Zillow’s climate risk reversal looks like a setback. It’s really a wake‑up call.

When private models sow confusion, it’s a flashing warning sign that Washington needs to fix federal flood maps,

Zillow has dropped climate risk scores from its residential real estate listings following complaints that the scores, which use data from the private risk-modeling firm First Street, were of debatable accuracy and were hurting sales.

My big takeaway from this news: We urgently need a fully funded, modernized federal flood mapping program that is designed to show not just today’s risk, but also how flooding is expected to change in the future. Robust federal mapping should be the default national platform, with datasets open to all and well-documented, with methods and sources behind them clearly explained. That’s the framework that can provide an official, verified basis for where floodplain development rules apply, where flood insurance is required for mortgages, and what elevation and flood-proofing standards building codes have to meet.

We are very far from this ideal today, and so proprietary vendors are stepping into a yawning information gap—and potentially confusing homeowners and officials when their numbers conflict with official FEMA maps. Developers and real estate agents have their own agendas, so we have ended up with a messy tangle of assertions that can obscure genuine climate risks for many prospective homebuyers.

The way to move forward is to double down on making climate‑aware, nationally consistent mapping information public. Our flood mapping program should be treated as critical national infrastructure, as important as the core weather forecasting that the National Weather Service provides. Proprietary systems can and will be built on top of these maps—just as the NWS has given rise to a host of applications—but should not replace widely trusted, gold standard, public information.

People who work on flooding every day, a top-notch cross-section of our country’s mapping and hazard-analysis community, strongly suggest this step. They’re not saying that proprietary work and innovation should stop. Far from it. What they are saying is that authoritative, federally led mapping must anchor the system. Without this grounding, risk communication is ad hoc and inconsistent, land use decisions are purely political, and people are relying on anything they can grab when making decisions about major purchases. Public and private can coexist—but without a regulatory backbone and open standards for methods and data, we are on a tyrannically risky path at exactly the same time that flood damages and dangers are increasing.

The clearest recent document I’ve located about all of this was filed with FEMA this past summer by the Association of State Floodplain Managers. It’s an “Issue Brief” that urges Congress to fund this critical piece of infrastructure.

The basic problem has been widely reported: FEMA’s flood maps, used to figure out who has to have flood insurance in order to get a mortgage, are often hopelessly out of date and don’t reflect much of the risk. They leave many places that can still flood badly “outside” official flood zones, do not show expected future flood levels, and don’t account for how heavier rains and higher seas will change flooding, except in a handful of special cases.

The flood managers know this. They’re dealing with these issues all the time. They know that FEMA officials have been pushing since 2019 for a “Future of Flood Risk Data” (FFRD) initiative that would fix these problems. A nationwide cross-section of experts on flooding and engineering called the Technical Mapping Advisory Committee (TMAC) was asked by Congress back in 2012 to advise FEMA about mapping future conditions, and has issued detailed reports providing step-by-step guidance. Its 2023 Annual Report pushes FEMA to produce probabilistic risk data for structures in the US. TMAC says FEMA should stop drawing lines between “floods” and “doesn’t flood” and instead provide watershed-wide data about the probability of flooding. In TMAC’s view, FEMA should use open, authoritative data inputs (like USGS river flow data, USGS topography data, and modern NOAA rainfall frequency data), and make the computer models it builds based on these data sources available to the public in a federal model library.

But Congress has not been willing to pay for this project, which as of 2020 was estimated to cost somewhere between $3 and $12 billion. So we don’t yet have public, modern, model-based maps that include crucial missing elements of risk. In the scheme of things, fixing this problem wouldn’t be a huge investment: Annual flood losses in the US now average over $40 billion.

In the absence of adequate federal investment over the last few years, proprietary flood analysis products have ...flooded the market. Users can’t inspect their models most of the time—not that users would necessarily know how. The models push out predictions that are sometimes “free” (as in the First Street scores) but wholly proprietary, and not designed for or certified for regulatory/statutory implementation.

The floodplain managers say that these proprietary models are “extremely valuable for raising national risk awareness, informing high-level planning, and screening for potential areas of concern.” They’re not trying to squelch them. The professionals are saying, though, that these proprietary efforts should be—must be—complementary to a well-funded, open federal system. Otherwise, confusion and anger reign—as we are seeing this week.

As a structural matter, a modern federal flood prediction data infrastructure would be similar to the way we treat earthquake data—the USGS has a “National Seismic Hazard Model” that is nationwide, publicly funded, probabilistic, and incorporates a host of data sources. But we could improve on the earthquake model by making greater investments in reducing the uncertainty of federal flood predictions (better topographical, water, rainfall, climate projections) and clearly communicating probabilities and scenarios. This just doesn’t happen across the federal government. Yet.

We need probabilistic, future-conditions, multi-peril mapping data that is public and subject to norms of oversight and transparency. It is not the job of private firms to do this. Their job is to maximize profits for their shareholders. Last summer, Axios reported that First Street, which began as a nonprofit, “transitioned into a public benefit corporation in February” 2024 and then raised a total of $46 million in Series A funding, marking its shift into a private, venture‑backed company. They have done us a real favor by illuminating blind spots in existing FEMA maps—these blind spots are so large they have blind spots of their own—but they are not a substitute for a strong national mapping program.

In addition to providing services to Zillow, First Street has embedded its products across the federal government. Other vendors are trying to do the same thing. Instead of relying on rented climate risk data from private vendors, Washington should build—and openly share—the gold‑standard data the country actually needs. In the long run, outsourcing our collective understanding of risk will cost all of us far more than investing in clearly communicated, trusted, public information.