Vol. 022: Where the Record Runs Out

Current Conditions

Three issues in this run have argued that the record deciding a deal is now cheap to read (Vol. 19, Vol. 20, Vol. 21). This issue tests that argument against a case where reading the record correctly and reading it completely turned out to be two different things.

On July 10, 2026, this firm ran a blind prediction: five written sections on 24CZ16, the Lawrence Property PUD in Apex, North Carolina, built entirely from records dated before the case’s real August 2024 submittal. Nothing about the actual seven-resubmittal history, the real hearing dates, or the outcome informed it. It was the second in a running series of dated, scored predictions, following an earlier case in Knightdale. The prediction sat untouched until the real record could be pulled and graded against it: Planning Board on May 12, 2025, Town Council on June 24, 2025.

Interestingly, the system flagged a historic farmhouse relocation, made to accommodate a collector street shift, as a red flag. From personal experience, we felt this commitment would likely not hurt the application and, if anything, might garner some goodwill. For this exercise, though, we wanted to test how a fully automated review would fare.

Signals Observed

The structural calls held up well. The predicted outcome, approval by a comfortable majority, landed and the actual result beat it: unanimous at both bodies. The primary precedent case, 23-CZ-22, and the density and buffer precedent, Castleberry Reserve, both matched the record closely enough that a June 24 council member’s own words tracked the prediction’s reasoning almost verbatim. Four of five predicted opposition themes, school capacity, stormwater, traffic, and road-funding burden, showed up in testimony essentially as written.

The farmhouse call missed, and it missed in the specific direction flagged in advance. At the June 24 hearing, a council member cited the historic relocation plan approvingly, saying he was “really excited about that part,” and the Transportation Plan Amendment tied to it passed unanimously with no debate. What the record-only prediction scored as the biggest exposure in the case turned out to be a selling point.

The other misses in the exercise cluster in the same category as the farmhouse: predicting how one specific person would weigh one specific issue. The prediction correctly identified which council member’s density and preservation lens would be most active in the case, then guessed the wrong direction, assuming that lens would produce a dissent when the plan was in fact built low density enough to satisfy it. The record showed which lens mattered. It did not show, on its own, which way that lens would point once someone actually looked at the plan.

Opposition volume again had no bearing on the vote. One public speaker’s testimony was openly adversarial, calling the project “a disgrace” and “a money game,” and it moved nothing, consistent with what this newsletter measured across 200 municipalities in Vol. 9 and on one Georgia parcel in Vol. 21.

Forecast

The structural layer of this kind of prediction, outcome, precedent, conditions, the recurring opposition themes, is likely to keep getting more reliable as the underlying record gets easier to query completely rather than sample thin. That is the part of the reading problem this run has been describing since Vol. 19, and it is closing fast for anyone willing to read the whole record instead of a version filtered through a phone call.

What is not closing, on the evidence of one case, is the layer where a specific person’s judgment turns an ambiguous fact into an asset or a liability. The record could show that a preservation-minded council member’s lens would be active in this case. It could not show, by itself, that the same lens would flip in this application’s favor once someone who has sat through enough of these hearings recognized which way the plan would read to a receptive board. That is not a data gap more meeting transcripts fix. It is a different kind of knowledge, the kind built from watching enough of these arguments land, or fail to land, in the room.

Watch Items

• Whether future blind-prediction exercises on other case types, commercial rezonings, variances, annexations, show the same hit-and-miss split, or whether the farmhouse result is specific to preservation issues
• Whether council-member voting histories, tracked at finer grain, eventually narrow the “which direction does this lens point on this plan” gap, or whether that judgment stays a human read
• Whether other firms running record-based tools start logging where a practitioner overrode the output, since that log is itself a dataset about where judgment adds value
• Turnover on the boards being modeled, since a member’s lens is only a reliable signal as long as that member holds the seat and reasons consistently case to case

Confidence Level

Medium. This is the second scored case in the series, not yet a study. The structural hit rate is a genuinely useful data point across both cases. The single miss here is suggestive, not proof, of where the boundary between record-reading and judgment sits generally.

An aside. The farmhouse is still standing, on the relocated site. The disagreement that almost got edited out of the report is the only part of this exercise that did not come from a database.

Why This Matters

The value of reading the full record, rather than a version of it filtered through relationships and a handful of calls, is not in question at this point in the run. Vol. 20 priced it. Vol. 21 showed it winning an actual case. What this issue adds is a boundary line. A complete record can narrow a deal to the small number of places where the outcome is genuinely uncertain. It took a less common kind of local knowledge to correctly weigh the one place that stayed uncertain here. Before a deal is signed or an application is filed, that split, what the record settles and what it still leaves for a person who has watched enough of these hearings to have an ear for the room, is worth knowing which side of it you are standing on.