Two Correct Systems. One Wrong Answer.

An LP elects a 1.50 percent management fee under a most-favored-nation provision, the rate a larger investor had negotiated and this LP was entitled to match. The election is recorded and sent to the administrator. Six months later the administrator's fee calculation uses 1.50 percent while an internal fee-reporting schedule still uses 1.75 percent. Both systems performed the calculation they were handed, and only one was handed the term then in force. The discrepancy reads like an accounting error, though what failed was propagation — the new rate reached one system and never arrived at the other.

Why both systems executed correctly

Neither calculation is wrong on its own terms. The administrator applied 1.50 percent because that is the rate it was given; the reporting schedule applied 1.75 because nothing had updated the input it reads from. Each is right about the rate it was given, and one of those rates happened to be out of date. This is how two careful people defend opposite figures in the same meeting, neither of them wrong: they are reading from systems that were told of the new rate at different times, if at all.

A calculation error is a mistake inside a system — a wrong formula, a mistyped number, a broken reference — and it surfaces when you check that system's own work. A propagation failure lives between systems: each one is internally consistent, every calculation checks out, and the fee is still wrong, because the operative term reached some systems and not others. Reviewing either calculation in isolation may not reveal it, since the failure sits in the handoff between them.

The tax on term propagation

Funds absorb an enormous amount of work catching these after the fact. The most demanding version has a name — shadow accounting, where a manager keeps a full second set of books alongside the administrator's so the two can be checked against each other. That comparison runs across capital calls, distributions, partner allocations, fee accruals, and the figures that feed investor statements, and it is among the highest-stakes routines a fund performs. For many managers, maintaining parallel records and continually squaring them against the administrator has become an ordinary operating control. Shadow accounting is broader than term propagation, but it exposes the same structural cost: discrepancies are discovered by comparing independently maintained records after both have already produced an answer.

Reconciliation of this kind never removes the condition that produced the break. It catches a divergence after it has happened, corrects it, and leaves untouched the reason the systems disagreed in the first place — no shared account of which term was operative when each of them last ran. So the next election, the next amendment, the next mid-life change to an LP's economics produces the next divergence, and the work repeats. Orivade is being built around that distinction, preserving the operative term, its provenance, and a record of which systems have adopted it and which have not, though the argument holds regardless of what closes the gap.

A source of truth is an architecture, not a policy

The instinct, once this is visible, is to declare one system authoritative and have the rest defer to it. As a policy it rarely holds, because the other systems keep their own live records — some of them, like the administrator's books and the underlying legal documents, must — and a policy does not reach in and update them. What is actually attainable is narrower and more honest than one record everything references. It is a single governed record of the operative term, tied by provenance to the document that established it, carrying its effective date and version history, propagated into downstream systems under some control, and reconciled where that propagation cannot be guaranteed. That last clause is the part most source-of-truth pitches quietly omit: some gaps close only by checking, because no fund controls every system its terms have to live in.

The hard part is rarely deciding which number wins once two reports disagree. It is establishing which term was operative, when it became operative, and which systems had adopted it at the moment they produced their answers. That problem sits upstream of reconciliation: it governs how a term becomes operational across a fund before competing answers have to be squared.