the ledger that stopped saying accepted

One entry in this cluster is a subtraction. The status marks were removed from the decision headers and replaced with timestamps. A record of decisions had been labelling each one with where it stood; afterwards it records only what was decided and when.

The reasoning is not written down as a principle, but the shape of the change is legible: a column called accepted implies an authority that accepts, and an authority that accepts is one that can also refuse, delay, or grant. Decisions in this design are numbered, dated, and superseded by later decisions. Nothing is blessed. Nothing is struck out.

The tension is real, because readers want to know what is binding. A numbered ledger is useful only if current can be told from historical, and the cheapest way to give a reader that is a status column — which imports a judge the design otherwise does not have. Marking a decision accepted also freezes an argument at the moment its authors were most confident, which is precisely the moment it is least likely to be revisited honestly.

So currency has to be established by reading forward rather than by looking up. Later decisions override earlier ones, reversals stay in the record with their reasons attached, and a reader who wants the current state reconstructs it from the sequence.

What it costs is work for every reader, and a ledger that cannot point at one row and say this is the rule. What it proves is that the team preferred an auditable argument to an authoritative list, in the one artifact where the temptation to be authoritative is strongest.