Gallup: Few Major U.S. Political Figures Rated Positively on Balance

“Secretary of State Marco Rubio, House Minority Leader Hakeem Jeffries, and Secretary of Health and Human Services nominee Robert F. Kennedy Jr. have favorable ratings that exceed their unfavorable ratings by seven or eight percentage points. However, at least one in four U.S. adults are not familiar enough with Rubio and Jeffries to rate them.” - Gallup

Discussion

I think it is clear that consistency is not the staple. Trump vowed to take politics out of prosecutions. A few weeks in, and the Adams case shows that is not the case, resulting in 7 attorneys in the Justice Department resigning over the steps Trump was taking. And Holman on Meet the Press this weekend, clearly outlining the steps that were taken were part of a quid pro quo.

Is President Trump overstepping his role or are the judges who are blocking his actions overstepping their role? If readers of SI are not able to agree to the answer to that question, then it leaves the rest of our nation in a bit of a limbo for the time being. That is why it takes time for these issues to be challenged and tested. Eventually they will be appealed to the Supreme Court and the court will either agree to hear them or not. Then if they hear them, they will be ruled upon. For us to adamantly declare that they are legal or not legal at this point just shows how much disagreement there is in this country about what the law actually does or doesn't allow a judge or a president to do. Thankfully we do have a process and the process is unfolding. Hopefully clear guidance will come out of that process so that future judges and presidents can follow the Supreme Courts findings. As it stands now, political pundits and political law experts are split about what is legal and what is not legal. (Edit: even federal judges are split). For most of my lifetime, Republicans would often back down once their opponents challenged the legality of what they were doing even if they thought it was legal. That has changed under Trump though.

That is why it takes time for these issues to be challenged and tested.

So why has Trump decided to issue executive orders at such a pace that the practicality of taking the time to challenge and test them is thrown out the window? It seems he is purposely trying to overwhelm the system in order to ram through his agenda.

Anyone who wants to better understand the lack of fraud when Elon claims from the Oval office that there are 150 year old people getting checks from Social Security, this is a good primer. Probably not surprising as most of the DOGE staff are under 24 and may or may not have had any college education. Of course, dump the data into an AI engine without any framework around the data and it is not surprising that this is the fraud they are finding.

https://www.wired.com/story/elon-musk-doge-social-security-150-year-old-benefits/

....OK, what we have, at the very least, is government computers about a quarter century old running programs on Cobol. I've worked most of my career in electronics reliability, in particular with computers, and the standard warranty period is about five years. If you get ten years out of a system, you are doing really well, and part of my job is to tell the service reps to tell the customer "You have had this system long past warranty--update! We cannot get you spare parts for this anymore.".

So what we have here is a lot of the same thing we had trying to keep the old space shuttles running; people would be running around trying to find X86 processors (or was it 8088? ) trying to keep the darned thing running instead of updating the systems.

Aspiring to be a stick in the mud.

  1. Do we know they dumped the data into an AI engine without any framework around the data?
  2. Is it a requirement to be college educated and at least 24 years old to be able to detect non-sensical and/or incomplete data?
  3. Other than conjecture from unnamed experts in a Wired article that the COBOL theory is correct and that the DOGE staffers did not account for the COBOL issue, if such an issue is even in play here?
  4. The Wired article cites an IG report and the SSA's website. What reason do we have to believe that the IG report and SSA's statement are based on something other than the (apparently) rickety system?
  5. If someone listens to the Oval Office "briefing," the context is data quality and completeness so that there is reasonable (as in any business would require it) assurance that money is legitimately distributed. Musk does talk about another problem related to SSA where payments are going out with no identifying information. So we are to believe that code is in place to prevent payments to people whose calculated ages is unreasonably old (and what would be the code that deals with actual exceptions to whatever the cut-off age is; who's maintaining the hard coding if that is the way it is handled?), but the same agency can't put in simple code that requires any value in some fields for payment to be issued? Or maybe Musk and the sub-24 non-college people saw a bunch of identifying information but when they threw it in the AI machine it all came out blank?
  6. PSA: Wired is owned by Conde Nast, same outfit that owns Vogue (two Kamala covers), Teen Vogue, The New York, Vanity Fair, Them (LGBTQ+ publication), Glamour, Pitchfork, etc. Not exactly an objective bunch. Not to say Wired's content is 100% bunk, but merits a gimlet eye.

  1. Yes. I implement AI at work. To get it to work properly, it needs time and data to provide context. They have been in power a couple of weeks. We spend months getting the data and the AI models to work very well and that is when we understand the data. I 100% guarantee that these young kids don't understand data that is very, very different from department to department. Why? Because I hire these same kids and it takes a while just to educate these young people on how the data operates, even before they get in to build the models.
  2. No. But it is a problem when no one has expertise on doing this type of thing and they just jump into it. I gave a company a bunch of data and it took their seasoned team more than a week or two to build it properly. Besides, we all sit here and use ChatGPT and other tools like CoPilot and Gemini and we scratch our heads sometimes as we get nonsensical stuff out of it. And somehow, we have to believe the hasty process this thrown together team has put together does not have any hiccups?
  3. Yes. Go look around the internet. Go visit a COBOL reddit forum. It is surprising that the data issues is exactly the same number that Elon throws out in the oval office in front of the news. It is also very typical of a team quickly looking at data. You find all kinds of issues and inconsistencies and you have to go through it all and trace the data to confirm there are really issues. A 20 year old kid hasn't gone through this so many times to even realize what they don't know.
  4. It is not a rickety system? Not sure where that is coming from?
  5. These have already been debunked by industry individuals familiar with the system. Of course, Elon offers no proof.
  6. The issue is not who owns them, it is whether the data is correct. Like I said, go on a COBOL redit forum.

"....OK, what we have, at the very least, is government computers about a quarter century old running programs on Cobol. I've worked most of my career in electronics reliability, in particular with computers, and the standard warranty period is about five years. If you get ten years out of a system, you are doing really well, and part of my job is to tell the service reps to tell the customer "You have had this system long past warranty--update! We cannot get you spare parts for this anymore."

Bert, this isn't entirely true for core transactional system. First, most banks and insurance companies as well as many other, still run on systems like the Social Security Administration. These consist of a mainframe for hardware and operating system which is still manufactured and developed today and COBOL for the application language, which is still supported with ongoing development. Manufacturers maintain these types of hardware for a much longer period than what you are stating above. It is highly doubtful that any of the components are that old. Some people may view an old software program written in COBOL on an IBM mainframe as old fashioned. And they may be right. But most companies don't have the luxury of rewriting core transactional system from the ground up. It is extremely expensive to rebuild this type of system and the money is better spent elsewhere. I have seen many of my peers loose their jobs after trying to go down this road. COBOL and mainframes have a lot of advantages over today's systems. The SSA is not unusual in its approach compared to many companies like Bank of America, Chase, Allstate, State Farm.....

Still a lot of conjecture. Just on #1, breaking it down into 2 parts:


1a). Do we know we they dumped it into an AI engine? Any actual evidence of this, or are we just supposing that's what a bunch of sub-24 non-college people do? Don't even see where it is suggested in the Wired article (itself a conjecture-heavy piece).

1b). If 1a is in fact "yes" ("in fact" being important), do we know they provided no framework?

For what it's worth, I suspect they didn't use AI at all, but I'm only supposing. Not claiming to know one way or the other.

Lot's more could be said about the conjecture and assumptions, but not sure it will generate more light than heat.

For what it's worth, I suspect they didn't use AI at all, but I'm only supposing. Not claiming to know one way or the other.

I'm curious why you suspect that they didn't use AI. Isn't it logical to use the most advanced, up-to-date computer search technology available? You say you don't know one way or the other, yet for some reason, you're leaning toward a non-AI process that would take a lot longer time than what has passed. Even for AI, this is a short time-frame, so I can't imagine what other process you suspect might have been used.

We will know for sure, when some of the lawsuits make it through the courts, as many of them have evidence of AI use by DOGE. But many of them have detailed information on what is being done leveraging AI and Microsoft Azure.

AI/not-AI: Just because AI is the newest tool doesn't mean it is the best tool. Again, I know they are all sub-24-years, but my guess is they know enough about data analysis to know when a couple of simple parameters would highlight unusual records.

Not at all surprised that they might be using cloud platforms like Azure, which have AI capabilities but not exclusively so.

Rickety: the SSA IG found $72 billion of improper payments over 8 years. I'd call that rickety. Or so thoroughly unreliable that maybe there is reason to be suspicious of the data and the systems that supposedly make up for all the data deficiencies.

Fundamentally there is far more we don't know about this than we do know. I, for one, am content to wait for the facts, and am not at all concerned about DOGE's access to the data. Those sub-24-year-olds non-college people are likely much more competent than the folks who let SSA become so vulnerable (might we say rickety?).

Well, regarding the hardware, I guess it is possible that they're not just doing the "eternal scrounge" in the way NASA did for the space shuttle, but having worked military contracting, I've seen otherwise, and repeatedly. Plus, the IRS has been doing the "eternal scrounge" as they've failed to update their systems, so the notion that Social Security (which apparently stores death records in an old mine) does the same is not entirely foreign to me.

This is especially the case when Wired argues they're still using what is arguably a 1960s era scheme for saving a few bits on date fields. To me, it's stunning to think that you would have an agency that went through a ton of difficulty back in 1999 to get those fields adapted to a better long term format, but then when they theoretically got new hardware where the hard drives cost less than a nickel per GB, they didn't avoid confusion by updating the date field to a modern format.

Aspiring to be a stick in the mud.

Rickety: the SSA IG found $72 billion of improper payments over 8 years. I'd call that rickety. Or so thoroughly unreliable that maybe there is reason to be suspicious of the data and the systems that supposedly make up for all the data deficiencies.

I'm not convinced that all 72 billion was actually improper. If what they are saying about COBOL is true, then a bad date on a form could be read by a COBOL system as a date in 1875, putting the person's age at 150, when they are actually younger and are proper recipients of SS. This isn't rickety if one understands COBOL.

Fundamentally there is far more we don't know about this than we do know. I, for one, am content to wait for the facts,

Yet you present the 72 billion number as if it were a fact showing the systems unreliability.

and am not at all concerned about DOGE's access to the data. Those sub-24-year-olds non-college people are likely much more competent than the folks who let SSA become so vulnerable (might we say rickety?).

And I'm sure Hillary Clinton felt she was competent when she put government emails on private servers, yet there were real security issues with those private servers. At least Republicans then felt there were security issues regarding servers. Do we even know now what servers the DOGE team is using for the government data? Isn't this something DOGE should be immediately transparent about?

Bert,

New doesn't mean better. Why do so many companies still run mainframes and COBOL? Because it works. Spending billions to transition to a new platform, offers minimal benefit. So most of these companies are focused on improving the experience of the platforms and adding additional functionality leveraging new technology. I have been a technology executive in a number of companies that use mainframes and COBOL. And to be honest, migrating off of them is not the best use of money.