“Curious where your infrastructure dollars are being spent? Today, accessing that information may require combing through obscure PDFs and press releases or submitting a FOIA request”
This hasn’t been true for a long time. USASpending.gov is, what, fifteen years old? And the FAADS and FPDS datasets it sits atop are older still. APIs, full text search, bulk downloads—it’s got all that stuff. If you can’t find where your tax dollars are going it is usually because the program you’re looking at makes block grants to states who have their own award systems. Alas, it will take more than four people to change that.
We worked on these kinds of things at the Sunlight Foundation and I watched with pride as colleagues went on to USDS, 18f and a host of agencies. They’ve done great work! But the problems are complicated. When I began working on these problems I couldn’t quite believe they weren’t solely due to a lack of will, talent, smarts, ambition. But I learned.
I mean this genuinely: it is always good to have more smart, driven people arriving to work on these problems. They are needed. But this play has been staged before.
(Re the blockchain thing—I am not a crypto basher but cmon, the whole point is trust in the absence of authority. Why would the world’s most powerful government authority need that?)
I shouldn't have used that as an example because, you're right, it's easily read as redundant with USAspending, but also because it doesn't capture the unique things you could do with spending on chain. That said, USAspending isn't real time and is notorious for missing, incomplete and wrong coded data. According to GAO, “one OIG reported that its agency’s submission did not include almost $10 billion of award data.”
So I asked o1 for a list of features an internal govt blockchain could enable above and beyond what existing open data projects have done:
Immutable, End-to-End Transaction Histories A blockchain ledger could create a permanent, tamper-proof record of each government financial transaction. Once entered, the record cannot be altered without leaving evidence. This immutability would make it harder to manipulate or retroactively “clean up” spending, improving transparency beyond what is currently published in tools like usaspending.gov.
Real-Time Updates and Unified Data Rather than quarterly or monthly data releases, a blockchain-based system could surface spending transactions in near-real time. The blockchain would unify data from multiple agencies into a single, standardized data layer, reducing mismatched or out-of-date information that arises from legacy systems.
Automated Auditability When transactions—like grants, contracts, or budget allocations—are added to the blockchain, embedded smart contracts could ensure each disbursement meets predefined criteria (such as compliance checks, budget caps, or milestone requirements). Auditors and inspectors general could then use standardized queries on the chain to follow the entire chain of custody for funds, potentially cutting down on the overhead of audits.
Granular Line-Item Visibility A blockchain could log not only broad categories (e.g., a contract to a vendor) but also every incremental “sub-transaction” (e.g., a subcontractor’s purchase of raw materials). Because every node sees the same ledger, the final paying agency can verify that spending aligns with the intended line items, providing more detail than is typically captured by existing systems.
Efficient Cross-Agency Coordination When multiple agencies co-fund programs, or when money flows from a federal department to states to local entities, a blockchain could automatically record each step of the process. This single source of truth would reduce duplication—currently, data is often siloed across numerous reporting systems. A government-run ledger could simplify data reconciliation and shed light on multi-agency initiatives in a unified manner.
Programmatic Verification and Alerts Smart contracts on the chain could automatically check key compliance triggers. For instance, if an agency exceeds a funding threshold, or if a contractor fails to report expenditures within a given timeframe, alerts could be raised immediately. This proactive mechanism enhances oversight and ensures faster responses to irregularities.
Public Trust Through Selective Transparency While certain sensitive data might remain private or partially encrypted on a permissioned government blockchain, portions of the ledger could be made publicly accessible in near real time. The public could then verify aggregated spending figures without relying on post-processed, potentially error-prone releases. This dual-layer approach offers significantly more insight than static reports on usaspending.gov without compromising security needs.
By integrating these features into an internal government blockchain, agencies could achieve a more trustworthy, detailed, and up-to-date record of federal spending than is possible with current systems.
I appreciate the response! Let me offer my own acknowledgement of my argument’s limits: usaspending does exist but you’re not wrong about the need to hunt for PDFs for usable detail. The data exists in parallel, but it’s often the case that the version normalized into a pan-government reporting standard less useful. I suspect this is an unavoidable byproduct of how Congress legislates and larger, Seeing Like A State-ish forces. AI could unlock this problem in meaningfully new ways.
Blockchain though… I am not sure how compelling o1’s ideas on this are. “A regular database can do this better” is a bit of a cliched answer to blockchain enthusiasm, but it’s still true here (with the exception of a hypothetical conspiracy to quietly amend spending data—big if true!). Nothing about normalizing different agencies’ (and contractors’) spending into a single data model is made easier by putting it on chain. I suppose it might be nice to deploy to a Solana contract rather than paying Booz to run an API for you, but then again you’ll be paying gas fees to a bunch of Chinese miners to run your infrastructure instead. Good grist for DC policy roundtables!
In my experience real time reporting is not as practically useful a window into the appropriation/obligation/disbursement process as I once imagined but perhaps others will find new angles. I feel more sure that “transparency creates trust” is wrong—one of the harder lessons I learned from my time in this field.
At any rate I remain skeptical of blockchain utility in this space. I was a booster when the World Food Program was an early adopter back in the day. But the benefits never materialized—it turned out to just be a way for technical staff to look like they were doing something interesting. There are many parts of the new administration’s policy pushes I could be surprised by, but the crypto stuff seems extremely legible—we’ve been at this for over a decade! At some point “Gensler was bad” will cease to be a sufficient rationale and it will become apparent that a tremendous amount of this is driven by venality and that many policies—like the ban on a central bank coin—are sacrificing US national interest in (imo) fairly outrageous ways.
I suspect the technology is needed less than the favor of taking the meeting with a blockchain company recommended by someone important (as far as I can tell this is all just overexcitement from reporting on one meeting, no? And it’s not like Elon has decided blockchain is important enough to use at any of his companies).
Perfectly understandable if so! But not exactly a break from business as usual.
I worked for and with Uncle Sugar's DoD for 25 years and my Dad worked for Social Sec Admin for like 35. The key is complexity. SocSec is actually very efficient. They take money in on a formulaic basis and send it out thusly. Simple and effective. If you don't like how they give or take money, take it up with your congressman.
But, we the taxpayers, put extra layers of expectations on the govt for everything else. The number one thing is that we want fairness and objectivity in decision making and contracting. This is because 100 years ago govt officials were just hiring or buying from personal contacts. Every time a program or purchase goes badly or there's a dispute there's the NYTimes or John Stossel exposing it as an affront to John Q Public. So, the bureaucrats try to beef up the process and the GAO says "more oversight". After a few rounds of this, you end up with a process like DoD 5000 (google "dod 5000 wall chart") and all of the govt employees are trying to just run a gauntlet of approval steps and reports and everythings slows down and gets mroe expensive. The actual product becomes secondary to the process!
Nevertheless, I am optimistic. I think that AI can lend some repeatable validity, to govt decisions while also being quicker. I also don't think that efficiency gains will necessarily come at the head of an axe. If things are the way they are, it is because of a web of laws and regulations. DOGE should focus on simplifying necessary rules and telling the public that 10% of your money will be wasted just because that's the cost of doing business with imperfect information.
The essay skillfully intertwines historical context, theoretical frameworks, and practical examples to argue for a tech-driven transformation of government. While the potential benefits are immense, the real-world implementation of DOGE’s vision will require navigating significant technical, political, and ethical hurdles. Nonetheless, the urgency of the challenges posed by emerging technologies makes this a timely and necessary endeavor.
Chatgpt think your essay is skillful,if that makes you feel better
I hope for great success. The DOGE can't do anything about the problem that people tend to be stupid about money, and EVEN WORSE if government money is involved, though.
“…federal government’s discretionary budget is simply too meagre relative to defense and entitlement spending”
Sorry, while the claim relative to entitlement spending is correct, the claim relative to defense spending is not. Right now non-defense discretionary spending and defense spending are roughly equal. The growth in non-defense discretionary under Obama was massive. Growth in both were massive under Biden.
And there’s no guarantee DOGE can’t impact defense spending as well, though I grant that it is far more likely to impact non-defense discretionary spending.
Yes entitlements are the biggest issue. And imo for the economy and long-term deficits, reducing regulation - Vivek’s focus - would have an even bigger positive impact.
But we should not let the perfect/best be the enemy of the good. Eliminating a ton of wasteful spending - much of which has built up only in the last 16 years (12 of which very liberal and/or hardcore leftists ran the administration) is a very good thing. Full stop.
Go look at the budget increases. Green new deal stuff, etc.
Huge budget increase in Great Financial Crisis under Obama. Most to leftist causes. (Rahm Emmanuel: “never let a good crisis go to waste”). Never went back down.
Huge increase under COVID. Most of it not taken away.
Thanks Andy. Thought maybe there was some specific data that was informing your opinion that non-defense discretionary spending exploded under Biden and Obama and that it was wasted on leftist causes; good to know it was actually just your reading the newspaper and watching the news (I'm guessing Fox).
Here is the actual data over time if you ever want to actually become informed:
The reality is that the Obama years did see a huge stimulus package in 2009-2010 to get the economy out of the Bush recession (the biggest, at least at the time, since the Great Depression) but was immediately reined in after that. And if you go through where that stimulus spending went (https://en.wikipedia.org/wiki/American_Recovery_and_Reinvestment_Act_of_2009#), it would be interesting to know what you consider leftist and wasted?
And then in 2021, the Biden Adminstration spearheaded a huge increase in spending to help the economy recover from the biggest recession since the Great Depression: the Trump/COVID bust. And while this spending likely played a role in the inflation which torpedoed his reelection hopes (though it's important to note inflation was even bigger in most countries that didn't pass large stimulus packages), it also helped leave America with by far the best economy in the world:
Good to see that with your snide remarks re Fox News - which I don’t watch - that you are explicitly signaling very clearly up front that you are an out-of-the-closet leftist. I appreciate it.
Citing the left-biased Wikipedia as a source of unbiased information is fairly laughable, or are you unaware of this well known bias (which would not surprise me, given your claim that I must be getting my information from Fox News…)?
Your claim that the 2009-2010 stimulus spending was “immediately reined in” is simply false. It is true that some fraction of it was cut back (forced on Obama by the GOP majorities in Congress from the 2010 tea party elections).
America had, and would have had, the best economy in the world regardless, so we do agree on that.
But be very clear, I am NOT actually here criticizing the temporary spending of 2009-2010, or of 2020 and 2021 - some of which I agree was necessary/wise. I am criticizing the fact that most of the increased spending did not in fact go away, and I am also criticizing what much of that increase was spent on. Most of your critique of/response to my comment was utterly non-responsive to my actual claim.
There is plenty of green new deal-like crap funding in Biden’s American Rescue Plan and the Orwellian-named Inflation Reduction Act, and in Obama’s American Recovery and Reinvestment Act.
There is tons of money that goes to “green” companies.
There is plenty of money that goes to “community organizer” groups, and has now since Obama.
You can deny the truth, but it doesn’t change the truth.
Note I clearly said “much of” the increased spending falls in this category; I did not claim it was most or even a majority.
I am someone who wishes DOGE well in its mission to cut the suffocating bureacracy. But I do wonder how it is going to get round the problem of human nature. If you don't know what I mean, look at this wonderful cartoon in the Daily Telegraph on 14 Jan: https://www.instagram.com/p/DExiPKNyczp/
Good article, and, until the end, I was onboard with the arguments for digitizing government a la Estonia. But I'm not clear how the possibility of billions of hostile superintelligent agents strengthens the case for "allowing government APIs to integrate with things like e-banking, business registries, electronic health records, child support systems and more." Hostile superintellent agents sounds more like a risk of digitizing government rather than a benefit of it. And that might suggest adding cybersecurity to the DOGE mandate.
You added the word “hostile” not me. I'm refering to the scalability problem posed by AGIs overwhelming institutions that aren't equipped to handle billions of additional agents interacting with them.
“Curious where your infrastructure dollars are being spent? Today, accessing that information may require combing through obscure PDFs and press releases or submitting a FOIA request”
This hasn’t been true for a long time. USASpending.gov is, what, fifteen years old? And the FAADS and FPDS datasets it sits atop are older still. APIs, full text search, bulk downloads—it’s got all that stuff. If you can’t find where your tax dollars are going it is usually because the program you’re looking at makes block grants to states who have their own award systems. Alas, it will take more than four people to change that.
We worked on these kinds of things at the Sunlight Foundation and I watched with pride as colleagues went on to USDS, 18f and a host of agencies. They’ve done great work! But the problems are complicated. When I began working on these problems I couldn’t quite believe they weren’t solely due to a lack of will, talent, smarts, ambition. But I learned.
I mean this genuinely: it is always good to have more smart, driven people arriving to work on these problems. They are needed. But this play has been staged before.
(Re the blockchain thing—I am not a crypto basher but cmon, the whole point is trust in the absence of authority. Why would the world’s most powerful government authority need that?)
I shouldn't have used that as an example because, you're right, it's easily read as redundant with USAspending, but also because it doesn't capture the unique things you could do with spending on chain. That said, USAspending isn't real time and is notorious for missing, incomplete and wrong coded data. According to GAO, “one OIG reported that its agency’s submission did not include almost $10 billion of award data.”
So I asked o1 for a list of features an internal govt blockchain could enable above and beyond what existing open data projects have done:
Immutable, End-to-End Transaction Histories A blockchain ledger could create a permanent, tamper-proof record of each government financial transaction. Once entered, the record cannot be altered without leaving evidence. This immutability would make it harder to manipulate or retroactively “clean up” spending, improving transparency beyond what is currently published in tools like usaspending.gov.
Real-Time Updates and Unified Data Rather than quarterly or monthly data releases, a blockchain-based system could surface spending transactions in near-real time. The blockchain would unify data from multiple agencies into a single, standardized data layer, reducing mismatched or out-of-date information that arises from legacy systems.
Automated Auditability When transactions—like grants, contracts, or budget allocations—are added to the blockchain, embedded smart contracts could ensure each disbursement meets predefined criteria (such as compliance checks, budget caps, or milestone requirements). Auditors and inspectors general could then use standardized queries on the chain to follow the entire chain of custody for funds, potentially cutting down on the overhead of audits.
Granular Line-Item Visibility A blockchain could log not only broad categories (e.g., a contract to a vendor) but also every incremental “sub-transaction” (e.g., a subcontractor’s purchase of raw materials). Because every node sees the same ledger, the final paying agency can verify that spending aligns with the intended line items, providing more detail than is typically captured by existing systems.
Efficient Cross-Agency Coordination When multiple agencies co-fund programs, or when money flows from a federal department to states to local entities, a blockchain could automatically record each step of the process. This single source of truth would reduce duplication—currently, data is often siloed across numerous reporting systems. A government-run ledger could simplify data reconciliation and shed light on multi-agency initiatives in a unified manner.
Programmatic Verification and Alerts Smart contracts on the chain could automatically check key compliance triggers. For instance, if an agency exceeds a funding threshold, or if a contractor fails to report expenditures within a given timeframe, alerts could be raised immediately. This proactive mechanism enhances oversight and ensures faster responses to irregularities.
Public Trust Through Selective Transparency While certain sensitive data might remain private or partially encrypted on a permissioned government blockchain, portions of the ledger could be made publicly accessible in near real time. The public could then verify aggregated spending figures without relying on post-processed, potentially error-prone releases. This dual-layer approach offers significantly more insight than static reports on usaspending.gov without compromising security needs.
By integrating these features into an internal government blockchain, agencies could achieve a more trustworthy, detailed, and up-to-date record of federal spending than is possible with current systems.
Thanks ChatGPT!
I appreciate the response! Let me offer my own acknowledgement of my argument’s limits: usaspending does exist but you’re not wrong about the need to hunt for PDFs for usable detail. The data exists in parallel, but it’s often the case that the version normalized into a pan-government reporting standard less useful. I suspect this is an unavoidable byproduct of how Congress legislates and larger, Seeing Like A State-ish forces. AI could unlock this problem in meaningfully new ways.
Blockchain though… I am not sure how compelling o1’s ideas on this are. “A regular database can do this better” is a bit of a cliched answer to blockchain enthusiasm, but it’s still true here (with the exception of a hypothetical conspiracy to quietly amend spending data—big if true!). Nothing about normalizing different agencies’ (and contractors’) spending into a single data model is made easier by putting it on chain. I suppose it might be nice to deploy to a Solana contract rather than paying Booz to run an API for you, but then again you’ll be paying gas fees to a bunch of Chinese miners to run your infrastructure instead. Good grist for DC policy roundtables!
In my experience real time reporting is not as practically useful a window into the appropriation/obligation/disbursement process as I once imagined but perhaps others will find new angles. I feel more sure that “transparency creates trust” is wrong—one of the harder lessons I learned from my time in this field.
At any rate I remain skeptical of blockchain utility in this space. I was a booster when the World Food Program was an early adopter back in the day. But the benefits never materialized—it turned out to just be a way for technical staff to look like they were doing something interesting. There are many parts of the new administration’s policy pushes I could be surprised by, but the crypto stuff seems extremely legible—we’ve been at this for over a decade! At some point “Gensler was bad” will cease to be a sufficient rationale and it will become apparent that a tremendous amount of this is driven by venality and that many policies—like the ban on a central bank coin—are sacrificing US national interest in (imo) fairly outrageous ways.
Came to comment and make this exact same point. Why in the world would a blockchain be needed for this.
And I AM a crypto basher. As usual, it's a solution in search of a problem.
I suspect the technology is needed less than the favor of taking the meeting with a blockchain company recommended by someone important (as far as I can tell this is all just overexcitement from reporting on one meeting, no? And it’s not like Elon has decided blockchain is important enough to use at any of his companies).
Perfectly understandable if so! But not exactly a break from business as usual.
I worked for and with Uncle Sugar's DoD for 25 years and my Dad worked for Social Sec Admin for like 35. The key is complexity. SocSec is actually very efficient. They take money in on a formulaic basis and send it out thusly. Simple and effective. If you don't like how they give or take money, take it up with your congressman.
But, we the taxpayers, put extra layers of expectations on the govt for everything else. The number one thing is that we want fairness and objectivity in decision making and contracting. This is because 100 years ago govt officials were just hiring or buying from personal contacts. Every time a program or purchase goes badly or there's a dispute there's the NYTimes or John Stossel exposing it as an affront to John Q Public. So, the bureaucrats try to beef up the process and the GAO says "more oversight". After a few rounds of this, you end up with a process like DoD 5000 (google "dod 5000 wall chart") and all of the govt employees are trying to just run a gauntlet of approval steps and reports and everythings slows down and gets mroe expensive. The actual product becomes secondary to the process!
Nevertheless, I am optimistic. I think that AI can lend some repeatable validity, to govt decisions while also being quicker. I also don't think that efficiency gains will necessarily come at the head of an axe. If things are the way they are, it is because of a web of laws and regulations. DOGE should focus on simplifying necessary rules and telling the public that 10% of your money will be wasted just because that's the cost of doing business with imperfect information.
The essay skillfully intertwines historical context, theoretical frameworks, and practical examples to argue for a tech-driven transformation of government. While the potential benefits are immense, the real-world implementation of DOGE’s vision will require navigating significant technical, political, and ethical hurdles. Nonetheless, the urgency of the challenges posed by emerging technologies makes this a timely and necessary endeavor.
Chatgpt think your essay is skillful,if that makes you feel better
I hope for great success. The DOGE can't do anything about the problem that people tend to be stupid about money, and EVEN WORSE if government money is involved, though.
“…federal government’s discretionary budget is simply too meagre relative to defense and entitlement spending”
Sorry, while the claim relative to entitlement spending is correct, the claim relative to defense spending is not. Right now non-defense discretionary spending and defense spending are roughly equal. The growth in non-defense discretionary under Obama was massive. Growth in both were massive under Biden.
And there’s no guarantee DOGE can’t impact defense spending as well, though I grant that it is far more likely to impact non-defense discretionary spending.
Yes entitlements are the biggest issue. And imo for the economy and long-term deficits, reducing regulation - Vivek’s focus - would have an even bigger positive impact.
But we should not let the perfect/best be the enemy of the good. Eliminating a ton of wasteful spending - much of which has built up only in the last 16 years (12 of which very liberal and/or hardcore leftists ran the administration) is a very good thing. Full stop.
"Eliminating a ton of wasteful spending - much of which has built up only in the last 16 years"
Can you point to the sources that informed this comment. That isn't my understanding so I'd like to see why you think it's true.
I’ve followed the news the last 20 years.
Go look at the budget increases. Green new deal stuff, etc.
Huge budget increase in Great Financial Crisis under Obama. Most to leftist causes. (Rahm Emmanuel: “never let a good crisis go to waste”). Never went back down.
Huge increase under COVID. Most of it not taken away.
it’s not rocket science.
Thanks Andy. Thought maybe there was some specific data that was informing your opinion that non-defense discretionary spending exploded under Biden and Obama and that it was wasted on leftist causes; good to know it was actually just your reading the newspaper and watching the news (I'm guessing Fox).
Here is the actual data over time if you ever want to actually become informed:
https://www.cbo.gov/content/discretionary-spending-options
https://taxfoundation.org/blog/federal-budget-deficit-2023/
The reality is that the Obama years did see a huge stimulus package in 2009-2010 to get the economy out of the Bush recession (the biggest, at least at the time, since the Great Depression) but was immediately reined in after that. And if you go through where that stimulus spending went (https://en.wikipedia.org/wiki/American_Recovery_and_Reinvestment_Act_of_2009#), it would be interesting to know what you consider leftist and wasted?
And then in 2021, the Biden Adminstration spearheaded a huge increase in spending to help the economy recover from the biggest recession since the Great Depression: the Trump/COVID bust. And while this spending likely played a role in the inflation which torpedoed his reelection hopes (though it's important to note inflation was even bigger in most countries that didn't pass large stimulus packages), it also helped leave America with by far the best economy in the world:
https://www.economist.com/special-report/2024-10-19
Good to see that with your snide remarks re Fox News - which I don’t watch - that you are explicitly signaling very clearly up front that you are an out-of-the-closet leftist. I appreciate it.
Citing the left-biased Wikipedia as a source of unbiased information is fairly laughable, or are you unaware of this well known bias (which would not surprise me, given your claim that I must be getting my information from Fox News…)?
Your claim that the 2009-2010 stimulus spending was “immediately reined in” is simply false. It is true that some fraction of it was cut back (forced on Obama by the GOP majorities in Congress from the 2010 tea party elections).
America had, and would have had, the best economy in the world regardless, so we do agree on that.
But be very clear, I am NOT actually here criticizing the temporary spending of 2009-2010, or of 2020 and 2021 - some of which I agree was necessary/wise. I am criticizing the fact that most of the increased spending did not in fact go away, and I am also criticizing what much of that increase was spent on. Most of your critique of/response to my comment was utterly non-responsive to my actual claim.
The only specific program you voted was not passed into law
🙄
There is plenty of green new deal-like crap funding in Biden’s American Rescue Plan and the Orwellian-named Inflation Reduction Act, and in Obama’s American Recovery and Reinvestment Act.
There is tons of money that goes to “green” companies.
There is plenty of money that goes to “community organizer” groups, and has now since Obama.
You can deny the truth, but it doesn’t change the truth.
Note I clearly said “much of” the increased spending falls in this category; I did not claim it was most or even a majority.
Hi Sam. Love to connect. How best to do. Thanks
I am someone who wishes DOGE well in its mission to cut the suffocating bureacracy. But I do wonder how it is going to get round the problem of human nature. If you don't know what I mean, look at this wonderful cartoon in the Daily Telegraph on 14 Jan: https://www.instagram.com/p/DExiPKNyczp/
Good article, and, until the end, I was onboard with the arguments for digitizing government a la Estonia. But I'm not clear how the possibility of billions of hostile superintelligent agents strengthens the case for "allowing government APIs to integrate with things like e-banking, business registries, electronic health records, child support systems and more." Hostile superintellent agents sounds more like a risk of digitizing government rather than a benefit of it. And that might suggest adding cybersecurity to the DOGE mandate.
You added the word “hostile” not me. I'm refering to the scalability problem posed by AGIs overwhelming institutions that aren't equipped to handle billions of additional agents interacting with them.