Special Edition 002: I Always Had the Maps. I Never Had the Record.

There is a split in due diligence that I lived with for so long I stopped noticing it. Half of the work was a structured lookup and the other half was a manual dig, and I treated that as the nature of the job rather than an accident of “which data happened to get organized first.”

The physical side got solved a long time ago. By the time I started running due diligence, the questions about the dirt already had answers I could pull in minutes. Parcels and zoning sat in GIS. Wetlands were in the federal mapper or discoverable in the field. Floodplains are on maps. Soils, topography, contours, easements of record, all of it queryable, layered, and standardized enough that two people looking at the same parcel would come back with the same picture. The site told me its story in structured data. I could be wrong about what the story meant, but I was rarely wrong about what the facts were, and I almost never spent real time finding them.

The municipalities never told their stories that way.

The governance record was the other half, the read, and it was the half that actually decided whether a deal lived or died. The plans, the ordinances, the conditions on the last three approvals, the thing a commissioner said in March that told you how the board was going to treat your product type in September. That record existed. It was public. And it had no index. Working it meant you could a) hunt meeting records, pulling minutes and video one file at a time, scrubbing hours of hearings for the handful of passages that touched my use, and chasing the governing ordinance and plan sections across a stack of separate PDFs that did not talk to each other, or b) go with what you already knew and figure out the rest along the way. Most of us chose b) and every deal started with incomplete political intelligence and due diligence.

What that asymmetry did to the practice is worth saying plainly, because it was never a personal failing and it was never laziness. When one half of the read is minutes and the other half is a week, you do not read the expensive half. You sample it. You pull the ordinance section that names the use, the two or three hearings someone flagged, the calls your own memory tells you to make, and you call that due diligence because under the cost you were facing, it was the only version of due diligence that fit inside a deal budget. The hours were never in the thinking. The thinking was the part I was good at and the part that took the least time. The hours were in the finding. The whole expense of reading a market was a sourcing problem wearing the costume of a knowledge problem.

What changed is narrow and it is specific. The governance record can now be structured the way the physical record always was. Once a jurisdiction’s meetings, ordinances, and plans are in a database instead of a folder of PDFs and a video archive, the long unstructured search collapses into a structured one. The week becomes an afternoon. Hours of looking become minutes of reading. I want to be careful about what that does and does not replace, because the easy version of this claim oversells it. It does not replace judgment. It moves my time onto the judgment. The senior attention that used to be spent locating the relevant ten passages now gets spent reading them and deciding what they mean, which is the only part that was ever worth a senior rate in the first place.

And because I am partly a skeptic myself, the part of this I trust least is the part that matters least. The objection that usually arrives here is that you cannot trust a machine to summarize a hearing, and that is fair. So… don’t. Pull the raw record and read it yourself. You still skipped the sourcing, and the sourcing was where the days went. The value is not in trusting a generated summary. It is in the disappearance of the dig. Trust the synthesis or don’t; the structured record removes the digging either way.

Here is the part worth slowing down on, because it is the part most people get backwards. The governance side lagged the physical side by years, and there is a real reason for it. GIS data was born standardized and geospatial. A coordinate is a coordinate in every county in the country. A code of ordinances is the opposite of that. Every jurisdiction writes it differently, organizes it differently, defines its terms differently, and stores it differently, and almost none of it is structured as fields. It is narrative text, written by lawyers, for a reader who already knows the local conventions. I can tell you how hard that is to fix because I have spent real time fixing it. Turning a thousand of those documents into something a person can query side by side is not a data-entry task. It is the actual bottleneck, and it was the bottleneck long before anyone was talking about AI. The model is the easy lever at the end. Standardizing the record is the hard work underneath it, and it is the work that was missing.

So the asymmetry I lived with is closing, and it is closing from the side that was always behind. For most of my career the physical record was legible and the governance record was not, and the entire shape of due diligence, the thin sample, the reliance on memory, the surprises that landed late, followed from that one imbalance. When both halves of the read are legible, the thin sample stops being prudent triage and starts being a choice. Not everyone will notice the moment it flips. The ones who do will read more of the record, earlier, for less, and they will price the deal and the risk more accurately than the ones still treating the governance half as something you sample because you have to.