Dave's Occasional Newsletter #2 (Updates! A theory of constraints! Medi-Cal!)
A theory of 3 constraints in services, some personal updates, and a little friction map/spelunk of Medicaid renewals from radio ad to web form
(Note: if you’re getting this email, it’s because you signed up here — though maybe it was some time ago! My first go at this was on Mailchimp, but I’ve decided to try out Substack.)
Personal updates: post-DOL, looking around at the world
In March, I left the US Department of Labor, where I was working on technology modernization in the unemployment insurance program, specifically around the $2 billion Congress gave USDOL for that purpose.
I originally signed up to be there about 6 months, but ended up staying closer to a year. While I was happy to get a few things done, I am happier about the things that will take some time to come to fruition on the government timescale.
I also learned quite a bit: both being inside government, but also for the first time operating with the federal vantage point, were instructive.
What am I doing next? Well, to start, I’m taking a bit of time to look around at the world, and think about what I think I’ve learned about it from the past few years, as well as what might have changed in the world (hint: I think LLMs like GPT-4 change a lot of the adjacent possible.)
In terms of “things” I’m doing, a few:
Experimenting in the build-measure-learn loop, getting back the tactile sense of building after operating in abstraction-land for some time (for example: here’s using OpenAI’s GPT-4 to summarize SNAP appeal docs; boy have I missed building things…)
Talking to users of various services I’m interested in (I’m attending a SNAP appeal today!)
Talking to orgs doing things I might like to do in a more proper job context (current skew: smaller orgs/start-ups with technology at the core)
Occasional consulting with orgs trying to do things I can help them with (most of this has happened through referral, but if you think you have a Dave-shaped problem, do feel free to drop me a note!)
A theory of 3 constraints in technology services (a sketch of a model)
Pattern-matching is useful and pattern-matching is dangerous: neat abstractions are friendlier than the mess that is the world. With that said, here’s a model I’ve been working out in my own domain that’s been useful to me. (Be careful with it! Specifics dominate generalities!)
A lot of my work has been around technology services, specifically online services around public benefits like SNAP (food stamps) or unemployment insurance. So, for example, most states have what would be called a benefits portal. This is an external user-facing online access point to engage with an agency.
From having worked on, in, and around such portals for close to a decade, I now believe there are 3 discrete constraints that must be tackled for these technology services to be Good1:
Situational awareness of problems (a high-signal view of what problems encounter now, the magnitude and frequency of those problems)
Design capacity (given a problem, the ability to scope a change that addresses it)
Technology change capacity (given a proposed change, the ability to make it)
I think I’ve come to believe that all three of these are necessary, and a focus on only one dimension is consistently insufficient and doesn’t move outcomes.
Let’s look at a few examples as to how a given constraint might shake out:
Constrained on #1: you can scope and deliver changes, but they’re constantly aimed at the wrong things
Constrained on #2: you see problems and you can ship, but every change you throw out there is ineffective
Constrained on #3: you’ve got a solid footing on problems and great proposed changes… and you can’t get anything done in less than 2 years (or worse, a once-a-decade procurement)
How does this help the practical work of making things better? Well, instead of looking at a service and saying, that’s bad, clearly everything is bad and everyone needs to be replaced, you can do a more precise diagnosis on which factor is holding the system back the most. You can identify the top bottleneck, and tackle that first.
An org that wants to build a native app might be better served with adding more capacity to situational awareness of problems (more user observation, more structured feedback channels, more instrumentation of/analytics on the existing service) to get to what the biggest challenges are.
An org that knows questions on multi-page forms need to be in simpler language — maybe due to academic research with rigorous findings on certain wording — might really need a technical path to make such changes in a lower-effort way.
What makes this a complex problem (rather than complicated) with something like a public benefits portal is that… these functions and capacity constraints may exist across varied actors, orgs, units, individuals.
One example: the priority of what the problems are in a public governance context might be a contested and negotiated space among actors like advocates, administrators, legislators, etc. So you’re not just looking at some better user research, you’re looking at shifting a consensus of what the top problems are. The Really Existing Design Process might actually be when a business analyst is given a broad 1-line description of the problem and has to generate requirements to give to a vendor based on some mental model of both the problem as described by leadership and a mental model of what is possible vs. impossible to be done by the vendor.
But what I’ve found useful about this mental model is that aspect of differential diagnosis on a specific user problem. Is a site not mobile responsive and lots of users are on mobile? Well, is Google Analytics installed to be able to see that number? Does the visual or interaction design of a help button make it so that many users don’t even know it’s there (let’s say, hidden way up in the top right) and so even though there’s quite useful help text, few users get to it? (Design constraint!) Does the org know of a page that errors out for users consistently, but simply isn’t in a technical-delivery-loop where that can be fixed in a timely manner?
Different problems, different interventions; and more specific starting points. That’s what this has given me.
As I said: consider this a sketch. It might be useful to you — but abstractions are dangerous and true ground truth should always win out!2
Friction mapping: a bit of a spelunk down the California Medicaid renewal path as unwinding happens
I was just in Los Angeles.3 On the radio I heard an ad about keeping your Medi-Cal! As lots of folks know, there's a thing called "unwinding" happening where people's Medicaid was kept going without the need for a renewal during the pandemic.
But with the end of the public health emergency, renewals are nigh!
Lots of thoughtful folks are working to avoid something many are rightfully afraid of: lots of people losing coverage because they have lost contact with the agency and can’t renew.
I’m a big nut for “friction mapping”: just looking at the actual path to doing a thing, and what friction people face in doing it. So let’s do it for this!
I heard an ad on the radio in LA — the takeaway I had from listening while driving (likely the most common case here?) was to do something about my Medi-Cal4
Okay, so I get on my phone and Google “medical renewal” (Why “medical” and not “medi-cal”? Because we do what’s easiest, and a hyphen is harder.)
Ah! Like I said, Google knows I’m in California, and shows Medi-Cal even though I said medical.
Okay the first result is a Covered California page, let’s go there:
Oh hrm… that’s kind of an annoying pop-up. And there’s a chatbot in addition to the pop-up. Okay, well let’s close it and see if the page answers our question.
Okay now this feels like a missed opportunity: this is the #1 search result for someone Googling about Medi-Cal renewal info. It gives generic info, but nothing about the public health emergency ending and the need to get in touch with your county.
Specifically the radio ad mentioned providing up-to-date contact info so that I — as a Medi-Cal user — would be sure to get notified when I need to do my much-delayed renewal.
But this page (1) lacks any of that timely info, and (2) has no action I can take to quickly do it. Think about the potential impact if right here on this page was a simple 3-question form to submit my name, latest phone number or email, and current mailing address.
This is not a blame or shame thing at all: I have no doubt that this is a helpful content page that CoveredCA has run for years that just rose in prominence. Also they’re not directly in charge of Medi-Cal renewals! But what an opportunity.
Which takes us to the second Google search result: the state agency over Medi-Cal itself! DHCS, we love you!
A nice thing at the top, though I notice that the contact your county option is a list of phone numbers. As part of this, I called one sizeable Bay Area county, and the phone tree was both long and there wasn’t a clear option to just update my info. After about 10 minutes of the phone tree and being on hold, I hung up. (Probably not uncommon for users taking this path?)
This I like! It has clear instructions, and gets me to my portal.
But I think I like EVEN MORE that #2 — Sign up for updates.
Unlike the other path — that involves logging into a portal account — I can just quickly sign up and get alerted.
The alerts are probably not going to be specific to my case, but this is a low-friction path to go from “I heard a radio ad” to signing up for alerts for action I’ll have to take. Nice job DHCS!
(Aside: if anyone reading this knows Covered California, feel free to suggest they do a similar thing on their page! This is the URL: https://www.coveredca.com/support/membership/medi-cal-renewal/ )
Say hello!
Okay that’s as much writing as I can do today. I am hoping to use this a bit more than other channels. I think this is probably better for my own brain than, say, oh I don’t know…Twitter.
But also there’s less of a feedback loop here! So I always enjoy a short email to say hello or something that resonated or just something that might interest me. I think you can just reply to this if you got it by email, or write me here.
“What is good” is an obviously contested space in a public governance context, but let’s skip over that for now.
I recently picked up the music producer Rick Rubin’s new book, The Creative Act, and it had this gem of an opener:
Los Angeles is great and I have been deeply wrong on that fact in the past.
Do I actually have Medi-Cal? That’s not the point. Your even asking is a HIPAA violation! (Narrator: it is not.)