Excellent framing and I hope more members of the public come to learn of and admire the commitment of the many Jed’s in govtech. That said, for any of the new special gov employees to learn from Jed, they’d need to WANT to learn from him. In other words, they’d need to share his goal of more effective government and…I seriously doubt they do. Their actions, the nature of their shock and awe engagement with the current staff, and the reports of who they seem to have been in private life all suggest that they are simply don’t care if they break things OR perhaps are there explicitly to break things. Dan Hon had an great thread on BlueSky that maintained it is actually the motivations of decision makers that determine the outcomes of such efforts - is there ANY indication that the folks currently involved have a goal of improving gov outcomes?
A very slightly more charitable explanation is that they are trying to fight the last war, in a sense, by replicating Elon's drastic and sudden slim-down of Twitter. That only worked because Twitter, like other large tech companies at the time, really had become overstaffed due to talent hoarding. People across the industry understood this well before Elon came along; he was just the first to act on it.
The prejudiced assumption that the federal government must be similarly overstaffed is thus likely a big part of their thinking. It is extremely inaccurate, indeed the opposite of the truth, but it will take a lot of misery inflicted on millions of innocent bystanders before they realize it.
Yes, the key difference here is the severity of the consequences from misguided policies/actions and/or any mistakes. At Twitter things could reasonably disintegrate with no real practical harm. Missed or blocked payments on prior obligations but the Treasury - or technical issues stemming from carelessness - could undermine the nations’ credit. Setting aside the constitutional issue of ignoring congress, our entire monetary and banking system are put at risk. The precise opposite of state capacity - this could crater gov effectiveness for a generation or more.
I don't know any of the people involved in DOGE but I did know a number of the people involved in the USDS work previously, who came over from having tech industry experience.
The common thread I heard is that in comparison to the tech industry, the technology groups in the federal government are very overstaffed. However, the average skill level is much lower. So, you end up with a team of 40 people working on something that in industry would typically have 5 people. You also might pay each of those 40 people 1/3 as much as you would pay in industry. So in terms of headcount, you might be 8x overstaffed. But in terms of budget, you're only spending 3x more than you need to. And the team doesn't *feel* overstaffed because they are struggling to achieve their current goals.
Typically in industry the way you would handle this problem is large scale layoffs, like firing your bottom 2/3 of the underperforming teams. And then raise your pay scale and your hiring standards, so that you can hire back to a smaller but more effective team with the same budget.
So, again I don't know the DOGE people, but to me this kinda seems like what's going on. Perhaps they are thinking from a tech industry playbook.
Again, the practices being deployed are indeed taken from the private sector but the only thing that really matters in these instances is the ‘why?’ - the goal of say, a PE firm or Musk type investor, is to ore efficiently and effectively achieve the stated goals of the organization. What we’re witnessing seems to be what would happen if a PE firm wanted to tank the effectiveness of their new acquisition. The playbook is similar but the game is completely different, and I fear, dangerous.
This post is an indictment not a heartwarming story. For decades no one cared about the plumbing, now there's intense scrutiny on it. While we should celebrate the Jeds across gov for doing more with less, the actual lesson is that a systems like Jeds never should have had to happen in the first place.
Now we have an opportunity to replace them with modern design standards, and many commenters seem convinced that Jeds systems are actually sacrosanct. It's the furthest thing from the truth. These systems are defined by their edge cases, rather than the overall execution of their core purpose. It would not be hard to improve on the vast majority of them.
What it's an indictment of is the U.S. public refusing to understand that replacing the plumbing costs money and time, not Jed or any of the people trying their darnedest to create something out of nothing. I do not believe for a minute that Jed's system or systems like this are sacrosanct. What I do believe is calling Jed and his coworkers evil because DOGE and Musk are too mean and stupid to realize that the problem is too few employees and too little money means that federal workers make do with tape and string and gum and prayers.
True but it's half the fault of bureaucrats refusing to let anyone else touch or connect with their holy systems and pushing policies isolating their data & tools anyone else. Look at the arguments people are making, people truly think that it's impossible to overhaul treasury's payment system or social security fraud reviews etc because the gov systems are so unique.
Maybe that was true in an era before billion user platforms, but it's downright ridiculous now. The line you're saying in that all these huge systems are actually just bootstrapped and duct taped was not the argument against modernization for decades and decades. It was quite the opposite.
I don't think you understand what you're talking about about here. The tip off is that you use the term "bureaucrats" pejoratively as a lazy descriptor of who's responsible, or that they're making personal decisions about allowing access to sensitive government data and data systems.
Also, when you toss off things like "defined by edge cases", well, those edge cases are there by law. You can't just not support them and focus on core operations. You can't just say you'll throw in the backlog for vNext, it has to be there or there are legal consequences.
If you look at the long history of big private sector tech companies with billion user systems getting government contracts to develop/modernize Gov systems, you'll see a lot of failure. Just look at how the EHR modernization at the VA is going with Oracle (Cerner), and the withdrawal of their previous partner, Google.
It's not the case that the Federal Government is full of people that are unaware of modern software and architectures. Their are many extremely smart people working in government. They spend too much time figuring out how to implement policy required by Congress, usually underfunded and with numerous overly prescriptive language, or overly vague language. Constant hold ups on funding don't help. Constant lawsuits where some judge who doesn't know anything decides they know better than anyone what Congress should have meant, and the constant meddling of idiots in Congress who know just as much.
I'm sorry my decade+ of work leading public sector tech implementations leads to opinions you don't like. The choice to use the VA EHR system is a funny one because it's precisely because they want to build on oracles extremely old and outdated systems that it failed! Google backed out because the VA ran over budget doing the same old stuff, and provided no path to do anything at all to fix the legacy issues with modern design! There was no project management plan! The whole debacle is a perfect use case of garbage technology and garbage system design, and then blaming the failures on a different vendor. Amazing.
Unfortunately it is absolutely the case that the Fed is full of people who completely missed the boat on modern enterprise cloud architectures. Absolutely full. If you don't think so that's your own ignorance. They resisted modernization for so long the world passed them by. They didn't bend for so long that any improvement would break their carefully constructed but wildly inefficient systems.
It's also not true that Congress has not been funding them. IT modernization is one of the only things Congress agrees on and passes regularly. TMF money consistently left on the table. DOD missed everything benchmark for IT modernization and hundreds of millions have gone unspent. Congress demanded DOJ reform PACER but they didn't because of these edge cases concerns and sent the money back. The GAO literally cites "lack of skilled personnel" frequently as a reason why these modernizations don't work!
Again the reflexive defense leads to simply lying about the reality of these initiatives. Id suggest not doing it.
Has it ever dawned on you that you and the previous commenter can both be right? That perhaps decades of red tape has prohibited the government from hiring the people and tools desperately needed to make these improvements and sustain them?
The VA replacing Vista with Cerner was a huge mistake. Epic would have done it right but lowest bidder always wins. Cerner is now even worse being owned by Oracle, whose customer support is atrocious.
Probably right... Vista I believe was MUMPS based, which Epic supports more natively (built on InterSystems Cache/IRIS, which evolved from MUMPS), might have made it an easier transition.
Thanks for this great perspective. I admire the dedication, but I also hope that we can move away from against-the-odds dedication and towards ways of working that actually… work fundamentally, not just through extraordinary duct tape. Maybe a few things need to break in order to foster that process. Having said that, I have little confidence that Doge will have a positive impact, but I am open to being shown I’m wrong.
Correct me if I'm wrong, but isn't Elon Musk's strategy for building companies to create an environment which demands, "against-the-odds dedication"?
Or, to put it another way, it's possible to find examples of organizations that have, "ways of working that actually… work" but I suspect that's the exception rather than the rule.
Yes, you do this through Congress. It's not like Jed didn't know any better, it's that his system never rose to a high enough priority to receive modernization funding, but along the way the legal process the system supports probably received multiple legal changes to it. What do you do? What magical thing happens that nobody thought of until Elon Musk came along?
Agreed. The dedication is remarkable and heroic, but only because he's paddling upstream with a system pumping water against him. Elon Musk's breaking things is not going to help but we have to find a way for things to work better. Only problem is that people want it for free - they don't understand the investment that needs to take place.
A friend sent me a link to this post and I read it and was instantly reminded of a book that some coworkers and I read: Recoding America, a relatively short book that is full of stories like this together with the context of how and why things came to be this way and the magnitude of the challenge of fixing them. It hit a little too close to home for all of us. I came to the comments to recommend it and stopped mid-draft to look up the author - who is of course Jennifer Palka.
I’m kind of surprised by the tone of the comments. The fact is that we don’t know what the motivations of these people are. Perhaps they want to make things better for the people of this country and the programs that they and their relatives depend on. I guess you could argue that their motives must be bad because they were hired indirectly, most likely, by people who …have questionable motives? I would have volunteered had I the skills and as a retired fed, will be sending suggestions to DOGE for improvement.
If they’d wanted to identify problems, they would have hired auditors, not programmers. And that actions like cutting funding bad eliminating AID were taken before results were in anolybdemonstrates the bad faith.
Recoding America has become my bedtime story (paperback arrived in the mail last week). Thank you for writing this. Once DOGE is done rampaging through the government systems and culture (and what they are doing is fundamentally aimed at breaking the culture, IMHO (am also reading Musk's bio right now)), I hope every one of the people who comes in afterwards to do the cleanup and rebuilding has read your book.
I don’t know Jed so I can’t speak to this particular case. But in my experience as a software engineer, usually the people who are completely indispensable for the operation of some ancient system, this not necessarily a positive sign about their ability. When you’re maintaining a system it’s part of your job to keep it operational and maintainable by people who aren’t you.
Sometimes people don’t have the skill to make a system easily maintainable. Sometimes they are reluctant to document it because they want to maintain their job security or because they are just shy or inexperienced. So it’s hard to say why.
Often the best way to proceed is to force them off the project. But again hard to say in this case.
Of course. Jed, by his own account, was happy to be forced off the project. He wanted to retire. Others hold on too tightly, for sure. But it's not his fault that there was no system to replace his for so many years.
Back in the mid-1990s I had a gig updating technical documentation for NY State HESC (Higher Education Services Corporation). That's the organization that handled student loans for NY State. The machine was an old Honeywell mainframe and the code was, of course, COBOL. They had to bring in programmers from India to work on it.
That was the first time I came face-to-face with the dreaded spaghetti code. I was documenting some module and had to trace my way through a string containing, let's say, a half dozen GOTO statements. The last one pointed to a block of code that had been commented out. Whoops!
I wonder of any of the DOGE kids have even seen COBOL.
Thanks for this. Part of my career in the biopharma industry dealt with electronic submissions of regulatory documents and drug safety. A lot of this focused on improving the IT capabilities of the Food and Drug Administration and implementation of industry-paid user fees helped modernize things at the agency. Later we established a pilot program to look for safety signals in large observational databases using open source tools as well as developing new ones. The FDA colleagues we worked with were all engaged in this effort but the key problem is that healthcare databases in the US focus on claims data for reimbursement whereas countries with single payer systems have core medical records. It's far easier to do the kind of work we were interested in the latter systems. Of course this is a result of policiy decisions about how countries provide for healthcare.
Thank you for writing this. I also just bought your book to get a better understanding of the complexities of government technology systems and the difficulty of improving them. Your empathy and that of others is inspiring, and stands in such stark contrast to the meanness of President Trump, VP Vance, Musk, and the whole array of cabinet members and Republican "leaders". But empathy is also PRACTICAL as you show here and in your book. That's what we don't see here, which convinces me that this is not reform but an outright steal by thugs.
It's worth noting that COBOL, for all its faults, is fast as hell when running on IBM mainframes. This matters for high-volume real time systems, like IRS processing tax returns, or many bank applications. When the youngsters come in to these environments with promises to recast old code into newer, more flexible and maintainable languages, the resulting code is often just too slow for the job. There's a reason lots of critical systems in the private sector are still running in COBOL, and that obviously can't be attributed to personnel policies or subpar IT strategies at the federal level. Likewise, the fastest code for massive microsimulation modeling is FORTRAN, not Python or Java or C++. It's not even close.
One of my friends is a SWE in civil engineering. According to him, a piece of commonly used software was made in the 90s by a German construction worker, who had no background in programming. Apparently you can see the evolution of his knowledge in the source code: he learns about object-oriented programming, then for awhile he experiments with functional programming before giving up on it, and so forth. So maybe there are a few Jeds outside of government, too! Unsung heroes.
I worked directly with the person you named on this exact modernization project for many months.
I write the following with immense respect for your extraordinary work, Jen:
- I'd like to know if he consented to his story appearing here. At a time when public servants are being targeted for dismissal, bullying, and harassment, using his name and agency affiliation may bring unwanted attention his way, especially as DOGE is working through the VA and Trump admin-adjacent foundations are creating watchlists of current and ex-feds. And if he indeed consented for this - more power to him, wishing him a well-deserved retirement!
- This is the time for tangibly defending public servants from DOGE's search-and-destroy-then-plunder mission. I hope you agree that we need to protect the identities of current and recent government employees in harm's way - unless they've consented to being publicly named at this moment.
- We need to prevent the interlopers from finding more linchpin systems and public servants to destroy and exploit. Empathy building exercises will not stop committed arsonists - as I wrote this comment, the actual DOGE staffer at Treasury just resigned after a Wall Street Journal reporter surfaced his history of deeply racist posts (railing against the Civil Rights Act and interracial marriage, to name a few examples).
I'd like to believe that the college kids that Elon Musk has empowered to vandalize the federal government are capable of the kind of introspection that Jen describes here.
Unfortunately, the early evidence is that they are MAGA'd up, overgrown teens who are hopped up on overconfidence and subsisting on a diet of Nazi memes and potentially a dose of the boss' ketamine.
The problem with this framing is that Jed is only valuable because the system is valuable. Our current overlords want to burn everything down.
If you want to burn a building down, you don’t need to stop and appreciate the architects.
Excellent framing and I hope more members of the public come to learn of and admire the commitment of the many Jed’s in govtech. That said, for any of the new special gov employees to learn from Jed, they’d need to WANT to learn from him. In other words, they’d need to share his goal of more effective government and…I seriously doubt they do. Their actions, the nature of their shock and awe engagement with the current staff, and the reports of who they seem to have been in private life all suggest that they are simply don’t care if they break things OR perhaps are there explicitly to break things. Dan Hon had an great thread on BlueSky that maintained it is actually the motivations of decision makers that determine the outcomes of such efforts - is there ANY indication that the folks currently involved have a goal of improving gov outcomes?
A very slightly more charitable explanation is that they are trying to fight the last war, in a sense, by replicating Elon's drastic and sudden slim-down of Twitter. That only worked because Twitter, like other large tech companies at the time, really had become overstaffed due to talent hoarding. People across the industry understood this well before Elon came along; he was just the first to act on it.
The prejudiced assumption that the federal government must be similarly overstaffed is thus likely a big part of their thinking. It is extremely inaccurate, indeed the opposite of the truth, but it will take a lot of misery inflicted on millions of innocent bystanders before they realize it.
Yes, the key difference here is the severity of the consequences from misguided policies/actions and/or any mistakes. At Twitter things could reasonably disintegrate with no real practical harm. Missed or blocked payments on prior obligations but the Treasury - or technical issues stemming from carelessness - could undermine the nations’ credit. Setting aside the constitutional issue of ignoring congress, our entire monetary and banking system are put at risk. The precise opposite of state capacity - this could crater gov effectiveness for a generation or more.
I don't know any of the people involved in DOGE but I did know a number of the people involved in the USDS work previously, who came over from having tech industry experience.
The common thread I heard is that in comparison to the tech industry, the technology groups in the federal government are very overstaffed. However, the average skill level is much lower. So, you end up with a team of 40 people working on something that in industry would typically have 5 people. You also might pay each of those 40 people 1/3 as much as you would pay in industry. So in terms of headcount, you might be 8x overstaffed. But in terms of budget, you're only spending 3x more than you need to. And the team doesn't *feel* overstaffed because they are struggling to achieve their current goals.
Typically in industry the way you would handle this problem is large scale layoffs, like firing your bottom 2/3 of the underperforming teams. And then raise your pay scale and your hiring standards, so that you can hire back to a smaller but more effective team with the same budget.
So, again I don't know the DOGE people, but to me this kinda seems like what's going on. Perhaps they are thinking from a tech industry playbook.
Again, the practices being deployed are indeed taken from the private sector but the only thing that really matters in these instances is the ‘why?’ - the goal of say, a PE firm or Musk type investor, is to ore efficiently and effectively achieve the stated goals of the organization. What we’re witnessing seems to be what would happen if a PE firm wanted to tank the effectiveness of their new acquisition. The playbook is similar but the game is completely different, and I fear, dangerous.
I think you’re imagining those numbers.
This post is an indictment not a heartwarming story. For decades no one cared about the plumbing, now there's intense scrutiny on it. While we should celebrate the Jeds across gov for doing more with less, the actual lesson is that a systems like Jeds never should have had to happen in the first place.
Now we have an opportunity to replace them with modern design standards, and many commenters seem convinced that Jeds systems are actually sacrosanct. It's the furthest thing from the truth. These systems are defined by their edge cases, rather than the overall execution of their core purpose. It would not be hard to improve on the vast majority of them.
What it's an indictment of is the U.S. public refusing to understand that replacing the plumbing costs money and time, not Jed or any of the people trying their darnedest to create something out of nothing. I do not believe for a minute that Jed's system or systems like this are sacrosanct. What I do believe is calling Jed and his coworkers evil because DOGE and Musk are too mean and stupid to realize that the problem is too few employees and too little money means that federal workers make do with tape and string and gum and prayers.
True but it's half the fault of bureaucrats refusing to let anyone else touch or connect with their holy systems and pushing policies isolating their data & tools anyone else. Look at the arguments people are making, people truly think that it's impossible to overhaul treasury's payment system or social security fraud reviews etc because the gov systems are so unique.
Maybe that was true in an era before billion user platforms, but it's downright ridiculous now. The line you're saying in that all these huge systems are actually just bootstrapped and duct taped was not the argument against modernization for decades and decades. It was quite the opposite.
I don't think you understand what you're talking about about here. The tip off is that you use the term "bureaucrats" pejoratively as a lazy descriptor of who's responsible, or that they're making personal decisions about allowing access to sensitive government data and data systems.
Also, when you toss off things like "defined by edge cases", well, those edge cases are there by law. You can't just not support them and focus on core operations. You can't just say you'll throw in the backlog for vNext, it has to be there or there are legal consequences.
If you look at the long history of big private sector tech companies with billion user systems getting government contracts to develop/modernize Gov systems, you'll see a lot of failure. Just look at how the EHR modernization at the VA is going with Oracle (Cerner), and the withdrawal of their previous partner, Google.
It's not the case that the Federal Government is full of people that are unaware of modern software and architectures. Their are many extremely smart people working in government. They spend too much time figuring out how to implement policy required by Congress, usually underfunded and with numerous overly prescriptive language, or overly vague language. Constant hold ups on funding don't help. Constant lawsuits where some judge who doesn't know anything decides they know better than anyone what Congress should have meant, and the constant meddling of idiots in Congress who know just as much.
I'm sorry my decade+ of work leading public sector tech implementations leads to opinions you don't like. The choice to use the VA EHR system is a funny one because it's precisely because they want to build on oracles extremely old and outdated systems that it failed! Google backed out because the VA ran over budget doing the same old stuff, and provided no path to do anything at all to fix the legacy issues with modern design! There was no project management plan! The whole debacle is a perfect use case of garbage technology and garbage system design, and then blaming the failures on a different vendor. Amazing.
Unfortunately it is absolutely the case that the Fed is full of people who completely missed the boat on modern enterprise cloud architectures. Absolutely full. If you don't think so that's your own ignorance. They resisted modernization for so long the world passed them by. They didn't bend for so long that any improvement would break their carefully constructed but wildly inefficient systems.
It's also not true that Congress has not been funding them. IT modernization is one of the only things Congress agrees on and passes regularly. TMF money consistently left on the table. DOD missed everything benchmark for IT modernization and hundreds of millions have gone unspent. Congress demanded DOJ reform PACER but they didn't because of these edge cases concerns and sent the money back. The GAO literally cites "lack of skilled personnel" frequently as a reason why these modernizations don't work!
Again the reflexive defense leads to simply lying about the reality of these initiatives. Id suggest not doing it.
Has it ever dawned on you that you and the previous commenter can both be right? That perhaps decades of red tape has prohibited the government from hiring the people and tools desperately needed to make these improvements and sustain them?
I wholeheartedly agree with that being a dynamic here. But a bad one!
The VA replacing Vista with Cerner was a huge mistake. Epic would have done it right but lowest bidder always wins. Cerner is now even worse being owned by Oracle, whose customer support is atrocious.
Probably right... Vista I believe was MUMPS based, which Epic supports more natively (built on InterSystems Cache/IRIS, which evolved from MUMPS), might have made it an easier transition.
You’re absolutely correct, and aside from being based on MUMPS, Epic is competent. Cerner is not and Oracle hasn’t improved things one iota.
Thanks for this great perspective. I admire the dedication, but I also hope that we can move away from against-the-odds dedication and towards ways of working that actually… work fundamentally, not just through extraordinary duct tape. Maybe a few things need to break in order to foster that process. Having said that, I have little confidence that Doge will have a positive impact, but I am open to being shown I’m wrong.
Completely agree we need to "move away from against-the-odds dedication and towards ways of working that actually"
Correct me if I'm wrong, but isn't Elon Musk's strategy for building companies to create an environment which demands, "against-the-odds dedication"?
Or, to put it another way, it's possible to find examples of organizations that have, "ways of working that actually… work" but I suspect that's the exception rather than the rule.
Yes, you do this through Congress. It's not like Jed didn't know any better, it's that his system never rose to a high enough priority to receive modernization funding, but along the way the legal process the system supports probably received multiple legal changes to it. What do you do? What magical thing happens that nobody thought of until Elon Musk came along?
Agreed. The dedication is remarkable and heroic, but only because he's paddling upstream with a system pumping water against him. Elon Musk's breaking things is not going to help but we have to find a way for things to work better. Only problem is that people want it for free - they don't understand the investment that needs to take place.
A friend sent me a link to this post and I read it and was instantly reminded of a book that some coworkers and I read: Recoding America, a relatively short book that is full of stories like this together with the context of how and why things came to be this way and the magnitude of the challenge of fixing them. It hit a little too close to home for all of us. I came to the comments to recommend it and stopped mid-draft to look up the author - who is of course Jennifer Palka.
I’m kind of surprised by the tone of the comments. The fact is that we don’t know what the motivations of these people are. Perhaps they want to make things better for the people of this country and the programs that they and their relatives depend on. I guess you could argue that their motives must be bad because they were hired indirectly, most likely, by people who …have questionable motives? I would have volunteered had I the skills and as a retired fed, will be sending suggestions to DOGE for improvement.
When they start with foreign aid and call USAID a criminal organization (Elon), it doesn’t inspire confidence.
They are locking federal employees out of their offices.
Please wake up. They are motivated by insane ramblings about government waste that gets fed to them through social media. Or raw power and greed.
If they’d wanted to identify problems, they would have hired auditors, not programmers. And that actions like cutting funding bad eliminating AID were taken before results were in anolybdemonstrates the bad faith.
What an inspiring and terrifying story.
Recoding America has become my bedtime story (paperback arrived in the mail last week). Thank you for writing this. Once DOGE is done rampaging through the government systems and culture (and what they are doing is fundamentally aimed at breaking the culture, IMHO (am also reading Musk's bio right now)), I hope every one of the people who comes in afterwards to do the cleanup and rebuilding has read your book.
Thank you! I hope so too. :)
Yes! I am telling everyone I know to read this book.
I don’t know Jed so I can’t speak to this particular case. But in my experience as a software engineer, usually the people who are completely indispensable for the operation of some ancient system, this not necessarily a positive sign about their ability. When you’re maintaining a system it’s part of your job to keep it operational and maintainable by people who aren’t you.
Sometimes people don’t have the skill to make a system easily maintainable. Sometimes they are reluctant to document it because they want to maintain their job security or because they are just shy or inexperienced. So it’s hard to say why.
Often the best way to proceed is to force them off the project. But again hard to say in this case.
Of course. Jed, by his own account, was happy to be forced off the project. He wanted to retire. Others hold on too tightly, for sure. But it's not his fault that there was no system to replace his for so many years.
Back in the mid-1990s I had a gig updating technical documentation for NY State HESC (Higher Education Services Corporation). That's the organization that handled student loans for NY State. The machine was an old Honeywell mainframe and the code was, of course, COBOL. They had to bring in programmers from India to work on it.
That was the first time I came face-to-face with the dreaded spaghetti code. I was documenting some module and had to trace my way through a string containing, let's say, a half dozen GOTO statements. The last one pointed to a block of code that had been commented out. Whoops!
I wonder of any of the DOGE kids have even seen COBOL.
Thanks for this. Part of my career in the biopharma industry dealt with electronic submissions of regulatory documents and drug safety. A lot of this focused on improving the IT capabilities of the Food and Drug Administration and implementation of industry-paid user fees helped modernize things at the agency. Later we established a pilot program to look for safety signals in large observational databases using open source tools as well as developing new ones. The FDA colleagues we worked with were all engaged in this effort but the key problem is that healthcare databases in the US focus on claims data for reimbursement whereas countries with single payer systems have core medical records. It's far easier to do the kind of work we were interested in the latter systems. Of course this is a result of policiy decisions about how countries provide for healthcare.
When did you do this work? Comprehensive EHR systems like Epic and Cerner are very common now, but less so 15-20 years ago.
We can hope the crew approaches this mountain with respect but there’s little evidence to support it.
The sticks of dynamite might be evidence of motivation...
Thank you for writing this. I also just bought your book to get a better understanding of the complexities of government technology systems and the difficulty of improving them. Your empathy and that of others is inspiring, and stands in such stark contrast to the meanness of President Trump, VP Vance, Musk, and the whole array of cabinet members and Republican "leaders". But empathy is also PRACTICAL as you show here and in your book. That's what we don't see here, which convinces me that this is not reform but an outright steal by thugs.
It's worth noting that COBOL, for all its faults, is fast as hell when running on IBM mainframes. This matters for high-volume real time systems, like IRS processing tax returns, or many bank applications. When the youngsters come in to these environments with promises to recast old code into newer, more flexible and maintainable languages, the resulting code is often just too slow for the job. There's a reason lots of critical systems in the private sector are still running in COBOL, and that obviously can't be attributed to personnel policies or subpar IT strategies at the federal level. Likewise, the fastest code for massive microsimulation modeling is FORTRAN, not Python or Java or C++. It's not even close.
One of my friends is a SWE in civil engineering. According to him, a piece of commonly used software was made in the 90s by a German construction worker, who had no background in programming. Apparently you can see the evolution of his knowledge in the source code: he learns about object-oriented programming, then for awhile he experiments with functional programming before giving up on it, and so forth. So maybe there are a few Jeds outside of government, too! Unsung heroes.
I worked directly with the person you named on this exact modernization project for many months.
I write the following with immense respect for your extraordinary work, Jen:
- I'd like to know if he consented to his story appearing here. At a time when public servants are being targeted for dismissal, bullying, and harassment, using his name and agency affiliation may bring unwanted attention his way, especially as DOGE is working through the VA and Trump admin-adjacent foundations are creating watchlists of current and ex-feds. And if he indeed consented for this - more power to him, wishing him a well-deserved retirement!
- This is the time for tangibly defending public servants from DOGE's search-and-destroy-then-plunder mission. I hope you agree that we need to protect the identities of current and recent government employees in harm's way - unless they've consented to being publicly named at this moment.
- We need to prevent the interlopers from finding more linchpin systems and public servants to destroy and exploit. Empathy building exercises will not stop committed arsonists - as I wrote this comment, the actual DOGE staffer at Treasury just resigned after a Wall Street Journal reporter surfaced his history of deeply racist posts (railing against the Civil Rights Act and interracial marriage, to name a few examples).
I'd like to believe that the college kids that Elon Musk has empowered to vandalize the federal government are capable of the kind of introspection that Jen describes here.
Unfortunately, the early evidence is that they are MAGA'd up, overgrown teens who are hopped up on overconfidence and subsisting on a diet of Nazi memes and potentially a dose of the boss' ketamine.