What real-world costs of the traditional "trust-based" payment model motivate a trustless electronic-cash system?

Because banks must be able to mediate disputes, no online payment is ever truly final — that reversibility raises costs, kills tiny payments, and forces merchants to distrust their customers.

Online commerce relies on financial institutions as trusted third parties. That trust model has built-in costs:

• Nothing is truly non-reversible — the institution can always reverse a payment to settle a dispute, so completely final payments aren't possible.
• Mediation raises transaction cost, which sets a minimum practical payment size and cuts off small, casual micropayments.
• Merchants must distrust customers, demanding more personal information than they'd otherwise need, and a certain fraud rate is simply accepted.

Physical cash avoids all this in person — but before Bitcoin, nothing did so over a network. The goal: a payment system based on cryptographic proof instead of trust, so any two parties can transact directly.

Go deeper:

• Bitcoin: A Peer-to-Peer Electronic Cash System — Nakamoto (2008) — the primary source; the abstract states the trust-vs-proof problem in two sentences.
• But how does bitcoin actually work? — 3Blue1Brown — builds the whole system from scratch, starting with why a shared ledger needs no bank.
• Bitcoin — Wikipedia — history and context of the trustless-cash idea and its launch.