fee, not order
Those producing blocks select transactions to include, and they generally prioritise by the fee offered per unit of data rather than by arrival time.
The result is an auction. When demand is high, the fee required to be included rises until enough transactions drop out to fit the available space.
A transaction offering too little is not rejected; it simply waits, potentially indefinitely, until competition subsides.
Demand arrives in bursts
Activity is not evenly distributed. Periods of sharp price movement, popular token issuances and automated activity all concentrate demand into short windows.
Because supply cannot respond, the entire adjustment happens through price. Fees can rise by a large multiple and fall back just as quickly.
This makes fees difficult to predict in advance, which is a practical problem for anyone timing a transfer.
Transaction size matters more than value
Fees are generally charged on the data footprint of a transaction rather than the amount being moved. A large transfer can cost less than a complex small one.
Transactions interacting with contracts consume more resources and therefore cost more, which is why simple transfers and complex operations price differently on the same network.
Consolidating inputs during quiet periods is a common way to reduce the footprint of later transactions.
Scaling approaches move activity elsewhere
Secondary layers batch many transactions and settle them onto the base chain periodically, so participants share the cost of a single settlement.
This reduces fees for activity on the secondary layer while leaving base layer pricing unchanged, since the underlying scarcity has not been removed.
Moving between layers has its own costs and its own trust assumptions, which differ considerably between designs and are worth understanding before relying on them.