Yes, definitely. This is similar to how large tech companies work - for example at Meta, roughly 1/3 of the engineering budget goes into "Infrastructure", rather than being tied to specific products.
So if you were in charge of developing a new product, like a new app, your team would not be responsible for:
1. Managing resources like servers or datacenters
2. Operating any standard services like databases, continuous integration, monitoring
3. Managing a developer environment
4. Multiple language support
5. Generic compliance
The list goes on and on. It's just much more efficient and high-quality to handle these things once instead of each product team. But it is expensive. To be worth it, you have to actually build infrastructure that gets used, not stuff that doesn't work. I think in an ideal world there probably would be something like 1/3 of the federal budget that goes to "any form of software development" going into the infrastructure team, similar to these big tech companies.
All of this standard stuff can easily take the majority of the time that a small team spends when they don't have an infrastructure team to handle it. And the other upside is quality of infrastructure can be better when the work is shared - at a large company you can have two people write an app in a month that scales to tens of millions of people.
As someone who's worked on critical internal tooling in industry, I've always felt that internally shared infrastructure and services are a sorely undervalued piece of what makes Big Tech companies able to work like they do. When your engineers and organizations don't have to repeatedly solve the same problems, and can put their shared learning/experiences back into one coherent thing, things can move so much faster. Beyond even just coherent public facing products like identity verification, shared codebases/deployment tooling etc really make a difference at scale. Interesting to me to read why this is harder in gov; something I've always been curious about.
To this amateur, Solitaire’s “practical roadmap”[1] looks like extremely effective communication: “here is a process for you to follow”. I wonder how far towards the “model legislation” template approach some of your initiatives can be pushed. Legislation-as-product?
I agree that building financial support for shared capabilities across agencies was difficult and tedious, but it was not impossible. Definately a grind, but very doable as many agencies had access to franchise funds for this purpose.
The true roadblock ( from my experience ) was data sharing agreements. The reason there are so many identity services is that agencies were unwilling to delegate control of data identified in agency level Privacy Impact Assessments or policies on records mgmt to operators of shared environments. Every general counsel at every agency remembers the OPM breach - it takes a lot more effort to manage legal/risk controls than financial controls.
Login.gov is a great success story, but tbf it builds on the data sharing framework put in place for the PIV card - which took a decade+ to put in place.
Great highlight on building shared capabilities, but I dont think money is the root cause. It's certainly a contributor.
To be honest, this is what I thought a lot of the PMA on CX was going to help fund in the last admin. I was surprised that there was not an effort to replicate the core personalization techniques used by all of industry, based heavily on underlying customer data platforms (CDPs) that enabled the cross-platform personalization that drove the vast digital success of industry over the last decade and a half. When I asked CMOs about the lack of these regular tools of industry in any of the CX PMA initiatives as the admin was wrapping up, they looked at me like I had two heads. One said that automated tools like those were bad because they increased "inequality of access." It's no surprise then that CX scores remained the lowest of any industry at the end of that large, multi year effort.
These technologies are not new, and login.gov, while important, is barely scratching the surface of how a cross-agency digital identity platform would be created and the value of would provide.
I would go one step further and say that the most valuable issue a service like this could resolve would be between levels of government appropriation, I.E. Fed>State>County>City, where fund allocation & outcomes can be understood far better, and therefore the reverse grant writing process would also be far more specific and actionable.
I know firsthand that the agencies furthest along on this path are the ones with large scale consumer marketing operations like USPS, Amtrak, the VA, etc. I worked heavily on similar systems for military recruitment. But all that golden data & systems are gated behind marketing contracts rather than integrated into the enterprise architecture like they should be.
It's a serious problem with unbelievable benefits across every level of government if we can get it right.
Congress is run by children, sychophants and drones. Let's start there. I sat on the digital council for the Federal government. My peers voiced their concerns and they went unheard. OMB was toothless and only looking for compliance. The way technology is going now, we don't require new infrastructure. AI tools negate the need for monolithic systems and with corporations being so embedded in government, it would be very simple to create technical standards and compliance protocols for use.
Great stuff. Honestly, this is why I hate the term product management, as it encourages focusing on a product, a specific application, not end-to-end solutions, which often requires working across multiple products to deliver solutions.
When I joined a startup as their first PM I even changed my title to Solutions Management, as we need to focus on solutions, not just specific applications, which is what the term product management often gets understood to mean.
The other challenge that I see organizations struggle with is even understanding that they are often working on similar capabilities that could benefit from becoming shared services.
At FedEx and elsewhere, I saw how different organizations each had created their own acronyms and names to describe the same thing, but since no one spoke the same language, they all assumed they were developing unique capabilities.
It was as if instead of everyone calling a tire a tire, they referred to them as Pirellis, Firestones, and Bridgestones, without ever realizing they were tires, and should be designed and developed holistically to work across cars.
Worse, as a product manager, you’re not rewarded for looking beyond your application/product, so you just rebuild more crappy 💩within your product as no one is empowered to build cross-cutting infrastructure.
Sorry, but your post triggered me. 😂
So love this for government, but it’s also a need for private sector as well.
I feel like this (great) concept butts up against the results of Loper Bright. If Chevron deference is no more, legislators have to sharpen language to narrow the amount of discretion an agency has, no? An agency head would need to find appropriation language that allows for this capability model while avoiding vagueness. Makes me want to go find the Post-Loper Bright drafting guides.
Yes, definitely. This is similar to how large tech companies work - for example at Meta, roughly 1/3 of the engineering budget goes into "Infrastructure", rather than being tied to specific products.
So if you were in charge of developing a new product, like a new app, your team would not be responsible for:
1. Managing resources like servers or datacenters
2. Operating any standard services like databases, continuous integration, monitoring
3. Managing a developer environment
4. Multiple language support
5. Generic compliance
The list goes on and on. It's just much more efficient and high-quality to handle these things once instead of each product team. But it is expensive. To be worth it, you have to actually build infrastructure that gets used, not stuff that doesn't work. I think in an ideal world there probably would be something like 1/3 of the federal budget that goes to "any form of software development" going into the infrastructure team, similar to these big tech companies.
All of this standard stuff can easily take the majority of the time that a small team spends when they don't have an infrastructure team to handle it. And the other upside is quality of infrastructure can be better when the work is shared - at a large company you can have two people write an app in a month that scales to tens of millions of people.
As someone who's worked on critical internal tooling in industry, I've always felt that internally shared infrastructure and services are a sorely undervalued piece of what makes Big Tech companies able to work like they do. When your engineers and organizations don't have to repeatedly solve the same problems, and can put their shared learning/experiences back into one coherent thing, things can move so much faster. Beyond even just coherent public facing products like identity verification, shared codebases/deployment tooling etc really make a difference at scale. Interesting to me to read why this is harder in gov; something I've always been curious about.
okay, you definitely caught up on writing. Thanks for this.
To this amateur, Solitaire’s “practical roadmap”[1] looks like extremely effective communication: “here is a process for you to follow”. I wonder how far towards the “model legislation” template approach some of your initiatives can be pushed. Legislation-as-product?
[1] https://www.niskanencenter.org/capability-based-budgeting-a-practical-roadmap-for-getting-started/
I agree that building financial support for shared capabilities across agencies was difficult and tedious, but it was not impossible. Definately a grind, but very doable as many agencies had access to franchise funds for this purpose.
The true roadblock ( from my experience ) was data sharing agreements. The reason there are so many identity services is that agencies were unwilling to delegate control of data identified in agency level Privacy Impact Assessments or policies on records mgmt to operators of shared environments. Every general counsel at every agency remembers the OPM breach - it takes a lot more effort to manage legal/risk controls than financial controls.
Login.gov is a great success story, but tbf it builds on the data sharing framework put in place for the PIV card - which took a decade+ to put in place.
Great highlight on building shared capabilities, but I dont think money is the root cause. It's certainly a contributor.
To be honest, this is what I thought a lot of the PMA on CX was going to help fund in the last admin. I was surprised that there was not an effort to replicate the core personalization techniques used by all of industry, based heavily on underlying customer data platforms (CDPs) that enabled the cross-platform personalization that drove the vast digital success of industry over the last decade and a half. When I asked CMOs about the lack of these regular tools of industry in any of the CX PMA initiatives as the admin was wrapping up, they looked at me like I had two heads. One said that automated tools like those were bad because they increased "inequality of access." It's no surprise then that CX scores remained the lowest of any industry at the end of that large, multi year effort.
These technologies are not new, and login.gov, while important, is barely scratching the surface of how a cross-agency digital identity platform would be created and the value of would provide.
I would go one step further and say that the most valuable issue a service like this could resolve would be between levels of government appropriation, I.E. Fed>State>County>City, where fund allocation & outcomes can be understood far better, and therefore the reverse grant writing process would also be far more specific and actionable.
I know firsthand that the agencies furthest along on this path are the ones with large scale consumer marketing operations like USPS, Amtrak, the VA, etc. I worked heavily on similar systems for military recruitment. But all that golden data & systems are gated behind marketing contracts rather than integrated into the enterprise architecture like they should be.
It's a serious problem with unbelievable benefits across every level of government if we can get it right.
Congress is run by children, sychophants and drones. Let's start there. I sat on the digital council for the Federal government. My peers voiced their concerns and they went unheard. OMB was toothless and only looking for compliance. The way technology is going now, we don't require new infrastructure. AI tools negate the need for monolithic systems and with corporations being so embedded in government, it would be very simple to create technical standards and compliance protocols for use.
Great stuff. Honestly, this is why I hate the term product management, as it encourages focusing on a product, a specific application, not end-to-end solutions, which often requires working across multiple products to deliver solutions.
When I joined a startup as their first PM I even changed my title to Solutions Management, as we need to focus on solutions, not just specific applications, which is what the term product management often gets understood to mean.
The other challenge that I see organizations struggle with is even understanding that they are often working on similar capabilities that could benefit from becoming shared services.
At FedEx and elsewhere, I saw how different organizations each had created their own acronyms and names to describe the same thing, but since no one spoke the same language, they all assumed they were developing unique capabilities.
It was as if instead of everyone calling a tire a tire, they referred to them as Pirellis, Firestones, and Bridgestones, without ever realizing they were tires, and should be designed and developed holistically to work across cars.
Worse, as a product manager, you’re not rewarded for looking beyond your application/product, so you just rebuild more crappy 💩within your product as no one is empowered to build cross-cutting infrastructure.
Sorry, but your post triggered me. 😂
So love this for government, but it’s also a need for private sector as well.
I feel like this (great) concept butts up against the results of Loper Bright. If Chevron deference is no more, legislators have to sharpen language to narrow the amount of discretion an agency has, no? An agency head would need to find appropriation language that allows for this capability model while avoiding vagueness. Makes me want to go find the Post-Loper Bright drafting guides.