Thank you for the public service of this essay (and 100x for the public service of the work it describes!). The detailed narrative creates understanding that probably couldn’t have been created any other way.
I love reading posts that aspire to improve government programs, but they tend to skip asking should the government be involved in the first place and if so, the level of government (federal, state, or local) that should manage that program.
For example, would federal oversight over state level licensing of hair salons prevented 80s haircuts? Even if I would have benefited from federally funded haircut regimes, that doesn't mean it's a good use of federal resources to set hair cut standards.
Federal oversight can course correct on regulatory overreach, but only if there exists well funded programs of checks and balances. For example, on IT security NIST sets the standards, federal agency CIOs craft the policies for risk management and enforcing standards, and IGs audit the results.
Lots of inspiration to draw from this post, but I hope future posts assess if and where a program belongs in government as part of considering opportunities for improvement.
Do we want the federal government to be involved in state level foster care licensing? Doesn’t the post illustrate an example of how the federal government’s oversight hinders foster care licensing? We’ve managed to live with state level licensing of doctors (which I understand doesn’t operate with the same level of federal oversight as foster care) who benefit from federal funding towards health care, yet we can’t live without federal oversight for foster care licensing?
It’s tough to agree when so many questions remain unanswered on what justifies federal oversight in place of state oversight.
I see your point. There are still good reasons for the federal government to play a role. Providing oversight and guidance and best practices is useful. This whole post is an example of federal coordination working well and accomplishing its goals - granted after a lot of difficulty up to that point.
Does a world with no federal coordination look better? What happens when a foster family has to move states and the two state foster systems are completely different?
My reading of the post: Federal oversight & guidance without accountablity lead to coping standards and cascades of rigidity. Plus, you don't need federal oversight and guidance when there exist model standards. States have the capacity to run their own programs and build national model standards on their own.
Someone else could read the post and conclude that every federal oversight program needs a program to build model standards.
I also read the post as an example of states working/leading together and not one of proactive federal leadership.
Lastly, there exist many examples of states expediting licensing for those previously licensed in other states. I wouldn’t use that issue as a justification for federal oversight.
Marina, your last line asks about procurement rules. That's where I've been stuck this year, so here's a field report from Finland.
Public buying has the same cascade, with money attached. The EU directive lets a buyer skip competition when it is absent "for technical reasons". National law copies that. The buyer's lawyer reads it as: if in doubt, write that. Nobody has ever been blamed for citing technical exclusivity. People get blamed for a tender that goes wrong.
The copy machine part is almost funny. This week Yle (our public broadcaster) went through ten municipal equipment tenders. In six of them only one manufacturer's machine fit, and some requirements were lifted straight from that manufacturer's brochure, down to the colour of the tractor. It's your trash-and-recycling line, except it came from a sales guide instead of HHS. And the law doesn't require the buyer to justify the criteria, so the brochure text just sits there looking like caution.
I got the software version this spring. A university direct-awarded an AI platform on technical grounds. I asked for the market analysis, the requirement matrix and the list of alternatives they had assessed, and all three answers were that the document does not exist. Every layer did the safe thing and nobody actually decided anything. Accountability sink, as advertised.
Your fix seems to translate pretty directly. A model justification for "only one supplier can do this" that spells out what evidence it requires. And a procurement version of your drop-table: publish the alternatives you looked at and why each one failed. If the list is empty, that's the answer.
(Disclosure, since it matters: I was the rejected alternative in that case, so weigh it accordingly.)
"We kept track of every requirement we dropped and why, in a table that's now longer than the standards themselves."
That table deserves a name, because it's the load-bearing instrument in the whole story.
In record-keeping terms it's a discount register: a struck requirement isn't deleted, it's printed with its reason and its source, which means the strike can be read backward — if conditions change, anyone can find the dropped rule, see why it was dropped, and argue it back in. That's what makes subtraction survivable in a system where, as you say, "nobody has ever been blamed for being too careful": the table converts "we removed a safeguard" into "we documented a decision." It's also most of an answer to Martin Sustrik's question above about who's accountable when a model standard backfires — the trail of what was dropped, why, and on whose reasoning is exactly what lets a state defend its standard, or reinstate a rule, without starting over. The cascade copies because each layer keeps what's checkable and sheds the judgment; your table is judgment made checkable. Thank you for tracing six identical trash-disposal sentences back to 1975 — the most convincing genealogy of a regulation I've read.
OMG… This is so insightful and rings so true. At FedEx, another large global bureaucracy operated the same way. Encounter an issue, write a new process. So many processes proliferated that everyone walked around with bubble wrap, protected from every kind of problem imaginable, but completely unable to move and suffocating from the bubble wrap.
We operated as if we needed to be wrapped to protect from the cold of a Minnesota winter, but instead we operated in the Arizona heat, and the woolen mitts and down parkas were killing us. Good news though, we would never freeze to death, only die from being unable to achieve our goals, but legal and the process bodies didn’t get credit for failing to achieve our goals, only punished if a risk that could’ve been prevented, happened.
Well shit happens, so for every incident, a new process, until the tape and bubble wrap asphyxiated.
And that is the biggest driver of all of this, as long as the punishment and reward structure, rewards compliance and avoiding risk/blame, then that’s what people will focus on.
If legal stops innovation, do they get punished no. If they approve something, and it’s a huge success, do they get rewarded, no. If they approve something, and it goes sideways, do they get blamed/punished, yes.
Net result, the default choice is to veto anything that might result in them being blamed, regardless of the potential organizational outcome.
And that’s the fundamental challenge, the rule creators are rewarded for avoiding risk, not for achieving larger goals that can’t be as easily attributed to each specific rule.
This sort of account shifts me back in the direction of my old dogmatically-anti-regulation libertarianism. I have no doubt that these thickets of rules are not only well-intentioned but actually prevent significant material harms to vulnerable people in some situations. But it's much less clear whether the amount of harm prevention exceeds the amount of harm caused by preventing or delaying what would otherwise be permissionless pro-social efforts.
Alas, as far as I can tell and despite the institutionalization of cost-benefit analysis, there does not seem to be a meta-rule in any jurisdiction requiring that the projected benefits (harm prevented or efficiency enabled) of any proposed new regulatory rule be weighed against the "time tax" (to use Annie Lowrey's phrase) of effort spent and worthwhile endeavors deterred by the burden of the new rule. Such a meta-rule would itself be a burden on those proposing new rules, likely greatly reducing the production of new regulations: and from where I sit, that looks like a feature, not a bug.
Real life is not software engineering. Libertarianism is fantasy. I have experience in IT engineering management and senior management experience in local public sector. Your political views will change the day you become responsible for actual public welfare. This “cost/benefit” myth is nonsense that comes from bad-faith actors.
This is an outstanding essay about important work!
Because measuring the projected benefits and projected harms would be exercises in futility. These are value judgements, and everyone's answers to what counts as each, and where to draw the line at when to stop counting benefits and harms would be different. It would do little if anything to improve the process, and could make it worse and more complicated.
This is a fascinating effort, and I wish you the best of luck.
From my own time inside a bureaucracy, I agree with your statement that adding a rule will typically be viewed as the safest course. And, as you say, this is not because people are evil. In my experience, one of the most important reasons is that people are afraid of being punished when things go wrong. We so overindex on the bad outcome and react very strongly to something going wrong. We often fail to act to address the ongoing burdens caused by rigidity and lost opportunities caused by failure to act.
In your case, a political test may come when an investigator empowered to use their discretion fails to spot a condition that later harms a child. It could be worse if the previous rules would have directed the investigator to note the condition and required that to be corrected before licensure. When the bad event happens, it may turn out that the bigger picture still shows tremendous positive benefits overall - many, many successful foster relationships occurring, against a much smaller number of errors. And yet, you still need to wonder if the system will acknowledge that and recalibrate without turning back to rigidity? Some politicians will want to bash caseworkers or the system, and strip it of discretion. It sounds like you're doing what you can to socialize the importance of discretion in key areas, and getting real feedback to get smarter and win buy-in, and I wish you the berst of luck as you try to rebalance discretion and rules!
Rules accumulate because the two costs sit with different people. Adding one costs the writer nothing and is carried by whoever operates the program. Removing one costs the remover personally the first time something goes wrong. Nobody owns the total. Every individual decision is correct for the person making it, which is why the stack keeps growing and why an audit of it changes nothing.
It is amazing and extremely depressing that this one area is fraught with all this excess of ridiculous rules. Thinking about how we turn this ship around - government effectiveness and efficiency - seems mind blowing. Do you get energized or depressed by the scale. Like most things, I guess, prioritizing is important. Does the Dems have plans - if they take over - on tackling important read? I know Pete talks about it, but coil plans? Thanks so much for your thoughtful piece. We need about 5 million of you
I've pictured Minnesota creating a standard. Then Wyoming decides to use it. When it backfires, can Wyoming shift the blame to Minnesota? If it can't, what's the incentive to adopt it in the first place rather than going down the rigidity cascade?
Thank you for the public service of this essay (and 100x for the public service of the work it describes!). The detailed narrative creates understanding that probably couldn’t have been created any other way.
I couldn't have said it better. This was an astoundingly informative, eye-opening, practical, and visionary piece.
I love reading posts that aspire to improve government programs, but they tend to skip asking should the government be involved in the first place and if so, the level of government (federal, state, or local) that should manage that program.
For example, would federal oversight over state level licensing of hair salons prevented 80s haircuts? Even if I would have benefited from federally funded haircut regimes, that doesn't mean it's a good use of federal resources to set hair cut standards.
Federal oversight can course correct on regulatory overreach, but only if there exists well funded programs of checks and balances. For example, on IT security NIST sets the standards, federal agency CIOs craft the policies for risk management and enforcing standards, and IGs audit the results.
Lots of inspiration to draw from this post, but I hope future posts assess if and where a program belongs in government as part of considering opportunities for improvement.
Seems like we should all agree that foster care licensing is something we want the government to be involved in, no?
Do we want the federal government to be involved in state level foster care licensing? Doesn’t the post illustrate an example of how the federal government’s oversight hinders foster care licensing? We’ve managed to live with state level licensing of doctors (which I understand doesn’t operate with the same level of federal oversight as foster care) who benefit from federal funding towards health care, yet we can’t live without federal oversight for foster care licensing?
It’s tough to agree when so many questions remain unanswered on what justifies federal oversight in place of state oversight.
I see your point. There are still good reasons for the federal government to play a role. Providing oversight and guidance and best practices is useful. This whole post is an example of federal coordination working well and accomplishing its goals - granted after a lot of difficulty up to that point.
Does a world with no federal coordination look better? What happens when a foster family has to move states and the two state foster systems are completely different?
My reading of the post: Federal oversight & guidance without accountablity lead to coping standards and cascades of rigidity. Plus, you don't need federal oversight and guidance when there exist model standards. States have the capacity to run their own programs and build national model standards on their own.
Someone else could read the post and conclude that every federal oversight program needs a program to build model standards.
I also read the post as an example of states working/leading together and not one of proactive federal leadership.
Lastly, there exist many examples of states expediting licensing for those previously licensed in other states. I wouldn’t use that issue as a justification for federal oversight.
As a former government worker deeply frustrated by our own bureaucracy, I found this essay riveting.
This post left me feeling really hopeful and inspired.
Marina, your last line asks about procurement rules. That's where I've been stuck this year, so here's a field report from Finland.
Public buying has the same cascade, with money attached. The EU directive lets a buyer skip competition when it is absent "for technical reasons". National law copies that. The buyer's lawyer reads it as: if in doubt, write that. Nobody has ever been blamed for citing technical exclusivity. People get blamed for a tender that goes wrong.
The copy machine part is almost funny. This week Yle (our public broadcaster) went through ten municipal equipment tenders. In six of them only one manufacturer's machine fit, and some requirements were lifted straight from that manufacturer's brochure, down to the colour of the tractor. It's your trash-and-recycling line, except it came from a sales guide instead of HHS. And the law doesn't require the buyer to justify the criteria, so the brochure text just sits there looking like caution.
I got the software version this spring. A university direct-awarded an AI platform on technical grounds. I asked for the market analysis, the requirement matrix and the list of alternatives they had assessed, and all three answers were that the document does not exist. Every layer did the safe thing and nobody actually decided anything. Accountability sink, as advertised.
Your fix seems to translate pretty directly. A model justification for "only one supplier can do this" that spells out what evidence it requires. And a procurement version of your drop-table: publish the alternatives you looked at and why each one failed. If the list is empty, that's the answer.
(Disclosure, since it matters: I was the rejected alternative in that case, so weigh it accordingly.)
"We kept track of every requirement we dropped and why, in a table that's now longer than the standards themselves."
That table deserves a name, because it's the load-bearing instrument in the whole story.
In record-keeping terms it's a discount register: a struck requirement isn't deleted, it's printed with its reason and its source, which means the strike can be read backward — if conditions change, anyone can find the dropped rule, see why it was dropped, and argue it back in. That's what makes subtraction survivable in a system where, as you say, "nobody has ever been blamed for being too careful": the table converts "we removed a safeguard" into "we documented a decision." It's also most of an answer to Martin Sustrik's question above about who's accountable when a model standard backfires — the trail of what was dropped, why, and on whose reasoning is exactly what lets a state defend its standard, or reinstate a rule, without starting over. The cascade copies because each layer keeps what's checkable and sheds the judgment; your table is judgment made checkable. Thank you for tracing six identical trash-disposal sentences back to 1975 — the most convincing genealogy of a regulation I've read.
OMG… This is so insightful and rings so true. At FedEx, another large global bureaucracy operated the same way. Encounter an issue, write a new process. So many processes proliferated that everyone walked around with bubble wrap, protected from every kind of problem imaginable, but completely unable to move and suffocating from the bubble wrap.
We operated as if we needed to be wrapped to protect from the cold of a Minnesota winter, but instead we operated in the Arizona heat, and the woolen mitts and down parkas were killing us. Good news though, we would never freeze to death, only die from being unable to achieve our goals, but legal and the process bodies didn’t get credit for failing to achieve our goals, only punished if a risk that could’ve been prevented, happened.
Well shit happens, so for every incident, a new process, until the tape and bubble wrap asphyxiated.
And that is the biggest driver of all of this, as long as the punishment and reward structure, rewards compliance and avoiding risk/blame, then that’s what people will focus on.
If legal stops innovation, do they get punished no. If they approve something, and it’s a huge success, do they get rewarded, no. If they approve something, and it goes sideways, do they get blamed/punished, yes.
Net result, the default choice is to veto anything that might result in them being blamed, regardless of the potential organizational outcome.
And that’s the fundamental challenge, the rule creators are rewarded for avoiding risk, not for achieving larger goals that can’t be as easily attributed to each specific rule.
Anyways, great story, and great work! Love this.
Thanks!
Kev
This sort of account shifts me back in the direction of my old dogmatically-anti-regulation libertarianism. I have no doubt that these thickets of rules are not only well-intentioned but actually prevent significant material harms to vulnerable people in some situations. But it's much less clear whether the amount of harm prevention exceeds the amount of harm caused by preventing or delaying what would otherwise be permissionless pro-social efforts.
Alas, as far as I can tell and despite the institutionalization of cost-benefit analysis, there does not seem to be a meta-rule in any jurisdiction requiring that the projected benefits (harm prevented or efficiency enabled) of any proposed new regulatory rule be weighed against the "time tax" (to use Annie Lowrey's phrase) of effort spent and worthwhile endeavors deterred by the burden of the new rule. Such a meta-rule would itself be a burden on those proposing new rules, likely greatly reducing the production of new regulations: and from where I sit, that looks like a feature, not a bug.
Real life is not software engineering. Libertarianism is fantasy. I have experience in IT engineering management and senior management experience in local public sector. Your political views will change the day you become responsible for actual public welfare. This “cost/benefit” myth is nonsense that comes from bad-faith actors.
This is an outstanding essay about important work!
Because measuring the projected benefits and projected harms would be exercises in futility. These are value judgements, and everyone's answers to what counts as each, and where to draw the line at when to stop counting benefits and harms would be different. It would do little if anything to improve the process, and could make it worse and more complicated.
This is a fascinating effort, and I wish you the best of luck.
From my own time inside a bureaucracy, I agree with your statement that adding a rule will typically be viewed as the safest course. And, as you say, this is not because people are evil. In my experience, one of the most important reasons is that people are afraid of being punished when things go wrong. We so overindex on the bad outcome and react very strongly to something going wrong. We often fail to act to address the ongoing burdens caused by rigidity and lost opportunities caused by failure to act.
In your case, a political test may come when an investigator empowered to use their discretion fails to spot a condition that later harms a child. It could be worse if the previous rules would have directed the investigator to note the condition and required that to be corrected before licensure. When the bad event happens, it may turn out that the bigger picture still shows tremendous positive benefits overall - many, many successful foster relationships occurring, against a much smaller number of errors. And yet, you still need to wonder if the system will acknowledge that and recalibrate without turning back to rigidity? Some politicians will want to bash caseworkers or the system, and strip it of discretion. It sounds like you're doing what you can to socialize the importance of discretion in key areas, and getting real feedback to get smarter and win buy-in, and I wish you the berst of luck as you try to rebalance discretion and rules!
Rules accumulate because the two costs sit with different people. Adding one costs the writer nothing and is carried by whoever operates the program. Removing one costs the remover personally the first time something goes wrong. Nobody owns the total. Every individual decision is correct for the person making it, which is why the stack keeps growing and why an audit of it changes nothing.
It is amazing and extremely depressing that this one area is fraught with all this excess of ridiculous rules. Thinking about how we turn this ship around - government effectiveness and efficiency - seems mind blowing. Do you get energized or depressed by the scale. Like most things, I guess, prioritizing is important. Does the Dems have plans - if they take over - on tackling important read? I know Pete talks about it, but coil plans? Thanks so much for your thoughtful piece. We need about 5 million of you
> You can publish competing model standards, and states can choose which one to follow.
Hm, how does that work? Who makes sure that the standard is compliant with the letter of law, and who's accountable if it isn't?
Why does having one or multiple standards change that?
I've pictured Minnesota creating a standard. Then Wyoming decides to use it. When it backfires, can Wyoming shift the blame to Minnesota? If it can't, what's the incentive to adopt it in the first place rather than going down the rigidity cascade?