Two Keʻeaumoku blocks. Two discoveries. Two KILO reads.
Walmart Keʻeaumoku, where construction began in 2002, and The Park on Keʻeaumoku, permitted in 2022 — two parcels about 150 meters apart — both encountered iwi kūpuna, and both were reshaped, under public dispute, after the fact. Run KILO on the two parcels today and it returns a read of each. Here are both, exactly as the engine gives them — including what it could not confirm.
A hindcast, run honestly.
KILO did not exist when either project broke ground — but the parcels did, and so did the screening logic the engine encodes: ahupuaʻa cultural sensitivity, coastal-substrate risk, historic-property density, heiau viewsheds. KILO reads the underlying land, not the buildings on it now, so a run today reflects what a parcel-level pre-fieldwork screen would have surfaced before excavation.
A retrospective only counts if you commit to the result before you see it. This page publishes the scores the engine actually returns — and is candid where the engine's confidence is low.
Two parcels, rated two points apart.
Same ahupuaʻa, same coastal zone, the same four rules — and the engine landed the two parcels within two points of each other. For a retrospective that is the signal that matters: not one dramatic score, but a consistent read across two parcels that shared the same outcome.
Walmart had been screened. Every screen came back empty.
The Walmart parcel is the sharper of the two tests. When iwi kūpuna were found there during construction, it was not for lack of looking. In Hui Mālama I Nā Kūpuna O Hawaiʻi Nei v. Wal-Mart, the Hawaiʻi Intermediate Court of Appeals reviewed the project's permitting and recorded that the parcel had already been through multiple environmental and archaeological assessments — subsurface excavations, borings, and testing among them — none of which had flagged a significant burial or historic site. The court found there was "no factual basis to know or reasonably believe" the project "may affect" a burial site.
That is the conventional screen — point samples in the ground — certified clean by a court. KILO reads the same parcel at MODERATE, rising to HIGH for commercial earthwork, because it reads a different signal: a filled coastal marshland inside ahupuaʻa ʻo Waikīkī, on the substrate where inadvertent discoveries most commonly occur. The land carried what the boreholes missed. The Park is a different test — there, a permit-stage assessment did surface the first iwi kūpuna — which is why this retrospective leans on Walmart: it is the parcel where every screen came back empty.
How KILO reached these reads — step by step.
KILO is a reasoning model, not a black box. Every score traces, in order, from public facts about a parcel to a final verdict. This is the chain it ran for both Keʻeaumoku blocks — and each section below substantiates a step.
A 972-unit tower is not a sewer lateral.
KILO scores every plausible disturbance footprint. Both composites — 51 and 53 — sit at the light end of the range, near a sewer-lateral tie-in. Neither project was that. A big-box store and a twin-tower complex are full site grading, commercial earthwork — and on that row both parcels read HIGH.
Four rules fired on each block — the same four.
Every KILO score is traced. The same four rules carried both parcels; their reasoning is quoted from the assessments, with a note on how each one weighed in.
The Park's read is softer — and the assessment says exactly why.
The Park scored 53 to Walmart's 51 — but at 66% confidence against Walmart's 78%, and the engine names exactly why: USDA's soil service returned nothing for The Park's parcel, so the substrate could not be confirmed and the coastal-substrate rule ran on the coastal-influence proxy alone. That gap is not the land. It is a data gap.
Read the distinction carefully — it hit confidence, not score. The substrate rule still contributed its full weight; The Park still rated two points above Walmart. The engine did not under-rate the parcel; it under-knew it, and disclosed the shortfall in the open, in the same report. A retrospective that buried that would not be worth running.
The panels sized the scenario. Both blocks lived it.
KILO does not decide what SHPD will require — only §6E-42 review does. What it does is size the process a parcel at this risk band could trigger:
Those figures are not forecasts — they are the cost of skipping the pre-decision survey that would have pre-empted a discovery. Both blocks encountered iwi kūpuna once their projects were under way; both were redesigned amid public dispute. That is the inadvertent-discovery scenario, lived — twice — and the panels describe it years before anyone breaks ground.
Screening protects more than the schedule.
Run before acquisition — or before the grading permit — these reads would have put a cultural monitor and an archaeological inventory survey in front of the excavation, which is exactly what the coastal-substrate and regulatory-cascade rules indicate.
That sequencing protects iwi kūpuna from inadvertent disturbance, and it protects the project from the freeze. Pre-decision diligence is not a way around the cultural-resource process — it is a way to meet it on time, and with respect, instead of colliding with it mid-construction.
What this page is, and is not.
This is an advisory pre-fieldwork screening — not an Archaeological Inventory Survey, and not a substitute for SHPD review under HRS Chapter 6E. KILO models cultural-resource sensitivity from public empirical correlates — never as a presence claim at any specific point. It does not ingest, aggregate, or store iwi kūpuna locations from any source.
And one limit this page cannot escape: two parcels is not calibration. Two hindcasts that agree are a reason to run the real test — KILO across every documented Oʻahu discovery, against a matched control set of parcels with no discovery history — not a substitute for it.
Source — the Walmart parcel's prior-assessment history and the "may affect" finding are drawn from Hui Mālama I Nā Kūpuna O Hawaiʻi Nei v. Wal-Mart, No. 28477 (Haw. Ct. App. Dec. 16, 2009).
Get the read before the decision — not after.
Access is invite-only during beta. Tell us about a parcel or a class of deals you'd screen — the more concrete, the more useful the first conversation.