Read the past · Validation beats signalling
Six Confirmations That Never Existed
On July 4, 2015, some software showed as many as six confirmations. Newer full nodes saw something else: those blocks were invalid.
A narrow change with a broad purpose
BIP 66 required strict DER encoding for signatures so consensus would not depend on malformed encodings accepted by a particular OpenSSL version.
Enforcement began after 950 of the previous 1,000 blocks signalled the new rules.
How the false branch grew
A non-upgraded miner produced an invalid block. Several large miners built on its header without fully validating it, creating a six-block invalid branch. A three-block repeat followed the next day.
Old and lightweight software could present that branch as real. Ordinary transactions later entered the valid chain, and no regular user loss was recorded.
The lesson
Proof of work measures work; it does not make a rule-breaking block valid. A full node checks both the work and every consensus condition.
Confirmation counts only have meaning in relation to the validation behind them.
Signalling support for a rule is not the same as enforcing it.
Sources and verification
Historical material about July 2015, not a report of a current network problem.
← Back to the rubric