DOGE To Rewrite SSA Codebase In 'Months' (wired.com) 314
Longtime Slashdot reader frank_adrian314159 writes: According to an article in Wired, Elon Musk has appointed a team of technologists from DOGE to "rewrite the code that runs the SSA in months." This codebase has over 60 million lines of COBOL and handles record keeping for all American workers and payments for all Social Security recipients. Given that the code has to track the byzantine regulations dealing with Social Security, it's no wonder that the codebase is this large. What is in question though is whether a small team can rewrite this code "in months." After all, what could possibly go wrong? "The project is being organized by Elon Musk lieutenant Steve Davis ... and aims to migrate all SSA systems off COBOL ... and onto a more modern replacement like Java within a scheduled tight timeframe of a few months," notes Wired.
"Under any circumstances, a migration of this size and scale would be a massive undertaking, experts tell WIRED, but the expedited deadline runs the risk of obstructing payments to the more than 65 million people in the US currently receiving Social Security benefits."
In 2017, SSA announced a plan to modernize its core systems with a timeline of around five years. However, the work was "pivoted away" because of the pandemic.
"Under any circumstances, a migration of this size and scale would be a massive undertaking, experts tell WIRED, but the expedited deadline runs the risk of obstructing payments to the more than 65 million people in the US currently receiving Social Security benefits."
In 2017, SSA announced a plan to modernize its core systems with a timeline of around five years. However, the work was "pivoted away" because of the pandemic.
Can't wait (Score:2, Insightful)
For Biden to get blamed for this when it fails catastrophically! This is for the Fox News viewers out there, did you know an Elon staffer with the nickname of "big balls" used to provide support for a cybercrime ring? https://www.reuters.com/world/... [reuters.com]
Re:Can't wait (Score:5, Insightful)
It can only fail catastrophically. I mean, even rough estimations put this at > 1000 developper years and that essentially means these cretins will try to do it "with AI". Now, AI is not very good at writing code. Simple, small things, yes, usually. But anything complex? No. Hence they will very likely end up with somethign that does not even run abd cannot be fixed.
Re:Can't wait (Score:4, Interesting)
For larger projects, there are a few tools out there that will loop the already looped LLM to basically coax what it wants out of it and autocorrect iteratively, and it's pretty hit or miss.
Sometimes it produces something pretty big and pretty complicated, and it's entirely functional.
A lot of the time, it's not though.
But one thing is certain- they had better use a memory safe language, because the C my best local LLM generates is very, very, unsafe.
Re:Can't wait (Score:4, Insightful)
Soo, lets do some more estimates. Let's say they end up with around 60MLoC as well. Lets say AI adds one show-stopping bug every 1000 LoC. Since nobody knows the new codebase and it is AI generated, lets say 10 developper days to find and fix one. Then we have 1500 developper years to find and fix these. But that is only for simpler errors, nothing fundamental. And it is a rater low estimate on all aspects.
My prediction is that they will not even manage to crash this project completely in the next few years. Even that will take more time.
Re:Can't wait (Score:5, Insightful)
The problem is getting the specification right and more importantly complete. A system like this will have to deal with many, many special cases the rationale for them being scattered in many different places. So: getting something that does roughly the right job for the really common cases might, with a great deal of luck, just about achievable.
Next is going to be importing the current data. This will involve some conversion, things will not be described the same way in old & new systems. There will be errors.
Then there is testing to make sure that it does what it is supposed to do. This step alone will take many months.
But, I suppose, as far as Musk is concerned those mostly affected will be poor people; about who probably do not much matter to him.
Re:Can't wait (Score:4, Informative)
The problem is getting the specification right and more importantly complete. A system like this will have to deal with many, many special cases the rationale for them being scattered in many different places. So: getting something that does roughly the right job for the really common cases might, with a great deal of luck, just about achievable.
Next is going to be importing the current data. This will involve some conversion, things will not be described the same way in old & new systems. There will be errors.
Then there is testing to make sure that it does what it is supposed to do. This step alone will take many months.
But, I suppose, as far as Musk is concerned those mostly affected will be poor people; about who probably do not much matter to him.
Having been involved in several large scale conversion efforts, you are spot on. In my experience:
In the end, most of the attempts were abandoned after spending boatloads of money. Messing up system as complex as SSA that impacts million of Americans, many of whom actually vote, is a prescription for disaster as they storm congressional offices demanding for heads to roll if their checks are late, wrong or delayed. “Mistakes will be made but we’ll fix them” is not what they want to hear, and angry voters screaming at them is not what representatives want to have to face. I suspect compnies will suck up a lot of money and then new system quietly abandoned as it becomes clear it is not working.
Re: (Score:3)
Re: (Score:3)
Now the real thing here is that morons like Musk and incompetents like his young "engineers" (they are not) are not doing is asking themselves _why_ COBOL (and FORTRAN) are alive and kicking...
The reason is simple: They are _needed_ to run these old workloads and keeping the language alive is several orders of magnitude simpler than reimplementing things in another language. In fact that reimplementation will usually be excessively expensive and excessively risky and may even be outside of what the human ra
Re: (Score:3)
Thanks. About what I expected. This basically means such a rewrite may well become impossible for all practical purposes quite a bit before you reach the SSA code-base size. This is probably also the reason why COBOL and FORTRAN are alive and kicking and get updated to include new things every 10 years or so: There is no sane alternative.
In fact, if I want to, I have already been asked whether I might be interested in doing some FORTRAN work when I retire in about a decade. I could even claim some experienc
Re: (Score:3, Insightful)
Re:Can't wait (Score:5, Insightful)
And you'll likely end up with a horribly architected and poorly structured solution - but it will be in Java, Rust, or Go or the hot language of the hour instead of COBOL - but by the time it's done, that language will no longer be the hot language of the hour and they will have to hire people out of retirement to fix it.
Giving AI the existing system even as a primary source of direction would probably be a mistake. Better to give it the statutes, regulations, policy manuals and, to a lesser extent, training materials for the current system. Perhaps during the validation phase AI might probe the existing system to create test cases to run on the replacement system. If AI can't do that, the "I" is missing in AI (and it is).
No matter what, given the apparent schedule (even with infinite resources - see The Mythical Man-Month: Essays on Software Engineering published around the time I was in college), the whole idea is nuts unless the scope of the project magically gets redefined to be trivial (perhaps a field office app will get rewritten and DOGE will declare victory). Sort of like "Tesla 'Full Self Driving'" or "ventilators" or Hyperloop, or "much more efficient tunneling" done by The Boring Company (which has every large tunneling project beating down their door to get The Boring Company to do the job - NOT).
Of course 240 months is, technically, "months" but I don't think it qualifies as a "few" months.
Musk has already demonstrated that he has no idea how legacy systems work or the risk of assuming they are "obvious" - such as when he seems to have assumed that because the "Dead" flag for a SSN holder in one of the SS databases wasn't set and the person was over 70 they must be receiving benefits even if they were now over the impossible age of 150 years. Then he kept doubling down on that apparently unwilling to learn or just figuring that if he repeated a lie enough times, the facts would get lost in the chaos and he could claim that he was "never wrong".
Re: (Score:3)
Indeed. The reference to Brooks is both factually appropriate and shows how _long_ we already know that projects of such size are exceptionally risky and require huge effort. It will definitely not be a "few" months. And whether they even can get something that works at all is up for debate. There is a reason so much old COBOL and FORTRAN code is still around and used and that reason is that nobody ever (to the best of my knowledge) has managed to transform a very large code-base successfully to a different
Re:Can't wait (Score:5, Insightful)
The ultimate goal is probably to abolish the social security system completely and this is just a decoy operation.
Re:Can't wait (Score:5, Funny)
Like many legacy government IT systems, SSA systems contain code written in COBOL, a programming language created in part in the 1950s by computing pioneer Grace Hopper.
And there we have the real reason, SSA is using some woke programming language created by an obvious DEI hire, of course it has to go.
Re: (Score:3)
Well spotted! With that, we need to of course get rid of all of it! Cannot have code in a language made by a _woman_ in a critical system!
Now, hmm, Alan Touring was gay. I think we need to re-think computers in general here!
Re: Can't wait (Score:3, Interesting)
Re:Can't wait (Score:4, Insightful)
I'm sure someone will find a way to work in Hunter Biden's laptop. They always do. I'm pretty sure I even heard our current Attorney General whatabouting Biden when asked if she was going to pursue an investigation into the fucking Signal fiasco.
I mean, really?
Yup really. When asked about a DOJ investigation into the Signal group chat incident, she "what abouted" Hillary Clinton's emails, Jo Biden's files -- ignoring the fact that those incidents *were* investigated by the FBI/DOJ.
She also appointed herself judge and jury declaring that there was no classified information in the chat so no investigation was needed and that people should focus on the fact that the mission was a huge success etc... Meaning (a) the ends justifies the means and (b) their crimes are okay in the name of the greater good.
Any low(er) level person texting this information on Signal would be stripped of our security clearance, probably fired/repremanded, perhaps prosecuted. For example: Trump Official Mistakenly Looped Journalist on Email About ICE Operation 2 Months Before 'War Plans' Group Chat: Report [people.com]
A US DHS employee has reportedly been placed on administrative leave and will have her security clearance revoked after accidentally adding a journalist to an email that included information about upcoming Immigration and Customs Enforcement operations.
(This administration is a clown car filled with amateur clowns.)
Re:Can't wait (Score:5, Insightful)
It becomes a lot more plausible once you realize that the goal isn't to develop a functioning system, but to extract as much money as possible while breaking SSA.
We are so screwed. (Score:4, Insightful)
We are so screwed. I honestly don't know what the end game will look like, but things are about to get very bad for many people in the US.
Re: We are so screwed. (Score:5, Insightful)
Re: (Score:2)
One of the advantages of having old systems is that they are so antiquated that almost nobody knows how to hack them. As far as "months" goes, it will take months. Maybe 96, at the minimum.
Agreed. Also... I like Java, but (a) don't want the SSA to pay (probably outrageous) licensing fees to Oracle / Larry Ellison -- you just know they won't use the Open Source version of Java if another Broligarch can earn some coin here -- and (b) something will inevitably go wrong because of an unpatched exploit or runtime issue.
Re: We are so screwed. (Score:4, Funny)
Oh this fucking things probably going to be written in nodejs by a graphic designer who did the javascript course on coursera and now thinks he's a backend coder lol
And its gonna get pwned within a week....
Re: We are so screwed. (Score:4, Insightful)
Elonia has no clue about software creation and is full-on delulu now.
Re: (Score:2)
Re: (Score:2)
Heheheh, nice statement on several levels!
Re: We are so screwed. (Score:5, Insightful)
The other thing about them is that they were written to run on hardware and operating systems that have been battle-proven for decades, and the runtimes that run the systems are the same. These things have been chunking away on some of the largest accounting undertaken by mankind for decades and they've never missed a payroll even once.
I'm sure that some "move fast and break things" dipshits will be able to successfully reverse engineer the entire system that handles literally trillions of dollars of transactions per year from a 60 million line codebase of a programming language they've never written anything in; with a stack of Linux servers, and Node.JS in a few months. No problem! It'll be fine, because it will be "cloud native!"
This is going to fail spectacularly, and not in an entertaining way. More like a few million senior citizens being thrown out of their homes kind of way.
Re: We are so screwed. (Score:5, Insightful)
This is going to fail spectacularly, and not in an entertaining way. More like a few million senior citizens being thrown out of their homes kind of way.
What makes you think they would consider that a failure?
Re: (Score:3)
Re:We are so screwed. (Score:5, Insightful)
We are so screwed. I honestly don't know what the end game will look like, but things are about to get very bad for many people in the US.
Reportedly, the SSA hasn't missed a benefits payment in 90 years, I'm betting that's probably going to change ...
Social Security has never missed a payment. DOGE actions threaten 'interruption of benefits,' ex-agency head says [cnbc.com]
Commerce Secretary Howard Lutnick says his 94-year-old mother-in-law ‘wouldn’t complain’ if Social Security missed a payment, but he's literally s billionaire and she lives with him and his family -- so, ya, she'd probably be okay. Other people who rely on their monthly SSI check for rent and food, probably not so much.
Howard Lutnick says his 94-year-old mother-in-law ‘wouldn’t complain’ if Social Security missed a payment — claims ‘real America’ will be rewarded while the fraudsters ‘yell and scream’ [moneywise.com]
(So much empathy, understanding and compassion from those in charge now... /s)
Re: (Score:3)
Already changed since Elon's teen boy band took over.
https://abcnews.go.com/US/man-... [go.com]
Re:We are so screwed. (Score:5, Interesting)
Indeed. My sympathies are limited to anybody that did not vote for this delusional crap though.
I mean, 60 million LoC and "months" to rewrite? I would estimate you need 10...100 years to even come up with a sufficiently complete and correct spec for that rewrite. And then some more time to do that rewrite and get it to work.
There are good reasons why basically no complex codebase gets rewritten: It is essentially impossible to do for the human race. Maybe in 100-200 years, when we have made software creation a proper engineering discipline.
Re:We are so screwed. (Score:4, Informative)
You only need a complete and correct spec if you consider it a requirement that the new system behaves similarly to the old one.
If, OTOH, your requirements for the new system are sufficiently reduced, then the new codebase can be straightforward also. For example, if you don't care whether the new system actually works or not, then the new code doesn't need to do anything at all.
Predictable outcome (Score:4, Insightful)
Re:Predictable outcome (Score:4, Informative)
Re: (Score:3)
Normally this kind of thing would be met with nationalistic mouth-frothing rageposting, but I'm pretty fucking sure anyone who has their eyes open knows you are speaking facts.
Re:Predictable outcome (Score:4, Insightful)
There's really only two options:
1. They want cover while they loot the system. "Our new system revealed there's hundreds of billions of dollars missing" and all the tracks of them stealing it are gone.
2. Elon is a complete and utter moron. Given hyperloop, those tesla semi trucks, etc, etc that's certainly likely.
Re: Predictable outcome (Score:3, Insightful)
Just remember that those options are not mutually exclusive.
Re:Predictable outcome (Score:5, Insightful)
Of course they are. The whole "250 year olds are getting social security!" thing was showing how ignorant of the whole thing they are.
Everyone is celebrating the "waste" and "fraud" they find, but a lot of these things have delayed effects that we won't see for months.
Things like USAID being disbanded is going to hurt a LOT of farmers because they were a big customer buying lots of surplus food to give away. That stuff won't happen until the fall harvest and suddenly farmers will go bankrupt.
They're panicking right now because the stopping of IRA (inflation reduction act) payments meant many of the improvements they paid for are no longer being paid for, so now they're stuck having paid for something they wouldn't have bought otherwise.
And now they're wanting to force SS recipients to visit in person an SSA office, many of which were closed during the mass firing.
This stuff is like a timebomb - it doesn't go off now, it'll go off way in the future.
Even more funny is that farmers, verterans, etc., the people most affected by all this, generally are hardcore Republican voters.
The "FO" part of "FAFO" is coming.
A thought (Score:3, Interesting)
If you want to watch mental gymnastics in real time wait for the usual MAGA fans to show up in the comments. They certainly know IT and know you can't rewrite a project of this scale in months. But will they be able to admit it's a bad idea? Of course not. They're too deep in the dogma.
Re: (Score:2)
They certainly know IT and know you can't rewrite a project of this scale in months. But will they be able to admit it's a bad idea? Of course not. They're too deep in the dogma.
Indeed. They are incapable of admitting their heros are really lying, scamming cretins and hence they start to hallucinate from all that cognitive dissonance.
Re:A thought (Score:4, Funny)
I'll be watching for the inevitable references to Hunter Biden. Because somehow he always gets dragged into it. Always.
Re:A thought (Score:5, Funny)
Hunter Biden's laptop was found to be full of Ancient, Cryptic COBOL - and no one purportedly knows what it's doing there... maybe running Hillary's email server?
Grok to the rescue (Score:5, Interesting)
What will really happen: The code will be simplified, the regulations ignored, so the SSA's new AI bot can decide who is 'Trumpy' enough to deserve a Social Security cheque.
Re: (Score:2)
What will really happen: The code will be simplified, the regulations ignored, so the SSA's new AI bot can decide who is 'Trumpy' enough to deserve a Social Security cheque.
That's fucking terrifying -- and, sadly, probably spot-on. ...
People with (continuing) Trump donations in their IRS files will get preferential treatment
Re:Grok to the rescue (Score:5, Insightful)
We're literally already seeing immigrants with legitimate, legal, visas, who haven't breached any rules or committed any crimes deported because they disagree with Trump over, say, Israel.
Regardless of political side, extremists do use the power of the state to punish people for disagreeing with them. The fact that you don't see your side doing that, when it's already going on, is a prime example of the willful blindness to their own side committing wrongs the GP (and others in this discussion) are talking about.
JAVA? (Score:3, Insightful)
Re:JAVA? (Score:4, Funny)
Yeah, I'm surprised that the 20 somethings on the DOGE team aren't telling them to rewrite in RUST. Isn't that the new programming language hotness right now?
Re: (Score:3)
Re:JAVA? (Score:5, Insightful)
Say what you will about Rust, but coding in Rust requires some discipline. If you don't have it, you'll quit after a few days of fighting against the borrow-checker.
I think they're going to aim for "vibe coding"; just tell an AI what you want the system to do, and whatever it comes back with, that's your code.
Re:JAVA? (Score:4, Insightful)
Being completely serious (I know, what's the fun in that?), Rust is a challenging language to use. It's popular among experienced programmers who understand the benefits, and are willing to put in extra work to do things the right way. NOT the sort of people who would think they could rewrite a 60 million line code base in a few months.
Again being completely serious, they probably intend much of their new code to be AI generated. They think they'll just write a few prompts and get a computer to do all the hard work for them. That's the trendy thing among novice developers who don't understand what's meant by maintainable code.
Re:JAVA? (Score:5, Informative)
So move from one antiquated, abandoned language to another?
As it turns out, moving from COBOL to Java is a common approach in the financial software world. While they are clearly different languages, Java, early in its life, implemented a BigDecimal class to precisely implement decimal arithmetic, which is critical in financial calculations.
umm.... which code base? (Score:4, Insightful)
Re:umm.... which code base? (Score:5, Informative)
Let's not forget the number of things that SSA deals with that are governed by state law (are these two people legally married?, is this person disabled?) to just add complexity to the mix. It's the sort of thing a junior developer with no experience would think they could do in months.
My father retired from SSA after nearly 40 years there, so I picked up enough over the years to understand that one might readily spend that long just understanding how complex a problem they were about to tackle.
I did meet someone who took care of that COBOL code in the late 80s/early 90s. They were about to start Y2K remediation. I did not hear of that anywhere else for oh, 8 years or so. So the folks taking care of it clearly took pride in their work.
Re:umm.... which code base? (Score:4, Insightful)
"My technical and management genius will easily make those problems go away" and not end us up in months of delays and fixes and a 2nd full rewrite.
Re: (Score:2)
They have zero chance of even understanding what you just explained, nor do they care.
What's more interesting is what the end game might be. We know the long term goal is the destruction of social security, destroying the database doesn't seem like a good path toward that end.
oufff (Score:2)
collective facepalm.
Do it! (Score:2)
Unfortunately probably not lying... (Score:2)
"Rewrite the SSA codebase" is only difficult if you implicitly or explicitly assume that the replacement must be equal to or better than the old system in terms of reliability, feature completeness, etc. if you are somewhere between depraved indifference and outright hostility about the
Superhero does miracle... (Score:2)
Or rather wannabe superhero will fail to make good on his promises. Signs of megalimania are getting stronger with this cretin.
In actual reality, doing something like that in that time is far out of reach of the human race.
No worries! (Score:2)
No worries; Elon's kids will use ChatGPT. Or... err... xAI.
When you have no idea what you need to do ... (Score:4, Insightful)
... it always takes just weeks or months to do.
If they actually read and understood what all that legacy code needs to do, and spent time to write a project plan or specification for an upgrade,
that would probably take a YEAR or MORE to accomplish.
Yeah, when you don't understand how something works, it always takes 'just a few months' to replicate it.
Rewriting what, exactly? (Score:2)
One could make an argument that rewriting the core number allocation and check writing services could realistically be completed in a few months. But we all know there are so many cases that are obviously being overlooked here, and they are most certainly going to try using AI. It will likely be a sad failure, and it all could be avoided.
At What Cost? Who's doing the job? This is graft. (Score:2)
The obvious answers are: X is doing the recoding at likely price of many billions of dollars, and they will "fail" because they did no work and si
Re: (Score:2)
Flat (Score:2)
If we were to switch over to a simple flat tax system I could write the code for them in a week in with a few hundred lines of QuickBasic code.
Re: (Score:2)
Re: Flat (Score:4, Interesting)
worked in social security software (Score:3)
For us the biggest problem was that different rules applied in different time of history. And border cases where not always clear.
Effect:
Feature estimated to take 3 months and 80-100 test cases ended up taking 15 months and over 460 test cases....
Ketamine (Score:3, Funny)
Do you manage all your finances on X? (Score:2)
“When I say payments, I actually mean someone’s entire financial life,” Musk said, according to audio of the meeting obtained by The Verge. “If it involves money. It’ll be on our platform. Money or securities or whatever. So, it’s not just like send $20 to my friend. I’m talking about, like, you won’t need a bank account...it would blow my mind if we don’t have that rolled out by the end of next year."
So easy. Here's some sample code (Score:2)
If ($user.party == "R" and vote.record.last == "Trump") then check.send()
else check.skip()
I wondered how they were going to get rid of Social Security. They'll break the code, blame it on the "deep state", bulldoze the whole thing and claim to be building it into "something great (tm)". Let the undesirable people get broken and/or die, and lose the ability to vote (no home address, etc.) Well played, team evil. Well played.
Utter lunacy (Score:4, Insightful)
They will probably use their screwy AI models to crank out code.
They only upside, maybe the nation wakes up once scores of old Americans starve on the street.
Then again all the avoidable deaths from Covid hardly moved the needle, so I am not holding my breath.
This has strong "smartest guys in the room" vibes (Score:3)
When (Score:4, Informative)
Will this rewrite come before or after the air traffic control rewrite?
Pick 2 (Score:3)
DOGE To Rewrite SSA Codebase In 'Months'
Good, Fast, Cheap
They've already specified "fast" so we're stuck with "good" and "cheap".
One of those is really great for them financially and the other is disastrous for us results-wise.
Which do you think they'll pick?
They very likely to hell... (Score:3)
I've spent some years in social security algorithms and it is a hell.
The problem is actually politics - politicians love to "make people lives better" so they add rules, exceptions and special treatments.
This leads to complicated rules that change every few months and to properly apply the law - your code needs to take into account the fact that for different periods of time different rules applied and sometimes it is not clear how border cases should be handled... Add manual fixes and exceptions to that
and you end up with 15 months of development for something that normally takes 3 months...
The question how much will they do/break... and how will they cover the f.up
Requirements documents? (Score:4, Informative)
I don't know about the SSA, but I went to talk at an Information Science conference about 15 years ago with someone from the IRS.
They said that the bulk of their time was trying to translate what the hell the laws that Congress passed actually meant. They had given Congress some madlibs-type templates to use, so they knew they could parse it cleanly and translate it into specific requirements in whatever language their systems used.
Even if someone were to try to write tests to duplicate the existing code, you have to ask which of these are actual requirements, and what's just quirks of how we're fulfilling those requirements.
(spent more than a decade doing data integration at NASA, and lots of legacy data stuff before then)
Re: (Score:2, Insightful)
If you're currently receiving social security payments I'm going to be laughing in six months when they stop.
Re: I'm optimistic (Score:2)
Re: (Score:3)
Spotted the idiot that has never learned that complete rewrites of incredibly complex logic almost always fails after taking inordinate time and expense to come up with something worse.
There's a reason why the existing system is still the existing system.
Re:I'm optimistic (Score:4, Insightful)
Sure, if you are willing to accept the revolution, mass murder, and mass suicides that will result.
You CAN'T make the requirements "less dumb" when talking about government, because government, unlike private sector, can't pick and choose its customers, it either serves them all potentially, or none.
Also, companies go bankrupt, people find new jobs, some guys are out money. Governments go bankrupt? Generally a prologue to mass death, carnage, potentially revolution and civil wars, along with invasions and other fun stuff.
Re: (Score:3)
What you actually are is called "delusional" or in modern terms "delulu". In that timeframe, it is not even possible to produce anything that works for a task of this complexity. Hence whatever will be produced will in fact be much worse.
Re: (Score:3)
Thisn is an absolutely key point. No matter how expensive it is to run the existing COBOL-based system, the costs of the compute and the complexity are absolutely miniscule by comparison with the costs of the actual social security payments. It's a complete irrelevance. Social security costs lots of money because it involves making payments to a large chunk of the population. The only way you change that is to make smaller payments to fewer people. The admin costs don't count.
Re: (Score:3)
No. For a software project of this size, that is not true anymore. What they claim they can to "in a few months" may well be larger than the whole US budget for 10 years ... or 100 years. Software project cost increases _exponentially_ with size.
Re:I'm optimistic (Score:5, Insightful)
Whatever Elon's crew does can't be any worse than the shit cranked out by the usual government contractor bloodsuckers. Also, over sixty million lines of code in the current system? Wtf, do they just write a new line of code for every retiree???
The SSA payment system is incredibly reliable and reportedly hasn't missed a benefits payment in its 90 years (Google it). Also, (a) COBOL is a very verbose programming language and (b) there are a myriad of laws, rules and regulations governing SSI payments that have to be handled and that's the fault of Congress not the SSA.
From a practical point, Elon hasn't demonstrated that he and/or his DOGE crew have any understanding of what people do or how things work in any part of the government, and no indication that they're interested in learning - and they're just slashing things willy-nilly at the moment, not trying to re-implement / re-engineer anything. They are most certainly going to fuck this up, probably hard, if it even ever actually gets attempted.
Re:I'm optimistic (Score:5, Interesting)
if it even ever actually gets attempted.
And we've arrived at the grift. I'm sure there will be lots of purchasing of AI compute power from xAI (it will not go to a bidding process, because any IGs that would oversee such legal requirements were sacked back in January).
Then, when it's not done in "months" (however long that is) and everyone has forgotten about it because of a few months of ever-increasingly alarming actions being undertaken combined with a press that has a memory reset every 24 hours, they'll quietly shut it all down after siphoning millions of dollars out of the Social Security Administration's budget and delivering nothing but increased revenue to xAI.
Re: I'm optimistic (Score:3, Insightful)
That's the very, very best case.
My fear is that they will start the new system, then break the old system and announce that they have no choice but to use what they have created, and it will have been designed to cut people off.
Re: (Score:3)
I think it should be stressed over and over again that nobody voted for the billionaire from South Africa.
Re: (Score:3, Informative)
Re:I'm optimistic (Score:5, Insightful)
There is a famous saying: if it ain't broke, don't fix it.
It's much cheaper to pay people to learn COBOL than it would be to rewrite that much code that is working fine. It's not like COBOL is rocket science.
But even if you did want it rewritten to be maintainable, the team you hire to do that should probably not be the same people who want to get rid of the whole thing. There's a bit of a conflict of interest there.
Re: (Score:2)
Whatever Elon's crew does can't be any worse than the shit cranked out by the usual government contractor bloodsuckers.
I have yet to see anything that could be called "competence" from Elon and his DOGE posse. Bring on the government contractors please.
Also, over sixty million lines of code in the current system? Wtf, do they just write a new line of code for every retiree???
Have you ever seen a COBOL program?
Re: I'm optimistic (Score:4, Insightful)
Re: (Score:3, Informative)
For code written over the space of five-six decades, it's already not surprising it's grown to millions of lines of code.
If you add to that it's COBOL, which typically takes 4-5 times as many lines to express the same content as a Python program of equivalent functionality, then it's even less surprising.
COBOL is very, very, verbose. Statements that C would encode in a single line are often 4-5 lines deep. Everything has to be declared in advance, and those declarations are themselves verbose. COBOL source
Re:240 months is "months" (Score:5, Interesting)
Also, don't forget that this is the "move fast and break things" crowd that would not understand the reason why they're using huge ass mainframes running real big-iron databases like DB2 instead of containerized node.js garbage.
They clearly have no understanding of the scale of the problem they're trying to solve. I don't even know if they could enumerate the problem beyond "green-field rewrite social security because somehow that's easier than learning COBOL for alleged software guys"
Failure is the point (Score:4, Insightful)
They aren't trying to produce a modernized system that does the same tasks just as well. They want to produce a buggy, broken version that denies people the benefits they earned.
Killing off a functioning system that many people rely on is best done in inches. If they migrate people over little by little, if the payments become unreliable rather than vanishing, fewer people take to the streets in any given month. And once people are contemptuous toward social security, it's easier to remove it via legislation.
Re: (Score:3)
MODS: Please mod parent up (Score:2)
zepg2020 is 100% on. I've been on several refactoring projects and keys to success are:
1. Defining the scope. To do that with an existing codebase requires understanding, diagraming, and documenting. That's a few man-months right there.
2. Defining the scope of the rewrite (refactor), picking a language, a data mine (db, flat, whatever, although currently SQL is the rule so Maria, My, Post, Oracle, NO, etc.)
3. Mapping out how to get from #1 to #2 without anything dropping in the middle.
4. Developing a tes
Re: (Score:3)
Back in the '90s when I was in college our BASIC programming instructor told us, "If you learn COBOL you'll never lack employment for the rest of your life."
Of course I ignored him, and went for RPG.
Re: (Score:3)
I was sure it meant "Single Sign All"
Private sector fails often (Score:3)
I think you have overoptimistic view of private sector...
Private sector fails oftern - you just do not see/remember the failures... some fail there and some succeed and you see and think mostly of success...
The problem is that the chance of failure acceptable in private sector is completely unacceptable in places like social security...