When a meter goes offline, what number do you show?

A three-phase meter on a factory reports every few minutes: cumulative import and export registers, per-phase voltage, current, power factor and power. Then the connection drops — a modem reboots, a site loses signal, power blips — and for three hours there is nothing. When the meter reconnects, its cumulative register has moved on. Energy was used. The meter knows how much in total; it does not know, and cannot tell you, the minute-by-minute shape of those three hours.

So you have a total you can trust and a timeline you cannot. The question is what to put on the chart between the last good reading and the first new one. It sounds trivial. It is actually the whole philosophy of the platform in miniature.

The first tempting answer is to fill the gap with zero. It looks clean — a flat line along the bottom — but it is false twice over. It implies the site drew no power for three hours, which almost certainly did not happen, and it throws away energy the register clearly recorded. Now your daily total is short, and it is short in a way nobody can see.

The second answer is to interpolate: draw a smooth line from the last reading to the next and spread the recovered energy evenly across it. This is worse, because it is convincing. It manufactures a plausible-looking curve for a period in which you have no measurements at all. Anyone downstream — an operator, a CFO, a billing engine — reads that smooth line as data, when it is really a guess wearing the costume of data.

The third answer is to silently drop the period, so the total for the day simply excludes it. That keeps every visible point honest but makes the day’s total quietly wrong, and hides the fact that anything was missing. A number that is wrong without warning is more dangerous than a number that is wrong out loud.

WattEY separates the two facts and treats them differently. The total is recovered from the meter’s cumulative register on reconnect, so the day’s energy stays correct — no unit is lost, none is invented. The timing, which is genuinely unknown, is shown as exactly that: an "offline · times unknown" band across the gap, visibly different from measured data.

The accounting identity we hold to is simple: measured energy plus offline energy equals gross energy. The measured part is what the meter actually reported, minute by minute. The offline part is the recovered total for the gap, carried without pretending to know its shape. Add them and you get a gross figure that ties back to the register. Nothing is smoothed into existence.

The effect is that a reader can always tell what the platform saw from what it inferred. The chart stops being a single confident line and becomes an honest one: here is what we measured, here is what we know happened but cannot place in time.

Businesses bill and make decisions from these numbers. If the platform quietly fills gaps, every derived figure — cost by source, peak versus off-peak, a tenant’s share of a shared bill — inherits an invisible fabrication. The error does not announce itself; it just sits in the total, and it surfaces the day a customer checks the platform against their own meter and the two disagree.

Holding to measured-plus-offline-equals-gross is what let WattEY reconcile a disputed dairy-farm bill to 0.015% against the meter’s own lifetime register. That number is only possible because the platform never invented energy it did not measure and never discarded energy it did. Trust is not a feature you add later; it is the sum of hundreds of small refusals to fake a line.