The Problem with Prop 1

On California’s enormous mental health ballot measure, or, what it looks like to try and steer a ship with no rudder

If you live in California, you may have heard about the state proposition that is on the ballot for Tuesday March 5—in fact it’s the only proposition on the ballot at the state level (thank God, because God knows there are enough other things on the ballot to figure out). 

Prop 1’s ballot text reads:

Ballot title: Authorizes $6.38 billion in bonds to build mental health treatment facilities for those with mental health and substance use challenges; provides housing for the homeless. Legislative statute.

Ballot summary: Amends Mental Health Services Act to provide additional behavioral health services.

Fiscal impact: Shift roughly $140 million annually of existing tax revenue for mental health, drug, and alcohol treatment from counties to the state. Increased state bond repayment costs of $310 million annually for 30 years.

Sounds pretty straightforward, right? Raise some money, house homeless people, bam! Housing crisis solved.

My guess is that most people reading the text of this proposition would see it as overwhelmingly positive (more money to solve a big problem? Of course!) and vote yes. I expect this proposition to pass, regardless of what I say in this post (polling currently seems to agree with me).

(Psych Crisis definitely doesn’t do lobbying or attempt to influence political outcomes; my interest in this post is not to convince you to vote a particular way but rather to point out the system dynamics surrounding something like this proposition and how those dynamics keep the same system-wide problems in place. So this piece is less ‘pro’ or ‘anti’ and more like ‘wow isn’t this system inadequately structured from the perspective of solving the problems it is tasked with solving? What could we do about that?’)

So there are (at least) two issues with the naive reading of this proposition:

1. it’s literally incorrect. Unfortunately, the text on the ballot leaves out key things the proposition will do if passed.

I only know this because of many hours talking to people who have poured over the full legislation the ballot prop is approving, and followed the saga of it getting to the ballot prop stage, and heard hours of debates between people with better understandings of the proposition than 99.9% of the population and I’m reaching the point where maaaaybe I am starting to understand what the ballot is literally supposed to do if passed. 

It is supposed to (to the best of my knowledge):

• Use an existing revenue stream (a 1% tax on millionaire income that funds mental health services, to the tune of $2-3.5 billion a year) to get a large bond (read: debt) to pay for building housing and institutional facilities for people who are homeless and have mental health and substance abuse problems. These funds will have quite specific requirements; like that 50% of the housing that is built needs to be allocated for veterans (who make up roughly 7% of the homeless population in California). 

• Divert $1 billion annually from a bunch of existing mental health programs (which will probably close down as a result) into these new funding priorities.

So one of the immediate things you notice is that although the prop bills itself as only increasing funding, it does decrease funding in areas that aren’t mentioned on the ballot in order to do so. 

The second issue is that the implicit promise of this proposition is that it will reduce homelessness.

Maybe even more specifically, more subliminally, it gives one the sense that after passing this measure, there would be fewer people living on the street of the kind who are loud, dirty, visibly addicted to drugs, or yelling to themselves (the ‘visibly mentally ill’ ones).

One can imagine an optimistic voter who walks through the Tenderloin in San Francisco every morning on their way to work in an office downtown seeing this proposition and imagining those streets absent of anyone disheveled or making any sort of fuss. Encounters with people who are disheveled and unpredictable can be stressful, so a ballot measure that promises they’ll be ‘housed and in treatment’ (whatever that means) sounds mighty appealing. 

This begs two obvious questions:

1. Can this proposition create that glorious, stress-free and tidy future that it promises?

2. Given that the promise I described is implicit, and stated nowhere in writing in the ballot, and the ballot will ostensibly affect (read: reduce) funding for all sorts of mental health programs, including ones that have nothing to do with helping homeless people, how might this ballot proposition affect how the system works for all those other people?

To answer those questions, one needs to dive into the labyrinthine and incoherent world of causality within government services. 

See, this is at the end of the day, a piece of legislation. Drafted by policymakers (likely, the staff of the California governor or the politician who formally introduced the bill into the legislature), the ballot measure is simply a request to the voting public to approve the long and complicated bill (actually two bills) that the elected politicians have already agreed upon and voted to pass. Aside from the part where California has to take on debt to do this (which they have to by law get approved by voters) this wouldn’t have even hit a ballot paper at all. 

As a piece of legislation, the measure is fundamentally a set of instructions for the administrative side of the government, that all the politicians agreed upon. ‘The department of X should distribute this money using process Y, while adhering to all the other rules we’ve ever previously passed except while making changes to bills Z, W, U and V.’ You can think of the legislation as code, run on the administrative part of the government as the computer.

Except in programming when the code is compiled in order to be run, the compiler checks whether it is grammatically correct and will throw errors if any of it literally violates the rules of how the programming language works. Similarly, a good programmer will develop some more code called a test suite, which is a set of tests that check whether the code does what it is intended to do and doesn’t do what it isn’t intended to do (i.e. have unforeseen side effects). 

Nothing like compiler grammar checks or test suites exist with legislation, so when politicians ‘push to production’ (pass a bill) there is essentially no way of knowing if the new rules are even possible or feasible to follow, whether they can even in principle achieve the intended outcome, and whether they are likely to create unintended or unwanted consequences. 

(In a lot of cases the legislators will check a bit with the administrators to make sure they aren’t making any glaringly stupid mistakes, but given that in many cases the administrators will take the legislation and then use it to produce even more detailed rules for other entities to follow, and those entities almost never have a say in the bill-drafting process unless they’ve very rich, there are a lot of opportunities for errors to creep in and very few opportunities to correct them before the bill is made into law.)

I’ve now had a few opportunities to closely read the text of a piece of current California legislation (on mental health) and compare it to my observations of how the thing referred to in the legislation is operating in practice.

One thing I will say is: there is tonnes of wording in legislation that ends up watered down in the implementation, implemented in a lip-service way, or straight-up ignored. How bad the deviation is, from my limited observation, seems to mostly be determined by a combination of how strong the consequences are for not following the letter of the law exactly, how much the entities implementing or following the law share the values the law was made to promote, and how much time and energy the entities involved have for actually keeping track of all the relevant changes to the legislation that affects them. 

In the case of laws that govern the behaviour of administrative bodies like state health departments and other issue-based departments, there really don’t seem to be that many actors whose job it is to closely track whether the legislation is being followed and then enforce consequences for noncompliance. And there certainly isn’t anyone whose job it is to make sure those departments are doing anything other than the bare minimum.

This isn’t to say that all such departments are slackers; but some certainly are, and there’s no ‘administrators doing their job’ police except for the legislators themselves and the legislators don’t always have the firmest grasp of what policies they signed into law anyways. (They do process an awful lot of it, on an absurd number of topics. Imagine needing to make expert decisions on matters around engineering, medicine, education, urban planning, finance, and ecological management all in the same day! The amount of things politicians and their teams have to keep track of to do their jobs is frankly absurd.)

---

So one of the questions we need to ask when asking ‘can this proposition succeed at its stated aim?’ is ‘is this legislation robust to being badly misinterpreted and implemented in parts very shallowly or with less-important pieces missing; is it likely to still succeed even if the administrators and the contractors and everyone else all the way down just fudged things a little bit, or interprets things overly rigidly or overly flexibly?’

One of the issues with making a decision like ‘let’s spend three billion dollars’ is you cannot possibly specify all the relevant tasks ahead of time. And even if you managed to, you would be leaving out relevant context that you could only get in the process of implementing it (that is probably known by only like one guy who is a specialist in one particular thing like three layers of management down).

So you don’t fully specify it; you leave the text broad enough that others can interpret it as they go.

The problem with this process is that from the perspective of the goal ‘end homelessness’ (which is ostensibly the measure’s true goal), the whole thing ends up like a game of Chinese whispers. Every time the legislation goes through a new layer of administrators it gets interpreted again, and in each case the administrators are trying their best to follow what their superiors said, not what they discern is best for accomplishing the goal, given everything they know from their vantage point.

Nobody actually wants this. Legislation gets passed and repeatedly interpreted and interpreted until the actual implemented policy bears little to no relation to a one-sentence description of the original topline goal the policy was designed to achieve. The bigger the legislation and the more departments and entities it affects, the more chance for such incentive-driven wobbles to get introduced. 

(For a great read on this phenomenon at the federal level, check out this piece by the Niskanen Institute.)

This is a very big piece of legislation. It will affect at minimum hundreds of private organisations, for-profit and nonprofit, a bunch of state agencies, and every single county in California (that’s 58 separate county governments). Each of those entities is a chance for the bill to get interpreted in a way that does or does not serve the original goal, without the person making the decision on behalf of the entity generally having any way of knowing whether they’re helping or hindering, or those who designed the law originally knowing how easy it was to follow or whether it led to the desired results. 

One of the reasons Prop 1 is being introduced and moving money around in the background is that 20 or so years ago a big bill was introduced that taxed people with millionaire-income a little bit and created a giant fund to give to counties to fix their mental health systems, and it has been 20 years and nobody at the top really has any idea of how well it has been working and so one has to assume it hasn’t been working particularly well, or at all.

It already funds around $3 billion worth of stuff a year, so if it were gonna massively lower the suicide rate or homelessness rate or addiction and overdose rate you would have expected it to happen already and the results to be pretty clear. Instead, it seems like there are mostly a bunch of new programs that maybe sorta work well for some particular group of people who happen to be involved in the programs currently but who can tell, because the counties can’t tell the legislators how well it’s working anyways?

So this measure seems, in some way, to be trying to fix the sins of the last attempt at ‘fixing the system’, this time targeting its focus on the most visible problem voters associate with the brokenness of the mental health system—people who are disturbingly disordered in the public spaces of cities, the people who are seen as ‘the mentally ill homeless’. 

It seems like this new iteration has about as much chance of ‘fixing the system’ as last time, because the functional mechanisms for how policy is implemented and legislators try to steer the system for change haven’t changed. If anything, in the last 20 years, it has grown more unwieldy, with more opportunity for errors and more entities able to misinterpret things.

---

But if misinterpretation were all we needed to worry about we’d have an easier problem on our hands.

The other, sort-of-obvious but never asked question is ‘will simply throwing more money at the problem actually help? Are the strategies we have at our disposal already for addressing these issues sufficient for solving them, if given more funding?’

I suspect if you talk to mental health professionals in the field they would disagree on this one, but my opinion here is clearly NO.

Just looking at the dropout rate from various publicly-funded mental health programs, or the relapse rates after various kinds of treatments; even if you look at it in the government’s language of relapse prevention and preventing hospitalisation, most of the default stuff we have doesn’t work, or works for some chunk of the population but definitely doesn’t work for some other chunk. 

It isn’t the case that 95% or so of people who get housing through housing programs stay housed for the rest of their lives, or that 95% of the people who go through rehab stay sober for the rest of their lives.

There are plenty of people who are rowdy and high in the Tenderloin who have places to live and come there to hang out and get drugs (I met one such guy on -I think- meth the other day who lives in the suburb of Bayview) so if you say you’re trying to solve homelessness but what you’re really trying to solve is an uncomfortable civic atmosphere in the downtown area when everyone’s doing drugs, housing isn’t going to fix that, and isn’t going to solve this perennial political issue where voters expect you to do something meaningful about the ‘homelessness problem’. 

(Leaving that aside, there is totally a housing shortage in California of course.)

I think that sadly even the literal billions of dollars California is about to throw at this problem are unlikely to make a dent in any of:

1. the ‘voters are unhappy with public disorderliness by ‘homeless people’’ problem

2. The ‘homeless people actually do need a place to live though’ problem

3. The ‘people with mental health problems need ways to get better’ problem

Ironically, the one problem I think this has a chance of solving is the ‘the system has tonnes of tiny programs that make it hard to navigate and no one can tell how good they are or even how many people they’re helping, particularly en masse’ problem; purely because all those complicated little programs will be cut. A tonne of ‘innovation’ funding will be cut too, funding which was meant to be a way to systematically experiment to find better alternatives to the status quo, but which in 20 odd years hasn’t turned up anything that has both worked better and been attractive enough to the government that they’ve wanted to fund it on a mass scale (except maybe early psychosis programs, but even those are tiny).

Ironically, perhaps the most use this ballot measure will have is in its power for destruction; cutting jobs and forcing those workers into other jobs (and thus maybe easing some of the industry’s terrible labour shortages) or simply reducing the options so making it easier to choose between a few crappy options rather than a million unknown ones.

Only time will tell, and maybe it won’t even; it’s not like there’s a built-in mechanism for evaluating the effectiveness of a piece of legislation for achieving its stated goal, at least not here in California. 

Maybe that’s the ballot measure we should really be hoping for; a ballot measure to simply track the performance of all other ballot measures against what they said they were trying to do. 

---

If you enjoyed this essay, please share it with your favourite policymaker; if you don’t have one of those at hand, please share with anyone interested in mental health reform. Your support is super important; and I’m glad you’re reading!