More Tales of Competence
This time: the Social Security Administration.
So, I saw this story referenced at LGM and it was so nutty I had to, well, Google it.
Via ABC News: New head of Social Security, hired from Wall Street, tells staff he had to Google the job when he was offered it.
The newly sworn-in head of the Social Security Administration told agency staff this week that when he was first offered the job in the Trump administration, he wasn’t familiar with the position and had to look it up online.
Frank Bisignano, a former Wall Street executive, said during a town hall with Social Security managers from around the country on Wednesday that he wasn’t seeking a position in the Trump administration when he received a call about leading the SSA.
“So, I get a phone call and it’s about Social Security. And I’m really, I’m really not, I swear I’m not looking for a job,” Bisignano said, according to an audio recording of the meeting obtained by ABC News. “And I’m like, ‘Well, what am I going to do?’ So, I’m Googling Social Security. You know, one of my great skills, I’m one of the great Googlers on the East Coast.”
The brazen indifference to expertise is something to behold.
I would hasten to add that this is a reminder that even if Trump ultimately proves to unsuccessful in some of his pursuits, he is doing real and lasting damage to key functions of the federal government.
Just stunningly stupid and irresponsible.
“I’m like, ‘What the heck’s the commissioner of Social Security?'” said Bisignano, who now oversees one of the largest federal agencies that’s responsible for distributing retirement, disability, and survivor benefits to more than 70 million Americans.
“Put that as the headline for the Post: ‘Great Googler in Chief. Chief in Googler’ or whatever,” said Bisignano, who throughout the meeting repeatedly bemoaned media leaks from within the agency.
On the one hand, improving service would be a good thing, knowing something about it all is kind of important.
While Bisignano, who previously served as chairman and CEO of financial technology company Fiserv Inc., brings experience managing large organizations and overseeing complex payment systems to his new role, he has no prior history working in government or with the Social Security system.
A Social Security official told ABC News that Bisignano is “working to transform the agency into a premier service organization.”
This, however, inspires little confidence.
Bisignano told the managers that they needed to believe that DOGE was “helping to make things better” even if “it may not feel that way.”
“Who’s heard of DOGE? Raise your hand, right? Your bias has to be, because mine is, DOGE is helping make things better. It may not feel that way, but don’t believe everything you read.”
He said DOGE would be involved in rebuilding the Social Security website and integrating artificial intelligence into the agency’s phone support systems.
The ongoing erosion of expertise at the highest levels of the federal government is concerning and will have long-term negative consequences.
BTW, there is audio if you want to actually here him say it.
Sounds like a perfect match for this administration. 🙁
“The brazen indifference to expertise is something to behold.”
I would go with “hostility” instead.
@Moosebreath: A fair edit.
Over a quarter century into the internet age, the ability to type a phrase into a search box is not something to brag about.
Seems like a dubious qualification.
But, okay.
I’ll challenge that…with the winner becoming the head of SS.
Well it could be worse, he could have been a weekend talking head on the One America Network that runs an Herbalife distribution network on the side.
@just nutha: What can I say. The man is a self proclaimed expert at “doing my own research”. A perfect qualification for the Trump administration.
My sense is that Bisignano is playing the “I’m so humble” game.
I can really appreciate the reasons for lack of trust in this administration, in DOGE, in Musk, and in anybody Trump might deign to appoint. This is a party that has repeatedly tried to destroy Social Security and loves to describe it as a “ponzi scheme”.
AND, the SSA’s job is simple, if enormous. It cuts checks, or direct deposits, and disburses them. It keeps employment records. It determines eligibility for disability. I think those are the main categories.
I expect their computer systems are severely outdated and this crushes their productivity. Updating government computer systems is really hard, because of the way that Congress allocates money for these things. (That’s a complex topic, but it’s why the US Digital Service was conceived by the Obama Admin).
So the most expected course forward for this is “stuff gets broken, people are hurt, and current administrators get roasted” After that, we will have no choice, though, but to modernize.
I think maybe 60-80 percent of customer support (if its fair to call it that) issues with SSA could be handled by a text-based ai, which I personally would far prefer to waiting on hold, or navigating a phone tree and getting tossed around from one office to another. Which is a thing that has happened, but not with SSA.
It sounds kind of awful, and yet, there’s a way this could work that would be better for everyone. Who knows if we will get that, though.
@Jay L Gischer: Even assuming this is true, that it’s possible to modernize Social Security, why take the risk, especially in a big-bang style modernization, which are almost always massive failures? Social Security is incredibly efficient today; its overheads are ~.5% of what’s spent. Its levels of waste, fraud, and abuse are non-existent. People are happy with Social Security. Having offices all over the place is a good thing, for a lot of reasons. What would “modernizing” it accomplish, exactly? Why does it need an AI?
@Jay L Gischer:
It is true that sometimes breaking things, in the long run, yields something better. I mean, in the original Hiroshima they didn’t even have WiFi. OTOH: Carthage.
@Kevin: I am not arguing in favor of this. Yes, that is a fairly low overhead. It could be lower. Customer experience is pretty good. It could be better, I expect.
Every year that we keep our stuff on old IBM mainframes the risk of some catastrophe increases, and the cost of fixing such a catastrophe escalates. Every year the number of people who know how to work on these systems effectively gets smaller.
Nothing here is insurmountable. My argument is more about observing what is happening, trying to “see through” they hype and emotion, and think about how things might play out. I’m often wrong.
The IRS, for instance, has badly outdated computer systems. There have been attempts to upgrade them which failed miserably. We – taxpayers and citizens – need those computer systems to continue to work effectively and reliably. We need to have a large pool of people who know how to tell when something is wrong, and how to fix it.
It is a fact of politics that crisis == opportunity. We might see that in the government IT domain soon.
@Michael Reynolds: Breaking rules is a hallmark of creativity. I think this is something that you are well aware of.
The idea that creation is linked to destruction is enshrined in many different parts of human wisdom.
@Kevin: I see I didn’t answer “Why does it need AI?”
I would avoid the word “need”. I think there are advantages to it, though. Whenever I engage with customer support on something, I spend a good 15-20 minutes dealing with crap that appears to be focused on solving issues I don’t need to solve. Eventually, the system will recognize that I have a problem that is unusual for them, and I get directed to someone that can help me. (or told that there is no way to help me, sigh.)
I have two takeaways:
1. I would much rather do this with text chat, as I can sit at my computer and do other things while waiting.
2. Most of the calls/requests can be addressed by very mundane answers, answers which I’m sure the humans doing the answering of find stifling and boring. Things like, “What is my balance?” or “What is my account number?” and “How do I pay my bill?” and so on.
AI would let people get a much faster answer to Category 2 issues. A very limited-domain AI like this would very likely NOT give wrong answers to mundane questions.
AI would probably get me to the person I was going to talk to anyway much faster as well.
But no, we don’t “need” it. Probably the SSA should not be the first organization to try this stuff out, either.
Trump has two areas of competence: grifting and bullying.
@Jay L Gischer: I’m biased, because I build those IBM mainframes, but the mainframes that the IRS, Social Security administration, Treasury department, and other entities that have large data processing needs use are not outdated. They’re cutting edge machines with technology in them that is ahead of anything you can get in commodity hardware. There’s a lot of stuff they aren’t good at, but at transaction processing, they are by far the fastest, most efficient, most secure hardware available. (There is, I’m told, a system from the 1970s running somewhere in the bowels of the DoD that is crucial to their accounting/inventory management, which is held together with duct tape and bailing wire, and would probably spill all its secrets if looked at funny, but that’s an exception.)
I’ll be the first to admit that they’re not straightforward to use, but they aren’t alien technology. They’re like industrial CNC machines, or container ships, or SCADA systems, or whatever. They’re specialized hardware that requires specialized training. At the high end, many attempts have been made by many companies to move off of them, and it’s never worked. Maybe DOGE will prove me wrong, and their current effort to rewrite the Treasury department software in Python and mySQL will work, and perform well, but I’m guessing neither of those things are going to happen. There’s sixty years of knowledge embedded in those systems.
I’m not saying they don’t need to be modernized, to make the data on them more accessible, but there are known strategies for doing so, which is not what’s happening right now. And unless there’s continuous investment into the systems, which is unlikely, they’re going to be written in programming languages that aren’t taught in school in a decade again. Trust me, I really understand the appeal of burning everything down and starting from scratch, as we’ve been working on modernizing our development and test environment for a decade, and it’s slow going. But the ROI of completely rebuilding a working system is never justified; you have to do it in pieces, over time.
@Kevin:
I used to work for a large bank. The guy in the cube across the aisle from me was the COBOL guy. A lot of the core systems this extremely large bank that isn’t Hell’s Embargo relied upon for their very basic day-to-day operations were decades old , the hardware and the software, and people with those skill sets were in short supply.
I literally could not have done my job without him. Brian produced the data I needed in a format I could actually use and load into a modern DBMS with minimal monkey business. He even frontloaded the integration of several disparate back-end systems into one smooth feed. He was a genius.
You are so very correct. Trying to untangle decades old code and legacy systems is damned hard, and is one of the most delicate, most painstakingly difficult things you can imagine. You cannot fuck up it up at all, even a little. You can’t crater a multi-billion dollar company because you introduced a new recursive loop.
I think the proper procedure is to peel off the systems and functionalities that got added to the core in reverse order. Can we peel off this widget that recalculates how much mortgagees are currently due if borrowers intentionally overpay?
Yes. But extraordinarily carefully. Test. Test. Test. Migrate over only if you are extremely sure. And have a fool-proof back up plan if and when it goes sideways.
My last corporate job was rolling out new purchased, licensed software updates to about 104,000 desktops and laptops. We knew exactly who had the existing version, but we didn’t know what they were doing with it. I didn’t actually do it, I managed the folks who did. It was deeply unsatisfying and the reason I went to consulting on things I was actually good at.
We’d send out questionnaires asking if there were processes built around said software. People didn’t lie, per se, but would omit information, or not know what their folks actually did.
By far the biggest kerfuffle was updating Microsoft Excel. People build extremely convoluted processes doing very complex shit in Excel that become core systems for that department over time. X feeds Y that Z uses the output from Ffs, just contact IT – chances are really high that they can automate that process, host it, make it more efficient, provide back-ups, and give you an SLA.
It is scary the amount of core stuff that happens off the radar that is key, core function. If Marsha in Accounting gets hit by a bus, they can no longer do invoicing. That’s bad.
Some folks are blind to risk analysis.
Excel migration to the new version was way harder than Access. Access users were much more upfront about what they used it for, and how it was integrated into the whole.
We pushed updates Sunday night 12 pm once or twice a quarter. We sat around, bullshitted, watched the progress. Crossed our fingers, hoping we aren’t about to break a big thing.
Monday 8am was make or break. We had roll-back plans, obviously, but that would likely be overnight. Push nights, and the next morning were very stressful. I’d go into work 10pm on Sunday night and maybe get home to crash 16 hours later.