Scenarios and the split matrix

Read Activation and fork paths first, since this article uses the state machine and the five heights throughout.

The axes that govern the outcome

The outcome is determined by at least four independent axes. Keeping them separate is what makes the matrix tractable.

1. Threshold timing: whether the header threshold (1,109 of 2,016) is reached in a voluntary period before the mandatory window, or not.
2. Parent choice during divergence: whether miners build on the BIP-valid parent when a divergence appears, not merely whether they set bit 4.
3. Actual enforcing hash power: how much hash power actually validates and mines RDTS-compliant blocks, as opposed to only signalling.
4. Economic acceptance: which history economic actors treat as settlement, and how that reallocates future hash power.

The first three are protocol-level and are the subject of this article. The fourth is economic and can change incentives, but it cannot change the validation relationships: an exchange choosing a ticker cannot make an invalid block valid.

Figure slot BIP 110 Branching Timeline figure pending

Two starting paths

Either the voluntary threshold is reached early, or the branch enters the mandatory window still STARTED.

Path A: the voluntary threshold is not met early

If neither an early period reaches 1,109 qualifying signals, the deployment is still STARTED as the candidate block at height 961,632 is considered, and the mandatory-signalling predicate applies. For blocks 961,632 through 963,647, every accepted BIP-branch header must signal. This is where a split can first appear, and it appears in the header, before any block body is downloaded.

Path B: the voluntary threshold is met early

If a voluntary period reaches 1,109, the mandatory interval is skipped, not merely eased. Non-signalling headers stay valid on that early path. The decisive test then moves to activation, at height 965,664 on that branch, when the seven transaction rules become consensus rules and the first incompatible block (if any) would expose whether signalling reflected real enforcement.

The threshold is a coordination mechanism, not a safety oracle. It certifies that at least 1,109 qualifying headers were mined in one historical period. It does not authenticate software, template sources, pool policy, or economic acceptance.

The mandatory-window outcomes (Path A)

Once the branch is in the mandatory window still STARTED, four mechanical outcomes are possible, depending on parent choice and hash power.

Outcome 2A: no aligned miner extends the BIP-valid tip

If the dominant network mines a non-signalling block at 961,632 and no miner builds a signalling alternative on top of block 961,631, then:

• BIP nodes reject the new header and remain at 961,631.
• Unmodified nodes accept the new block and continue.
• There is consensus divergence but not two advancing assets: one growing permissive chain, and a stalled population of BIP nodes.
• A later bit-4 block built on top of the rejected block does not repair anything. Its parent is still invalid to BIP nodes.

The distinction between "sets bit 4" and "mines on the BIP-valid parent" is the essential one. Observed signalling rate is only a rough proxy for hash power capable of sustaining a separate BIP branch.

Outcome 2B: a minority mines an all-signalling alternative

If some hash power mines a signalling block on block 961,631 and keeps using that ancestry, two live histories exist from the last common block. Unmodified nodes regard both as valid and normally follow whichever has more cumulative proof of work. BIP nodes cannot select the permissive branch, whatever its work, because it contains an invalid ancestor. If the BIP branch is minority hash power, it advances slowly at the inherited difficulty. If it later overtakes in cumulative work, unmodified nodes can reorganize to it; the reverse reorg is impossible for a BIP node.

Outcome 2C: signalling dominates during the window

Miners can collectively begin setting bit 4 at 961,632 even without an earlier threshold. If the most-work chain contains only signalling blocks throughout the interval, both node populations can remain on one history. Every accepted BIP-branch block in the interval signals, so a complete mandatory period necessarily exceeds 1,109 signals, and at 963,648 the branch reaches LOCKED_IN. This proves only that acceptable headers were constructed on that ancestry. The transaction rules are not active during this phase, so a miner can cross the mandatory window without running the validation that would reject an RDTS-invalid transaction later.

Outcome 2D: the BIP-valid branch has majority work

Non-signalling blocks can still be produced, but if the all-signalling branch consistently has more cumulative work, unmodified nodes reorganize onto it and the non-signalling blocks become stale. The split can therefore be brief, or visible only to monitoring infrastructure rather than durable. Around a 50/50 work split, unmodified nodes can reorganize as the lead changes, while BIP nodes behave asymmetrically: they move among BIP-valid candidates but never onto the branch with the invalid mandatory-window ancestor.

Figure slot BIP 110 Activation and Fork Paths figure pending

The activation outcomes (both paths)

Whether the branch reached ACTIVE by the early threshold or through the mandatory window, activation itself does not force a split. At height 965,664 the seven rules become consensus rules, but a visible split needs an actually incompatible block. The possibilities:

• Every mined block is RDTS-compliant. Both software populations can stay on one chain for the entire active interval. The rule difference remains latent.
• A violating block appears, enforcing work dominates. BIP nodes reject it. Unmodified nodes may accept it briefly but normally reorganize to the higher-work BIP-valid branch, and the bad block becomes stale.
• A violating block appears, permissive work dominates. Enforcing miners must build a competing valid block on the last common valid parent. A persistent two-chain split results if both sides continue.
• A violating block appears, but no enforcing miner builds an alternative. BIP nodes stall at the last valid block rather than inventing a chain by themselves.

An active-period violation normally requires the block body and the UTXO or script checks to detect, so its header can look valid. This is unlike the mandatory-window failure, which is visible directly in the header.

The full permutation matrix

The table below is the compact form. Read each row as "if the signalling-and-enforcement condition holds, and the trigger occurs, then the likely protocol mechanics are as stated". These are mechanical consequences, not probabilities.

The economic permutations

Economic behaviour can change incentives but not the validation relationships. Three cases are worth keeping distinct, and none of them is a prediction:

• If economic actors prefer a lower-work BIP branch, an unmodified node still selects the higher-work permissive branch. Following the lower-work branch requires enforcing the rules, a deliberate local invalidation or checkpoint policy, or a future work crossover.
• If economic actors oppose BIP-110 but the BIP-valid branch has dominant work, an unmodified node still follows that branch, because it is valid and has more work. Protocol-level opposition requires alternative rules plus sustained mining of an incompatible history.
• If markets value both histories, prices and fees can continually reallocate hash power. The initial split percentage is not necessarily the long-term equilibrium.

The null result is a valid outcome

One outcome deserves naming, because it is easy to overlook: no split at all. If every mined block through the window and the active interval is RDTS-compliant, the rule difference never surfaces on-chain, and one shared chain persists. This is the "anatomy of a fork that did not happen" case. It is not a failure of the event to be interesting; it is one of the mechanically valid results, and the site treats it as such.

What to watch

On the Live dashboard, the scenarios above become observable in this order:

• Signalling by period: decides Path A versus Path B. Bit 4 is a header claim, not enforcement, so a high count still does not settle the enforcement question.
• Chain tips: the first place a mandatory-window or activation split becomes visible, as a second tip nodes disagree about.
• Split telemetry: appears only if a split actually materialises, capturing the rejected block or the competing branch. Absence of this panel is the null-result case, not a verdict.

Sources

This page restates the BIP-110 technical walkthrough sections Scenario 1: the voluntary threshold is not met (all four mandatory-window outcomes and the active-period outcomes), Scenario 2: the voluntary threshold is met, and The full permutation matrix (including the economic permutations). All statements are conditional mechanics. The walkthrough's own evidence-based base case is discussed, as a base case, in the monitoring article, not asserted here as a prediction.