The ADU legalization index — the instrument, the floor, and the two rows we can stand behind

Why an index, when there is already a study

The legalization study answers what the routes are and how they work. This page answers a different question: how would you compare one jurisdiction with another without cheating?

The obstacle is not access to jurisdictions. It is that jurisdictions describe the same thing in incompatible language.

One city publishes an ordinance chapter with a defined amnesty; another publishes a permit checklist that never uses the word; a third publishes nothing at all and applies the state statute directly at the counter.

Read those three side by side and it is not obvious which offers the homeowner more, or whether the differences are real.

So the instrument comes first, and the rows come second. What is published here is the coding schema: eight indicators, each with a written rule for what earns two points, one point and zero, and a fixed convention for what happens when a source cannot be retrieved.

That is the part that has to be right, and it is the part that can be checked before a single jurisdiction is coded.

The second reason is that the floor matters more than the index does. Most of what a homeowner with an unpermitted unit needs comes from state statute and arrives whether or not their city has a program.

An index that scored only local programs would imply the opposite — that a city with no amnesty offers nothing — which is false and is the most damaging false impression in this subject.

So the floor block is published first, in full, and the index is explicitly the layer above it.

How to read the floor if you have an unpermitted unit

Four of these levers are routinely absent from the advice homeowners receive, and each of them changes what is worth doing first.

The inspection-before-application provision is the one that changes behavior. Section 66311.7(d)(2) requires the agency to inform homeowners that, before submitting a permit application, they may obtain a confidential third-party code inspection from a licensed contractor to establish the unit’s condition and the likely scope of work.

The word doing the work there is “confidential”. The whole reason people leave a unit unpermitted is fear that asking the question starts an enforcement file, and the statute puts a route between the question and the file.

The fee position is stronger than most homeowners assume. Section 66311.7(e) provides that a homeowner applying to permit a pre-2020 unit shall not be required to pay impact fees or connection or capacity charges, with a narrow exception where utility infrastructure is required to comply with § 17920.3 and the charge is authorized by § 66311.5(e).

Impact fees and utility connection charges are the largest single line in most legalization quotes.

The enforcement-delay route exists independently of any amnesty and has a deadline. Government Code § 66331 requires the agency, on the owner’s request, to delay enforcement of a building standard for an ADU built before 1 January 2020 — or built later in a jurisdiction whose ADU ordinance was noncompliant at the time and is compliant now — subject to Health and Safety Code § 17980.12.

That section sets the delay at five years, requires the notice of violation itself to tell the owner the right exists, and runs only until 1 January 2030.

That sunset is the single most consequential date in this subject and it appears almost nowhere in consumer material.

And the main house cannot be held hostage. Health and Safety Code § 17980.12(a)(3) provides that a local agency shall not require, as a condition of ministerial approval of an ADU or JADU permit, the correction of a violation on the primary dwelling, provided correcting it is not necessary to protect health and safety.

Government Code § 66322(b) and § 66336 make the parallel point for the ADU and JADU permits themselves: a permit may not be denied because it corrects nonconforming zoning conditions, code violations or unpermitted structures that do not threaten health and safety.

HyreADU analysis. Read together, these are not eight separate concessions. They are a single design: remove the four things that make a homeowner rational to stay hidden — the enforcement trigger, the fee bill, the retrofit-the-whole-house demand, and the immediate correction order — while leaving the life-safety boundary intact.

Whether that design works is an empirical question this desk cannot answer from the sources it holds, because nobody publishes counts of legalized units.

What can be said is that the levers exist, that they are statewide, and that the 2030 sunset means the enforcement-delay half of the design has a shelf life.

The design decisions, and the ones we would argue about

Equal weights. All eight indicators carry the same weight, and that is a choice rather than a finding.

A reasonable person would argue that J2, the eligibility window, matters more to a homeowner than J1, whether the program has a name.

We have not weighted, for two reasons: with eight indicators and two coded rows there is no evidence base on which to justify a weighting, and an unweighted instrument is one anyone can reweight from the published cell values. The cells are published for exactly that reason.

A three-point scale, not a binary. Most of the interesting variation in this material is partial: a program that mentions fees without stating them, a statutory citation that exists but is superseded. A binary would collapse those to the same value as silence, which is wrong in both directions.

The index is a percentage of the coded maximum, not of the full maximum. Long Beach is coded on two indicators and scores 75 on those two; Oakland is coded on all eight and scores 81.

Those two numbers are not comparable and the page will not present them as a ranking. The coded-indicator count is printed in its own column precisely so that nobody reads the two as a table of winners.

What the instrument deliberately does not measure. Whether the program complies with state law — that is a finding for the Department of Housing and Community Development under Government Code § 66326, and a preemption conclusion belongs to a court.

Whether it is generous. Whether anyone has successfully used it. How long it takes. How much it costs in practice.

Every one of those would be a better thing to measure and none of them can be measured from published pages, which is what this instrument reads.

The failure mode we have designed against. An index of this kind naturally rewards jurisdictions with good web teams and punishes those whose material sits in a PDF behind a search box. That bias is real and it is not fully removable.

The n/c convention limits it; the coded-indicator column exposes it; and the explicit framing of the score as published transparency rather than program quality is the honest way to live with it.

The number nobody publishes

There is no published count of legalized ADUs in California, and this desk has not found one. HCD’s Annual Progress Report captures permits and completions; it does not carry a field that distinguishes a permit for a newly built unit from a permit legalizing a unit that has stood for fifteen years.

Municipal permit extracts record work descriptions, not amnesty status. So the obvious question — is any of this working — cannot be answered from the sources this desk holds.

That absence is why this page is an instrument and not an evaluation. An evaluation would require outcome data that does not exist in public form.

What can be built without it is a defensible way of comparing what jurisdictions say they offer, plus a complete register of what the state guarantees regardless.

Both of those are on this page and neither is dressed up as the thing that is missing.

HyreADU analysis. If the enforcement-delay route matters — and Health and Safety Code § 17980.12 sunsets on 1 January 2030 — then the absence of an outcome measure is a policy problem and not only a research inconvenience.

There is no way, on published data, to know whether the sunset will arrive with the route heavily used or barely used.

We note that as a gap, not as a criticism of any agency, and we would revise this page immediately if a count were published.