What the fuck is the rush!? They should do this, but there is zero reason to rush it unless you want people to suffer needlessly.
It’s easy. Rush to “fix” things, but break them in the process. Then tell the American people… look we tried and it’s not working! Let’s privatize! Moving all of our SS money into hedge funds and wall street scum. Welcome to Trump’s America.
Right… That’s covered in the “making people suffer needlessly” part.
A sad fact about programming projects is that you can and have to lie about how long it’ll take or cost because nobody will ever accept the truth about those things.
Why would Elon hate his mother? She kept him from being beaten up by that facebook guy.
Someone need to back up this codebase ASAP so it can be restored when thing surely go to shit because of Elon and some random LLM.
Avoiding Gell-Mann amnesia! Unfortunately, to even get to this question you have to be expert enough in one area to see through the BS… Not a luxury everyone is afforded.
I’ve been a programmer for around 20 years now and I can tell you with 100% confidence that this is going to be an absolute fucking disaster. I’ve built project features—not even full projects—that took months. Hell, we launch one on Monday. This would realistically take years, and after seeing how garbage the DOGE site was built, I have zero confidence in the incel team Elon scraped off his Steam friends list.
Yeah, the ideal process is something like this:
- Write a bunch of automated tests
- Rewrite a chunk that’s well tested
- Go to 1 until the project is done
I would budget a couple years. Even a smaller codebase would take more than a couple months.
Yeah, this. They need this to be completely test-driven. Build a foundation and build out/finalize in pieces. If they had an actual dev team that knew what they were doing, someone would set them straight. They also need a ton of time to gain an understanding of what every part of the existing codebase is doing.
well years are made of months
I vaguely remember the same story when he bought twitter and got called out by a senior dev for not understanding shit. History repeating itself speedrun any%?
Someone at work was saying people should download their work records from the ssa web portal before they mess it up/lose it.
Don’t worry, I’m sure they’ll make a full backup before they do something as drastic as try to re-write the system. /s
I fully expect them to celebrate launch day by taking a sledgehammer to the old machine like that scene from Office Space. All I can say is that I really hope the SSA has been virtualizing their old COBOL applications and not running them on an ancient mainframe.
I just tried to check that out and got “online service not available”
I’m sure they have their remaining engineer on the job and it will be back soon o_-
Imagine the overload on the systems as everyone rushes to do just that…
If they are genuinely stating months as a realistic timeframe then they either know nothing or are fully aware and this is a deliberate killshot.
This is the typical Elon “Fully self driving soon” of 2016.
Bud, this isn’t “typical” anything.
No. It is a killshot.
He stated his resentment of social security MULTIPLE times.
Sabotage is the purpose.
You might be right. Cutting expenses and saving money, but not in the way that any normal person would want.
The black market for social security data is gonna have a field day.
It will kill the black market for that kind of data. What good is a black market if the data is trivially available to everyone?
How much you want to bet they are going to feed the code base into ai and ask it to re-write it in a different language?
You know what? I’m slightly more confident about the AIs chances compared to his team of 20 year old lead by Big Ballz.
even ChatGPT agrees this is stupid:
even his own Grok is against the idea:
Between this and the other instances of Elon’s AI correctly calling a duck a duck (or rather, a Nazi a Nazi and a disinformation spreader a disinformation spreader), Elon might be able to replace himself with Grok as the head of DOGE and see significant improvements almost overnight.
Tbh, replacing our unhinged leadership with LLMs would probably make things a lot better in the near term, though not the mid or long term, as IME they tend to be way too cautious about things they shouldn’t be and way too confident about things they shouldn’t be.
In general, I agree with your conclusion. In this specific case with DOGE, I would argue otherwise.
Elon is overconfident about things he shouldn’t be (which is everything) and not cautious about anything. With a LLM being dumb as hell and confidently wrong, that’s actually not any worse than what we have now. If it somehow manages to be cautious about anything at all, it’s an honest to goodness improvement.
As if “accidentally” breaking it during this process would be a bug, not a feature.
“Oh no, we completely broke the SSA, it’s all gone. Soorrrrrryyyyyy”
Fr. They’re gonna “accidentally lose” all the records of who was receiving social security and how much they were getting, and they’re gonna make people reapply for it in person, but they have to bring their birth certificate, social security cart, passport, driver’s license with real id, 5 years of tax returns, and probably some new special certificate they start issuing to prove you’re
magaa real citizenThey’re gonna break it, then when people need to reapply, they won’t be able to because they cut funding to phone lines and ssa offices
Oh fuck this… I can’t even look at my own code from a year or two ago and understand what the hell I was doing.
I learned some very basic COBOL back in high school (using punch cards), and that shit was confusing as hell.
There is no possible way that this ancient codebase could be migrated by a bunch of 20 year olds in such a short amount of time.They’re just going to feed it to some LLM and then rely on the slop it outputs.
I need to stock up on popcorn for this
As someone who’s actually been part of a team that did this exact migration, it’s gonna be shitshow with tons of bugs when the mainframes are decommissioned and the new system goes live.
It’s not the wrong move cause the tech is increasingly obsolete but the level of effort will be underestimated.
I think this type of project would take at least five years. A minimum of three to develop, 1 to do internal QA, and at least a fifth as a slow roll out from the smallest states while progressing to the larger. If bugs are revealed during that fifth year, slow the roll while getting them patched up.
It is the sort of thing that requires long term vision, effective governance, and patience. Things that Trump and friends do not possess.
Surprise surprise, that’s a similar time frame the last attempt at this gave. Apparently the they never got too far off thr ground because they had to divert resources to help against covid
it is only obsolete if you have a better currently working solution
anything else is a fucking pipe dream
Just toss it all into an Access database. What could go wrong?
This comment has spoken to me!
Nope it’s also obsolete if the cost of maintaining an old system is far higher than a buggy new system. That’s what accountants tell me anyway.
people say this at a manager level, then go over budget by 3x in time, 2x in cost and only mostly get the same result as the original
ask me how I know
accountants, business majors and leadership with 0 experience doing the actual job need to fuck right off
fucking wanker muppet elmo
