LibraryConsensus2015Design paperCorpus record
Secure High-Rate Transaction Processing in Bitcoin
GHOST. Yonatan Sompolinsky and Aviv Zohar.
At high block rates the longest chain throws away too much honest work. GHOST follows the heaviest subtree, so blocks off the main tip still count for the fork choice.
GHOST picks the heaviest subtree rather than the longest chain, so honest blocks that lost a race still count as work for the fork they support.
- Evidence
- Primary paper
- Re-measured
- No
- Assumptions
- 4
- Records linked
- 6
Code
Source, repository, and what is actually hosted
- ePrint 2013/881https://eprint.iacr.org/2013/881
- The source URL is not a GitHub repository. No code remote is invented.
- Full text is not mirrored. Public availability is not a licence to copy the file onto this site.
- Add to a reading list
01 Claim ledger
What the paper is allowed to say
Each row is a sentence already in the study. The status is the same on every row: a model claim, not a live measurement.
Claim 01 · Paper model
Fast blocks collide
If the interval is short, two honest miners often find a block before either has seen the other. Longest-chain throws one away. The attacker, publishing in private, collides less and keeps more of their work.
Claim 02 · Paper model
Count the subtree
At a fork, follow the child whose descendant tree contains the most work. An uncle that lost the tip still adds weight on the side that mined it.
Claim 03 · Paper model
The attacker must beat the forest
Overtaking means outpacing the honest subtree, not a single path that got lucky with orphans.
Claim 04 · Paper model
Ethereum's uncles are not this protocol
Paying uncles and mentioning them is related. GHOST is a fork-choice rule. A chain can pay uncles and still follow the longest chain.
02 Three cuts
Observation, model, falsifier
A desk does not stop at the summary. Each claim is cut three ways, using only this study's own assumptions and checks. Nothing here is a new figure.
01 Fast blocks collide
Observation
What the study says
If the interval is short, two honest miners often find a block before either has seen the other. Longest-chain throws one away. The attacker, publishing in private, collides less and keeps more of their work.
Model
What has to hold
Work in the subtree is visible. A block that never propagates adds nothing.
Falsifier
What would retire it
What block interval is assumed, and what is the measured orphan rate?
02 Count the subtree
Observation
What the study says
At a fork, follow the child whose descendant tree contains the most work. An uncle that lost the tip still adds weight on the side that mined it.
Model
What has to hold
Work in the subtree is visible. A block that never propagates adds nothing.
Falsifier
What would retire it
Does the fork choice compare subtree weight or chain length?
03 The attacker must beat the forest
Observation
What the study says
Overtaking means outpacing the honest subtree, not a single path that got lucky with orphans.
Model
What has to hold
Work in the subtree is visible. A block that never propagates adds nothing.
Falsifier
What would retire it
Does the fork choice compare subtree weight or chain length?
04 Ethereum's uncles are not this protocol
Observation
What the study says
Paying uncles and mentioning them is related. GHOST is a fork-choice rule. A chain can pay uncles and still follow the longest chain.
Model
What has to hold
Confirmation depth is still chosen by the user. The fork choice does not pick it.
Falsifier
What would retire it
Does the fork choice compare subtree weight or chain length?
03 Sequence
One action, as an operating tape
- 01Miners publish blocks that may arrive to different nodes in different orders.
- 02A node builds the tree of all blocks it has seen.
- 03Starting at the genesis child, it chooses the branch with the heaviest subtree.
- 04That path is the chain it mines on and the chain it uses for confirmation counts.
- 05A block off that path can still add weight. It is not deleted from the security argument.
04 Load-bearing
The argument, and where a pitch drops it
Orphans were a security hole, not just wasted fees
The cut
The paper's claim is that at high rates the longest chain quietly changes the honesty assumption. GHOST is a repair of the fork choice, not a throughput feature you can advertise on its own.
Why it carries weight
If this cut is skipped, the paper's name is being used without the mechanism that makes the name mean anything.
Where it is dropped
The paper is not PHANTOM and not SPECTRE. Those are blockDAG protocols with different ordering rules.
Weight is not finality
The cut
A heavier subtree is still a probabilistic claim. It does not give the accountable checkpoint Casper later asked for.
Why it carries weight
If this cut is skipped, the paper's name is being used without the mechanism that makes the name mean anything.
Where it is dropped
Heavier subtrees do not by themselves give fast finality.
05 Register
What has to be true
Model · Not re-measured
Work in the subtree is visible. A block that never propagates adds nothing.
Model · Not re-measured
Miners build on the heaviest subtree they know. Strategic withholding is outside the basic claim.
Network · Not re-measured
The network delay is what creates collisions. The paper does not remove delay. It changes how collisions score.
Model · Not re-measured
Confirmation depth is still chosen by the user. The fork choice does not pick it.
06 Divergence
What happened after the paper
Ethereum cited GHOST in its early research and then implemented a simpler uncle-aware chain. PHANTOM and SPECTRE move from a tree to a DAG. If a team says GHOST, make them show the subtree comparison. A list of uncles is not enough.
A later client, parameter set, or reward formula is a different object. Cite this paper for the mechanism. Cite a primary release for the network. This desk has not re-run the proof.
07 Pre-mortem
What to check before you use the idea
- 0 of 4 marked on this browser. A mark is a reading note, not a pass, a rating, or a recommendation.
08 Anatomy
The paper, in the order a builder needs
The problem it names
If blocks are fast, two honest miners often publish at once. Bitcoin's longest-chain rule orphans one of them. An attacker who does not suffer the same collisions gains an edge. The paper is about that edge.
What the design proposes
- Choose the chain by the weight of the subtree, not by the number of blocks in a single path.
- Honest blocks that lost the race still contribute weight.
- The security claim is about the fraction of hash power, at rates where orphans are common.
How the mechanism is specified
- From the genesis, at each fork pick the child whose subtree contains the most work.
- An attacker must outpace that subtree, not a thin path of lucky blocks.
- Ethereum's later uncle rewards are a cousin of this idea. They are not an implementation of this fork choice.
What this page does not treat as proven
Why the desk still reads it
When a design shortens the block time and still cites 'the Bitcoin rule', ask what happens to honest blocks that arrive second. GHOST is one answer. It is not the only one.
09 Lexicon
Terms, opened into the record
- Subtree weight
- The total work in a block and everything that builds on it, including branches.
- Orphan
- A valid block that lost the race to be on the canonical path.
10 Repository
Every linked record on this page
Underlined words open a page that already exists: a concept, a protocol profile, a failure record, or another paper. If a word is not underlined, this desk does not have a record for it.
This is Blockchain Lab's reading of a public design paper. It is not the paper, not a copy of it, and not an offer of tokens, equity, custody or a partnership. Later network behaviour can diverge from the text. Nothing here is investment, legal or technical advice.
Research status: Design paper. Last reviewed: 1 October 2026. This is a reading of a public paper, not investment, legal or security advice.