TAL, PLO, and the Hidden Cost of Accumulated Micro-Adjustments
Batch reference:
Colombia / Sidra / 111
Research status: mechanism confirmed by direct observation. Single confirmed case.
Many roasters trust a smooth curve.
No crash.
No flick.
No dramatic correction.
No obvious mistake.
But some failures do not announce themselves in the graph first.
Colombia / Sidra / 111 is one of those cases. The operator made small, careful, intentional adjustments. Nothing looked reckless. Yet the final drop landed above the intended observation window, and the cup showed the cost: compressed structure, reduced differentiation across temperature stages, and intensity without arc.
This is what SUNNY M Lab describes as PLO, Probe Lagging Overcorrection.
The point is not that the operator made one large mistake. The point is that no single adjustment looked large enough to be the cause.
In professional digitally logged sample-roasting equipment, bean temperature is read by a probe. The probe does not display the full thermal effect the instant an adjustment occurs. The bean has already begun absorbing energy, but the probe has not fully reflected that change. The reading is therefore not purely real-time. It contains delay.
SUNNY M Lab describes this delay as TAL, Thermal Absorption Lag.
TAL is not an equipment defect in this context. It is the physical behavior of heat moving through a dense object. Bean surface, bean core, and probe reading do not become the same thing at the same instant.
The real risk appears when the operator reads “delay” as “no response.”
One small adjustment is made.
The probe has not shown the effect yet.
Another small adjustment is made.
Then another.
Each decision may look reasonable on its own. But the cup does not experience these decisions separately. Their effects may accumulate before the probe displays the combined result.
This is PLO.
The probe lags.
The operator corrects.
The correction becomes an overcorrection.
Not because one dramatic error occurred, but because multiple careful-looking decisions arrived together in the later part of the roast.
This case matters because it challenges a common assumption in curve-based roasting: if the visible curve is smooth, the roast process is stable.
The curve may look controlled, while delayed thermal effects continue to accumulate. The log may look calm, while the cup has already been shaped by decisions that had not yet fully appeared.
The cup from this batch was not collapsed. It was complete but pushed beyond the intended window. The sensory signature was compressed structure, reduced temperature-stage differentiation, and intensity without arc. It was not undrinkable. It was simply not the cup that this window was designed to produce.
The purpose of this note is not to provide a correction method.
It is to show that multiple small decisions can complete their accumulation before the probe displays their combined effect.
This is why SUNNY M Lab places failure records inside the research archive. Failure is not only error. It can become a boundary. It can reveal where a model is too simplified. It can explain why the curve record looks acceptable, but the cup disagrees.
The graph did not show the problem.
The cup did.