IMO this is indicative of the kind of questions anyone with DOGE on their resume will need to answer at their next interview.

- why did you do it?

- why did you not quit as soon as you realized how dangerous this was going to be?

"I did what I did, I knew there were mistakes, I don't really know what was done with the output of my work and I never tried to find out" are absolutely disqualifying statements for anyone above associate level engineering.

Incompetent and destructive as they are, DOGE only did what every company and startup in the world is trying to do, aggressively integrate AI in a way that radically upsets the societal paradigm and makes a ton of money. And much of tech buys into the politics and narrative. They're rock stars. They'll never even need to interview to find work, work will come to them.

> And much of tech buys into the politics and narrative.

While I agree I think that's a fatalist perspective. It's on all of us working in tech to push back on that. In the unlikely event that someone with DOGE on their resume comes across my desk I certainly will be.

And if folks are able to clear their mind of ideology, this is an account of someone that was straightforwardly bad at their job. It's rare to get such a confession from an applicant, we should use it!

zero chance that "all of us working in tech" will push anything.. worse, aggressive opportunists know this very well

Again, back to that being a fatalist perspective. It won’t work so there’s no point in trying. IMO it’s always worth trying.

The opportunists are a minority and they should be ostracized. They’ll always hire each other of course but the reality, as we’re seeing, is that they aren’t actually very good at what they do. For years Musk was seen as an Iron Man genius but by any metric you choose DOGE has been an absolute failure.

Ego > engineering.

I’d do it, gladly.

I worked on the Obamacare site after it rolled out, crashing. I gave it my best effort.

I’m in favor of progress and will work for the common good. I think that’s the best way.

> disqualifying statements for anyone above associate level engineering.

Only if you're an engineer. The Silicon Valley folks love this sort of lawless disregard for wider society.

I'm not even talking about the wider societal implications. I mean the job of being a software engineer, including in SV. "I knowingly wrote code riddled with mistakes and didn't ever look into how it's being used" is, IMO, disqualifying. The latter more than the former in many ways, being a good engineer above entry level is a lot more than just bashing out code, you need to be understanding the context in which your code operates.

It would be one thing if we were talking about a fresh out of college newbie. But Lavingia was employee #2 at Pinterest and founded Gumroad. It appears he learned absolutely nothing in either place.

Let me clarify. I agree with you. My point is rather that Silicon Valley bros WONT agree with us. To them it's fine actually if some veterans didn't get healthcare and die, because they don't matter in their grand scheme.

I believe, as a ground truth, that good engineering requires working in the real world. Silicon valley guys do not. They believe their responsibility start and end at the computer.

I understand this sounds extreme, I think so too, but it is the only explanation that explains their fascination with "long-termism". A philosophy that any harm now is justified if it has even a .1% chance of doing something very good in the long term.

> A philosophy that any harm now is justified if it has even a .1% chance of doing something very good in the long term.

Even then, our descendants, who'd finally have the privilege to live in that faraway future that we sacrificed everything in the present for - they'd also be obligated to treat their present as a disposable resource in service of an even further away future.

If you took the "philosophy" at face value, it would actually guarantee that the entire future of humanity will be miserable. But we'll colonize space and number in the trillions. Yay...

I live in Silicon Valley and I don't love lawless disregard for wider society.

"I needed money." Why is that not a good enough answer?

"I needed money." - Charles Ponzi "I needed money." - Bernie Madoff "I needed money." - Kenneth Lay "I needed money." - Qusay Hussein

Because he got paid $0. They hired him as an “intern” or volunteer.

It's responsibility-washing. Like GP said, those answers are totally unacceptable coming from an engineering professional. Which is exactly why they took on interns to do it.

Ah, then we got what we paid for.

Well:

(a) It doesn't appear to actually have come with a salary.

(b) Even if it paid, it slightly stretches credulity given his background that it was the only job that he could get.

They will probably get a favorable role at a thiel or yc associated company

Employers inclined to ask those questions will more likely just toss any resumes mentioning DOGE at an earlier stage of filtering.

The only silver lining here is his transparency about being wrong -- and providing evidence of that publicly. Given DOGE's culture of prioritizing avoiding accountability and transparency measures like FOIA [1], it's almost sadly laudable.

[1] https://www.cjr.org/feature-2/trump-100-legal-battle-for-dog...

But he still doesn't seem to get the seriousness of the situation, at all. Like, he's saying no-one should have used his stuff, but also "It would have been cool if the entire federal government used this script".

For instance, imagine you have surgery, and die of a complication. Your relatives will feel very differently, and be in a very different legal situation, if the hospital's explanation is "surgeon was the bottom of their class" vs "yeah, we just handed a visiting 5 year old a scalpel to see what would happen". Neither are ideal, obviously, but the latter is _criminal_.

The idiot in the article is clearly closer to the 5 year old than the scraping-a-pass surgeon; he is simply not in any way a proper person to have been given the sort of access or authority he was, this would have been obvious to whoever gave it to him, and whoever gave it to him is at least morally culpable.

I mean, hindsight is 20/20, but that's probably why it's a bad idea to give a gaggle-load of inexperienced tech sisses/bros the ability to do stuff like this.

Thinking back on it, a disaster should have been the expected outcome. Drawdowns of the federal government should be measured and deliberate and done by grey haired people not given to rash action. Not by a kid with a python script he thinks is kind of cool.

A disaster was the expected outcome to anyone paying attention. We were screaming to the rooftops that this was going to end in disaster! Nobody should be playing the "shucks nobody realized this was going to be bad" card

Exactly! Many people on HN and elsewhere (including myself) could see that this was never a serious effort to make the government more efficient and called it out from the start. The evidence was overwhelming.

It was dismaying how many people took the Doge claims at face value, hopefully at least a few people will remember the chain of events and be more skeptical of these actors in the future.

If anyone thinks drawdowns should be done of the US federal government they aren't anywhere near the levers of government. There hasn't been one of those in the best part of a century as I recall.

> Thinking back on it, a disaster should have been the expected outcome.

I doubt anyone with decision making power is surprised - the flow of events has been both predictable and predicted. Whatever the goals with DOGE were it wasn't spending cuts or finding efficiencies. The first is in opposition to the preferences of the US Republicans as expressed in their actual voting patterns and the second is largely irrelevant and they weren't set up to do that.

> There hasn't been one of those in the best part of a century as I recall.

That’s just saying your media diet doesn’t cover it. Agencies adjust all the time (every year has a new budget and priorities from Congress), and there was a big Clinton effort which cut hundreds of thousands of jobs and restructured agencies.

https://en.wikipedia.org/wiki/National_Partnership_for_Reinv...

That tends not to get mentioned because it wasn’t disruptive – they actually took time to plan it – and because it happened under a Democratic president, the Republicans aren’t going to acknowledge it.

And it had no impact [0]. Increases in US Federal spending look like a jump from 5% to 20% of GDP. "Cuts" look like a flat line at 20%. Although I don't know what the situation might have changed post-COVID (higher again).

If we're defining "drawdowns" to mean "hold steady" then sure, lots of those. If it means drawing spending down then there technically was one in the 1940s. The proper terms for what we're seeing are "reallocation" or "expansion" depending on the congress.

[0] https://en.wikipedia.org/wiki/Government_spending_in_the_Uni... / https://en.wikipedia.org/wiki/Government_spending_in_the_Uni...

Take a closer look at that chart. Notice the pronounced downward trend on the blue line (federal) starting from a peak in 1992 and declining until the Bush administration unbalanced the budget? In 1998-2001, we had a budget surplus!

https://www.cbo.gov/sites/default/files/cbofiles/images/pubs... https://www.cbo.gov/publication/61181

The big thing, of course, is that total spending is a crude metric lumping discretionary and mandatory spending together. Federal employment has been getting a lot of attention this year, but it's at historic lows and there's no real way to significantly lower total federal spending there without cutting social programs millions of people depend on.

> Take a closer look at that chart. Notice the pronounced downward trend on the blue line (federal) starting from a peak in 1992...

So? That is noise on a flat line. A drawdown doesn't mean sticking to the consensus flat trend. The plan here seems to be a "drawdown" is something where I can say "spending will be about the same".

If I talk about increasing government spending I can point to spending growing from 3% -> 20% GDP with a really clear trend and level change. When we talk about drawdowns it is supposed to mean maintaining a trend at 20%? That isn't a drawdown. If that is a drawdowne then DOGE may turn out to have successfully drawn down US government spending! Even though spending seems to be going up.

If you’re arguing that going from 10% to 6% is flat, it sounds like you’re starting from a predefined conclusion and adjusting terminology to match.

Says the person whos missed most of the spending - US Federal spending hasn't dropped below 10% in 80 years. You might not have read the title of the graph you're looking at?

US Federal spending over any recent time period is basically sum(GDP over however many years)*20%. You're talking about a fraction of a fraction of a fraction that, which likely at no point involved actually lowering spending since we're talking a % of something that is growing.

Y'know, well done Clinton and all that - balancing a budget is challenge. But if that is going to get called a drawdown it was about as effective as DOGE. It didn't bump Federal spending off the trendline and the small amounts render it largely meaningless in the big picture. It is literally statistical noise.

A lot of us knew it was going to be a disaster before they started. There’s a reason bureaucrats do things in a measured way. So to say “in hindsight” belies the point.

This disaster was exactly predicted by a ton of people, with foresight! To treat this as an unexpected outcome belies the exact lack of seriousness that characterized this whole ordeal

DOGE isn't SpaceX where getting it wrong is a chance to learn for the next iteration. These are human lives being affected by all of these mistakes that they are so nonchalantly making as if the humans are just NPCs in whatever game they are playing. Not only are these mistakes directly affecting humans with the loss of government jobs, they are also affecting citizens that depend on the services these employees worked. I don't care about your silver lining. The damage is done, and it is not easily unwound. All because they had no shits to give and too young to have the life experience to even come close to understanding just how damaging their decisions were.

@sahillavingia is a member and wrote about his DOGE experience recently: https://news.ycombinator.com/item?id=44116061

> did he just think LLMs were _magic_?

If you happened to follow this particular developer, yes he does think they're magic.

It doesn't even take very long messing around with the things to quickly realize how much smoke and mirrors is involved. Next time you get a good response from your favorite LLM, go ahead, hit "regenerate" a few times. Look at how different the responses can be, from the same initial prompt, with just a liiiiitle push in the RNG weights. Sometimes it's remarkably correct! Sometimes it's remarkably *incorrect* and confident about it. Which one did you get this time?

But, like, in this case, _even if the LLM was a superhuman AGI_, there is _still_ no way it could do what he wants based off the prompt he gave it because it simply would not have the sort of in-depth knowledge of the inner workings of the VA required to make these determinations.

Even if we waive the "LLMs are bit shit" stipulation, it still doesn't really help much.

To be fair, a superhuman AGI should've told him that what he's trying to do is impossible.

(Not that an LLM will ever get there though.)

... and get paid to do it while sidelining anyone with actual experience in the sector/topic.

I called this when it started. They would come in, be totally ineffective to the point of breaking things, and then we would get a "shucks, sorry" later.

He didn’t get paid.

Azure OpenAI absolutely got paid.

I just can't believe how totally incompetent this whole thing is. Even for a Musk/Trump venture, I didn't expect quite this level of just complete nonsense.

Have you not followed the Twitter acquisition saga and the firing process that followed? It was the same thing

I remember the follow up. That Twitter was doomed and Threads, then Bluesky were absolutely going to crush it.

Those were not only the popular takes here; but any claim otherwise was derided as harmful fiction.

And still is.

But with such fiction for a memory would you even believe this whole thing is totally incompetent?

thats the only reasonable way to do business. Its not a grocery store. if you don't get enough money, you need to cut your fixed costs. So he did. After buying, a selection of highly competent individuals knowing their systems decided on the (future) tech-stack and on that what is not needed and what can be done recoded/rebuild to save money on tech-stack even more. And then you just don't need that much work force anymore.

rule no. 1 in business: If you pay, you decide.

rule no.2 in business: If you pay more than you earn: close the company.

key-insights:

- it wasn't musk alone to decide. He had twitter's qualified employees and system architects -> pick the most skilled ones

- it was a power meeting where the groups & members brainstormed excessively for longer then just 3 days in a row -> let them think about a better, simplified system design on a minimal viable tech-stack

- they were accompanied by qualified stuff from tesla to uncover blind spots in the design and to bring fresh ideas / integration-knowledge / synergies & work-force -> check & start building. pivot if neccessary

I really don't undersand, why arguments like yours are still & always brought up. Is it the lack of understanding how to do business? Is it pure "hate" & ad hominem @ musk, because it's musk? What is it that is strong enough to let one show absolute lack of either economical or technical thinking publicly? The -SANE- economical and technical mindset craves for Optimization of tech and fin. Thats the only differentiator to the competition: do it better, do it cheaper.

So why was it bad exactly?

You'll never understand why such questions are brought up because you're too deep into your own fiction. That's not what happened at all, and your primitive business rules aren't real either.

> if you don't get enough money, you need to cut your fixed costs.

No you don’t (in general), a common business model in tech is to spend investor’s money covering this deficit. Also another harder option is to get more money. Also, it wasn’t just fixed costs. And if you cut too deep and lose a lot of revenue as a result as with Twitter you’ll still not get enough money. "So he did"

Instead of all this "thinking hard about the stack" in reality here is how people got fired:

> it wasn't musk alone to decide

Unless it was, of course

- like Frohnhoefer without notice, without "brainstorming", by Musk alone, for daring to say the "technical stack" truth to ignorant power

- engineers were told today to print out their last 30 to 60 days of code, so they could show it to Elon Musk himself.

- or there was a guy who couldn’t even get info from the dedicated function for such stuff - HR - whether he was fired. That also had to be resolved by Musk in a Twitter exchange

> minimal viable tech-stack

How come you ignore that a huge portion of employees aren’t involved in any tech stack? Think 80% of employees were fired.

> After buying, a selection of highly competent individuals knowing their systems decided on

What’s your proof of competence? Like in this case: on paper, you could call this engineer competent "Lavingia has nearly 15 years of experience as a software engineer and entrepreneur", yet the work tells the opposite.

Or even if they were competent, if you give them just a week they’ll do the same happening in this story.

> - they were accompanied by qualified stuff from tesla to uncover blind spots in the design and to bring fresh ideas / integration-knowledge / synergies & work-force

These are just empty words, care to give specific examples of what fresh ideas/synergies and other work-force positive things this great process resulted in? Doge was also accompanied by "qualified stuff from tesla/twitter/etc", but instead of uncovering blind spots they were mostly acting blind.

> The -SANE- economical and technical mindset craves for Optimization of tech and fin. > So why was it bad exactly?

Because you’re supposed to maximize revenue and profits, not minimize them?

> What is it that is strong enough to let one show absolute lack of either economical or technical thinking publicly?

You tell me, why do you do this?

Sorry, you're wrong in your thinking. You took the time to discuss and now it's my turn to do the same. I'm German and we're direct in our arguments. No intent of offense or to come at ya.

> No you don’t (in general), a common business model in tech is to spend investor’s money covering this deficit.

A common INVESTMENT model is to accept (risk!) money burned for a perspective of (sustainable) returns in future. The dicision on investment is usually driven by figures and CPIs. A common BUSINESS model is to become fin-sustainable as possible, with steady minimal spendings and reoccuring earnings that cover the spendings in the least. This is called by you "primitive business rules". And yes, they are. Thats the most primitive rules.

Investors don't just give money. They get a share of the company and become partial-owner. No investor will give money, if there's no return to expect in a set time horizont. If you want, have some read on that [1].

Musk & Friends are THIS very investrors you're talking about. They invested together into full ownership of 100% shares. They have the solely power to decide. They decided not to dillute their shares by getting others onto the boat - in fact, Twitter has been bought to squeeze out other deciders (taking off public listing). So, what you actually asking investors for is to burn money and to accept the losses resulting by creadit loans and interest fees, because its common business-sense. Interesting economical thinking you have..

> Also, it wasn’t just fixed costs.

Variable costs can't be cut by definition. They can be steered/controlled/altered/reduced, like the monthly electricity bill in winter/summer time. But not cut. Variable costs MAY occure irregularly. Fixed costs WILL occure regularly. The aim is to cut the fixed costs, such as rent, salaries & loans, interest rates. Coleterally, variable costs change. Of course, "it wasn’t just fixed costs" - they are correlated!

> And if you cut too deep and lose a lot of revenue as a result as with Twitter you’ll still not get enough money.

Pardon, what??

profit = earnings - expenses

cutting expenses, while earnings stay unchanged = the profit rises. Full-cut of expenses, while earnings stay unchanged = earnings are the profit. If earnings go down but expenses are not cut' = profit sinks or becomes net negative in case earnings < expenses (loss!). Cutting expenses while earnings go down = profit is proportional to the cuts, but will never turn neg, as long earnings > spendings.

When the revenue goes down one has to cut expenses or increase the revenue by any means. Or both. Thats "what he did" - staff reduced drastically and system redesigned to run unattended, with self-healing-features. That is done, BECAUSE Twitter's yearly revenue only once was enough to cover the expenses in whole time being, without perspectives to change in a mid-distant future. It was a money-sink. So your argument backs my argument ;) Thank you. Even more, the new calculations showed even higher financial needs: amounted and running expenses by ex-Twitter and additionally costs of buying credit loan and its interest rates are to be served now, too. And we're back to:

> Instead of all this "thinking hard about the stack" in reality here is how people got fired: > ....

So, can you suggest a different way? They are fired like all employees in a hire & fire society are likely to be fired. Some without or with a few minutes notice, some days ahead their planed exit, ..., some in a friendly way. Sometimes owner decides alone, sometimes he gets pre-filtered decisions, sometimes the decision is delegated to lower menagers. Absolutely business as usual.

> - like Frohnhoefer without notice, without "brainstorming", by Musk alone, for daring to say the "technical stack" truth to ignorant power

The idea is to have people who truelly belief its possible to achieve results working on something everyone denotes as "impossible". Frohnhoefer did not align with that vision.

I'm in Germany and was "fired" like that: "sorry, do you have a minute?" during me holding a customer meeting. Left meeting, went into boss' office. "We don't want to prolonger our relationship together and thus decided to fire you. Sign here and here.." then 30sec of small-talk and some legal advices later "Take your backpack and your jacket. The other things you can collect at some other time. Please leave now." I was "running" that company's production and education for 14 years and was the only senior with wide skill set, actually - I was thrown out in 2 mins without a thank you or given time to say good-bye to others. You tell me something about Frohnhofer ... lol!

> - engineers were told today to print out their last 30 to 60 days of code, so they could show it to Elon Musk himself.

Musk understands what it means to code. So to select who can code, who do code, who efficiently can and do code and who are up to propper selected algorithms and efficiency - there is just only way:

you let them show what they can / did in the last xx days and let them explain why they coded what they show. Thats an very efficient knowledge and performance test. To let them print it out is actually important. It shows the amount of LOCs written and problems solved - its not always good to write a lot. Each application at FAANG, or others, nowadays consists of such assessments, whether live-coded or problem-assigned. Nothing different what musk did, too. Just think about your argument once again.. How do you want to assess whom to keep and whom to fire in a fast manner under time pressure?

> - or there was a guy who couldn’t even get info from the dedicated function for such stuff - HR - whether he was fired. That also had to be resolved by Musk in a Twitter exchange

Seriously? I can't even get the info if my train is delayed in a working public transport system and you expect everything to simple function, while the whole HR and loan filebase is being restructered and turned upside down?? ...

> How come you ignore that a huge portion of employees aren’t involved in any tech stack? Think 80% of employees were fired.

I don't ignore them. They're inluded. The tech-stack means the technology twitter is running on and the ressources (human, non-human, financial, ...) needed for running it. Including the ones who do not directly operate the servers, code or design the systems. Who's not needed for restructure, will go. If needed, the position can be filled later again.

> What’s your proof of competence? Like in this case: on paper, .... .

Exactly this has been done. They [all] were asked to present, explain and argue on their works, not on their CVs. It doesn't matter what reputation one has gained. Its pure skills that are here accounted for. And pure skills can usually work under time pressure w/o problems. Only the necessary for restructure will be kept.

> These are just empty words, care to give specific examples of what fresh ideas/synergies and other work-force positive things this great process resulted in?

You can't imagine that there are or might be some? From [3]:

> Among the dozens who Elon Musk enlisted specifically from Tesla are: director of software development Ashok Elluswamy, director of Autopilot and TeslaBot engineering Milan Kovac, senior director of software engineering Maha Virduhagiri; Pete Scheutzow, a senior staff technical program manager, and Jake Nocon, who is part of Tesla’s surveillance unit, as a senior manager of security intelligence.

What could the benefits to pull in some directors & seniors into the process only could be? ...

There is clearly a bias in your thinking, if such a basic thing like "a bunch of people are thrown into problem solving under time pressure" do provoke a question as "what benefits does this great process bring", then I can't help. But sure, you are not that interested in this topic as it would be enough to get the information for your argumentation by yourself. May be you should read up the news and not ask others to do the work for you?

> Doge was also accompanied by "qualified stuff from tesla/twitter/etc", but instead of uncovering blind spots they were mostly acting blind.

How does DOGE fits in here? Ah... "Musk and his qualified not-able'ers from tesla run DOGE".

> Because you’re supposed to maximize revenue and profits, not minimize them?

I think, I answered this excessively already. Forbes and Bloomberg write about the "magical" turn-around. You should read up the actual news about X and X.ai. Instead going ad hominem at insert some enemy here, you could account what has been reached at Twitter/X - its still there and running with less costs and rising revenue. It dio not collapse as it was foreseen by all.

[1] https://www.investopedia.com/articles/personal-finance/10201...

[2] https://www.scu.edu/media/mobi/Fixed-and-Variable-Costs.pdf

[3] https://www.cnbc.com/2022/10/31/elon-musk-has-pulled-more-th...

Imagine hiring a consultancy, and getting stuck with a team of people fresh out of college without any wisdom to guide them... I think that's what the American people got.

Not just "fresh out of college", but people who think AI can solve everything. COBOL too "ancient"? Just throw AI at it! Can't write code? Throw AI at it! He literally wrote an AI prompter using AI:

> Under a time crunch, Lavingia said he finished the first version of his contract-munching tool on his second day on the job — using AI to help write the code for him. He told ProPublica he then spent his first week downloading VA contracts to his laptop and analyzing them.

This is entry/junior-level levels of incompetence being given carte blanche access to production, but where "production" includes people's livelihoods.

Did you see how two billionaires "leaders" feuded yesterday over disagreements (one was the President of the US and the other was the world's richest man)? They were the ones behind DOGE. That was the leadership behind it. We are in the fucking Twilight Zone. It's surreal.

There was a epic book about the whole Iraq situation that had the most apt title:

FIASCO

Then there are these military terms:

FUBAR - Fucked up beyond all recognition

&

TARFU - Totally and Royally fucked up

&

SNAFU - Situation Normal: All fucked up

It's the people that have been to war that seem to have the words for what's going on, no one else.

except VA is a byzantine hellhole of self-serving dysfunction. Your outrage is only partially appropriate. Anyone around the system knows this, but has various incentives whether they will say it.

Maybe there's a balance between not doing anything about the problems in the VA and ignorantly breaking in in 48 hours?

Maybe this engineer is just saying, like all big projects, there are mistakes, and there should be more vetting, debugging. It was rushed, so yes, there are mistakes, but not, "generally, all my stuff is bug ridden crap"

This isn't your typical SV startup. Move fast and break things works for garbage like Juicero or Uber for Dogs or whatever silly idea gets VC funding this week, it does not work when you're talking about the lives of millions of people

But in this case it WAS all bug ridden crap. Casually rolling the dice with people’s lives like this means you probably won’t be able to live without some fear of people coming looking for you when this is all said and over.

Maybe they don’t realise it yet but their lives have just changed for the worse in a very significant and hard to reverse way. This is going to catch up with them on a reasonable timescale.

I was making a comment on software projects, not a political statement. Thanks for the downvotes. Of course the situation is awful, but not every programmer gets to cherry pick the picture perfect project to work on.

The job market is difficult right now, so who wouldn't take this one? That's the more insidious message from capitalism, there are a lot of cogs in the machine that are fully aware they are cogs, but you still have to eat, so you keep doing cog like things.

I think you are misinterpreting the comment here.

>I would never recommend someone run my code and do what it says.

Using this as a tool to signal can be a huge time saver. But not taking the code output as "truth" is what the engineer meant.

> It’s unclear how many have been or are on track to be canceled.

This article, just like you, are taking all of these out of context.

>Google Maps says drive into the lake. Do not drive into the lake.

He's using LLM's as a potential flag, then you go and human review those contracts with lawyers, supply chain, business specialists, nurses on the ground, etc. etc.

Calling him an idiot for being transparent is a personal attack, and has no place here.

Directly from the github: Tools I have written to do stuff more effectively at the VA. Code exists to make humans more efficient; not to replace them. All code leads to human review. Feedback appreciated!

Spoken like a developer who doesn't understand how real world works.

> Using this as a tool to signal can be a huge time saver. But not taking the code output as "truth" is what the engineer meant.

Lavingia didn't provide any provable metric that this tool was going a "huge time saver". Additionally, if the output isn't the "truth", what can kind of time saving is expected? That a person needs to run the code, get result and then spend time analyzing results and cross verifying contracts again?

You know what will be a good time saver? Open the contracts and cross check it if it is valid. No AI slop required.

> This article, just like you, are taking all of these out of context.

Real people are impacted by VA contracts. If there is even 1 contract impacting even 100s of people that is enough damage. Saying that this is people taking things out of context shows your of empathy for other human beings, at least if they are not coders.

Because apparently Lavingia should get kudos for his transparency. But calling him an idiot for his lack of self-awareness that he had no clue about government contracts and should not be writing AI and creating this mess, all the while impacting people's life? Well we can't have that.

> He's using LLM's as a potential flag, then you go and human review those contracts with lawyers, supply chain, business specialists, nurses on the ground, etc. etc.

If human review is the point then start from there. Don't write AI slop, generate wrong result and then ask for human review.

To the larger point - Benjamin Franklin said - 'It is better 100 guilty Persons should escape than that one innocent Person should suffer'. In the same vein every well meaning contract/legislation which is helping 100s will have 10s exploiting the loophole. People like Lavingia and from DOGE seemed to have decided that rooting out those 10s is worth hurting the 100s. So, lets write an AI for that.

using an LLM is a very VERY small part of the code he published. 12 lines out of around 13k lines. You seem to be disagreeing with the concept, not the actual code. you prolly could have reviewed the code in the time it took you to google that ben franklin quote.

If you look at the code, he explicitly filters for sole source before he even sicks a LLM on it.

Tell me what you know about far part 6

I'm not sure exactly where you get 12 lines of code from. There are at least a hundred lines of LLM-related "prompt engineering" alone: see https://github.com/slavingia/va/blob/35e3ff1b9e0eb1c8aaaebf3..., for example.

Regardless, number of lines is not an accurate metric of how "essential" a component is to the functioning of the overall system. How about this, try removing the LLM-related code from the script... what functionality is left, exactly?

> Mistakes Were Made

https://en.wikipedia.org/wiki/Mistakes_were_made