What Happens in a Deep Reorg

A one-block reorganisation is normal. A deep one, reversing many blocks, is an attack or a serious failure, and it has happened.

What a deep reorg does

Transactions in the discarded blocks return to pending. They may be included again, or they may conflict with transactions in the new chain and never execute.

A payment that was confirmed becomes unconfirmed and potentially never confirmed. That is the mechanism of a double spend: the attacker pays, the payment confirms, the recipient acts, and then the payment is erased by a longer chain in which it never occurred.

Where it has happened

On smaller proof-of-work chains, repeatedly.

The economics are straightforward: if the hash power securing a chain can be rented for less than the value of what can be double spent, the attack is profitable. Several chains with small hash rates have suffered successful attacks of exactly this kind.

On large chains it has not happened, because the cost is prohibitive and the attacker would destroy the value of the asset they were attacking.

What protects against it

Confirmation depth. Each additional block makes reversal exponentially harder. This is why exchanges wait, and why the number they wait differs by chain.

Economic finality. Proof-of-stake designs with a finality gadget make reversal past the finalised point require destroying a large fraction of stake, which converts the question from feasible-but-expensive to catastrophic.

Chain security relative to what is at stake. A chain secured by a small amount of hash power or stake is cheap to attack regardless of its design.

What a user should take from this

Confirmation requirements are not arbitrary. They reflect an assessment of how expensive reversal is on that chain.

Small chains carry genuinely different risk. The same number of confirmations does not provide equivalent assurance on a chain with a hundredth of the security budget.

Larger amounts warrant more confirmations. The appropriate depth scales with what is at stake.

The venue perspective

Exchanges set confirmation thresholds per asset, and those thresholds are the market’s practical assessment of settlement quality made by parties with money at risk.

An asset requiring a large number of confirmations on a venue is one that venue does not trust quickly. Comparing those numbers across assets, published by platforms including a venue that supports immediate withdrawal, is more informative about chain security than most published comparisons.

The broader point

Security here is economic rather than absolute. Nothing is impossible; things are expensive.

The question for any chain is what reversal would cost, measured against what could be gained. Where that ratio is unfavourable to an attacker by orders of magnitude, the chain is secure in the only sense the word has in this field.