New Jersey’s governor recently went on TV asking for volunteers who know COBOL, a more than a 60-year-old computer programming language, because the state’s unemployment benefits system runs on a decades-old mainframe that can’t keep up with the volume of requests for financial help.
New Jersey isn’t alone, however. Several states, not to mention a lot of financial institutions, are all at the mercy of these very old systems running on a programming language that isn’t even taught in schools anymore. I spoke with Joseph Steinberg, a cybersecurity consultant and author of the book “Cybersecurity for Dummies.” I asked him if this is the digital equivalent of crumbling roads and bridges. The following is an edited transcript of our conversation.
Joseph Steinberg: First of all, there are many financial institutions that use mainframes with COBOL for back-end processing of transactions and batch processing. I think they probably have things in a lot better shape because they have the resources in-house. In some cases, they’ve replaced the systems at great expense over recent years. In some cases, they’re doing it now. And in some cases, they’ve made the decisions to actually train people to be able to support these systems going forward. When it comes to the government — I would say probably even worse in the state governments than in the federal — you have many systems that seem to hang around for a very long time. One of the questions that I posed in the last week is simply if, God forbid, this pandemic continued for several more months and all Americans had to vote in the presidential election remotely, do all 50 states actually have computer systems that could accommodate that volume of ballots going out and in to be counted, processed in a secure fashion, in any reasonable time frame? I’m afraid of what the answer to that question might be. We simply have a lot of legacy systems involved in various state processes, and they were never designed to accommodate mass loads that could suddenly happen now. It’s a lot harder to scale these legacy systems. Newer technologies scale much better. I’m afraid of what’s out there at the state level, and I’m sure there’s some in the federal government, but after this pandemic is over, it’s really going to be time to take a serious look at a lot of these older systems and evaluate whether they really are reliable only because we haven’t had to make changes, or they’re really reliable? I think in many cases, we’ll find it’s the former.
Molly Wood: It seems like this is an invisible infrastructure problem in America, [which] in some ways is as expensive and difficult to fix as roads and bridges.
Steinberg: It’s as expensive and perhaps, in some ways, more dangerous because you don’t notice it until the problem happens. You have people who desperately need money because their work suddenly ended. They place an unemployment claim, and they can’t get it processed — that can be dangerous. People could be low on food and other supplies, and it’s not like they can easily obtain them now, either. I think that’s precisely the issue with these older systems. Often we assume that they’re reliable until it’s too late, and the problem has hit, and at that point you’ve got a tactical problem — how do we process these unemployment claims? — and a strategic problem — how do we make sure this never happens again in the future?
Wood: You’re saying that even if we were able to bring a lot of programmers out of retirement, that these systems simply may not ever be able to handle the load that is being put on them now?
Steinberg: It may be possible to scale them up with improved hardware, but that’s going to be very expensive, No. 1, and they don’t scale as simply as more modern systems do. The fact that you’re hiring programmers out of retirement — people don’t live forever — you’re not going to have an endless supply of programmers who programmed in the ’70s in the workforce. I’m not saying that these systems aren’t reliable. They may be very reliable for what they do now, but we have to accept the fact that systems may need to be changed and may need to scale up suddenly, and we may not be able to do it with the existing technology and human resources.
Wood: I have been at companies that have replatformed, and it is hard and it’s really expensive. What does that look like at a state and national level? What investment, in terms of both technology and money, would that take?
Steinberg: The thing to keep in mind is there’s also a political problem with things like this. If you’re a state elected official, and you have to budget a large amount of money for replacing a system, no user of that system initially benefits from it. So you need to have the political will, which I think after this, we will, to invest the money before problems happen, understanding that you’re investing for the future. We should have our infrastructure set up that if there is an emergency in which people are dying, the governor should be focused on people, not on technical glitches in an unemployment system.
Wood: This is a speculative and vaguely nightmare scenario. At what point do you think companies like Amazon, Microsoft or Google might approach states and the federal government and say, “Maybe you need a private solution, a privatized infrastructure for some of this”?
Steinberg: I don’t know if they already have. They may have approached them years ago and been told that the system is working, i.e., “We have no need to replace it.” I don’t know. I’m sure that after this is all over, as the [New Jersey] governor himself said, we’re going to have that post-mortem. After that post-mortem, there’s going to be evaluations of many different options as to how to proceed. Some of those will involve utilizing third-party resources, potentially. By the way, you could have a cloud that’s run by the government — it could be a private cloud system, it could be their own infrastructure. There are many, many options, but moving away from the COBOL code running on the mainframe to a more modern platform might allow for greater scalability suddenly if needed, as well as for easier maintenance if code changes need to be made. You’re certainly not going to need to go begging for programmers at a press conference.
Related links: More insight from Molly Wood
Here’s the link to a 2016 report from the Government Accountability Office, which found that among other things, the Department of Defense uses 8-inch floppy disks to coordinate some of the country’s nuclear forces. You can see why they want to get that big cloud project moving, I guess.
In other infrastructure news, MIT Technology Review has a piece this week about how, generally speaking, the internet is holding up shockingly well in the midst of our transitions from work and school to home. It points out that cloud service providers, and also streaming media and telecom companies — Amazon, Comcast, Netflix, Microsoft, Google — put a lot of money into new internet infrastructure in the past decade or so in order to keep delivering their services. The current crisis, the piece says, is generating even more investment. If only we could do the same for the power grid, bridges, the state and county mainframes.
It was a brutal week at Yelp. The company that’s basically built on small businesses is laying off 1,000 people and furloughing another 1,100, according to Axios. Yelp’s CEO will take no salary for the rest of the year and is cutting executive pay by 20% to 30%.
On Thursday, Microsoft released a report about remote work trends based on data from its Teams product. The company, of course, reported a huge increase in the number of people using Teams. It also has some interesting data. For example, the report notes that people are using video in their meetings a lot more than they ever did before — 1,000% more in March, in fact — but the U.S. lags a little bit in terms of video calling. In Norway and the Netherlands, about 60% of calls include video, but only 38% of American meetings have video on. I’m just going to confess that I refused to join a meeting by video Thursday because I can no longer stand the sight of my own forehead. I see you, America. Or rather, I only hear you.
The future of this podcast starts with you.
Every day, the “Marketplace Tech” team demystifies the digital economy with stories that explore more than just Big Tech. We’re committed to covering topics that matter to you and the world around us, diving deep into how technology intersects with climate change, inequity, and disinformation.
As part of a nonprofit newsroom, we’re counting on listeners like you to keep this public service paywall-free and available to all.