top of page

The Missing Layer: The FAA Lost About 940 Journeyman Technicians in 11 Years, Has No Long-Range Plan to Replace Them, and Just Repeated a 2009 Failure. What the Evidence Says, and What It Doesn't

Writer: Patrick Duggan
Patrick Duggan
11 minutes ago
6 min read

Before you cut, outsource or "streamline" the people who keep a system running, write down what they know that nobody else does. Then count how many of them are within five years of retirement. If the second number scares you, the first list is your real risk register.


Last week six big systems went down in six days, and none of it was an attack. In our series on it we kept landing on the same pattern: a change, a forgotten lesson, a repeat. This post asks the obvious next question. Is the layer of experienced people who remember those lessons disappearing? We went looking for evidence both ways. It holds up well for one case, the FAA, and stays unproven for the rest. Here is all of it.



The FAA: the evidence is strong


The experienced middle is shrinking. In testimony to the House aviation subcommittee on March 4, 2025, Dave Spero of PASS, the union for the FAA's systems specialists, gave the agency's own workforce numbers:


  • Technicians in the 2101 job series fell from 5,303 in 2014 to 4,871 in 2025.

  • Journeyman-level technicians, the experienced ones, fell from 3,638 to 2,699. That's 939 fewer.

  • Trainees rose from 5 percent of the workforce to 14 percent.

  • 34 percent are 55 or older; 14 percent are 62 or older.

  • The FAA's staffing model shows a shortfall of at least 800 systems specialists.

PASS is the union, so it has an interest in the conclusion. The numbers come from the FAA. PASS's argument is the part that matters for this post: the FAA does not hire ahead of retirements, so the experience walks out the door before anyone can learn from it.


Nobody is planning for it. The Department of Transportation's Inspector General reported in May 2023 (AV2023027) that the FAA has ten-year workforce plans for air traffic controllers and safety inspectors, but not for the maintenance technicians who keep the equipment running.


The same failure, sixteen years apart. In November 2009, a contractor's router configuration reached 129 facilities, blocked about 75 percent of the FAA's telecommunications routes and delayed 820 flights. The Inspector General's letter to Congress (CC-2010-012, June 2010) said the FAA "tends to have a reactive, rather than proactive approach." According to GovInfoSecurity, an independent review in 2010 recommended tools to automate router configuration and a model of the network to test changes on before they went live, and the FAA accepted them. In November 2026, data flooded the same kind of network during a router replacement on a Friday afternoon. The FAA says about 12,462 flights were canceled, and has since switched off a broadcasting function on more than 60 routers. GovInfoSecurity reported that tracing the source took about 45 minutes, partly because maintenance is still logged on paper. Whether the 2010 fixes were ever fully built is, by every account we read, an open question.


Contractors keep making the change. The 2009 router was configured by a contractor's engineer. In January 2023, the FAA said contract staff unintentionally deleted files while fixing a database, and the NOTAM system went down. The FAA's network is run by a contractor. None of that makes contractors the problem. The question is whether the FAA kept enough of its own experience to check their work.


What we can't say: that staffing caused the November 2026 failure. The root cause isn't final, and the router maker says its routers "appeared to work as intended." GAO's September 2026 report on the FAA's modernization program (GAO-26-107992) points to schedule and cost planning, not workforce. And a separate 2024 GAO review traced the FAA's unsustainable equipment to parts and funding, not staff.



Everywhere else: plausible, not proven


We looked just as hard at the other outages, and the evidence doesn't carry the same weight.


  • Microsoft. Its preliminary review blames a bug introduced in July that sat dormant for months, load tests run on weaker hardware than production, and 14 percent of capacity already offline. Microsoft did cut thousands of jobs in 2025, but we found nothing tying those cuts to the teams that run Microsoft 365.

  • Chase. JPMorgan spends about $18 billion a year on technology and runs a large in-house engineering organization. It's hard to call that hollowed out.

  • The NSA. Nextgov reported the agency cut about 2,000 civilians in 2025, mostly on anonymous sourcing, with no breakdown of IT staff and no link to the outage.

  • Schwab cut 5 to 6 percent of its staff in 2023. There's no link to Tuesday's outage, and no cause has been given.

The broader research cuts both ways. Google's own analysis of its postmortems (2010 to 2017) puts code and configuration pushes at the top of the trigger list, and staffing isn't even a category. The Uptime Institute says outages per data center site have fallen five years in a row. On the other side, Uptime's annual surveys, as summarized in trade press, say failure to follow procedures is a rising share of human-error outages, and that the talent shortage has worsened. Academic work on software teams (Mockus, 2010) found that when experienced people leave, customer-reported defects go up. The effect is real, but smaller than the effect of plain complexity.


So the honest line is this. At the FAA, a measurable loss of experienced people and a contractor's change repeated a 2009 failure, because the 2010 safeguards didn't stick. Everywhere else the record shows change and complexity risk, and staffing is a plausible contributor we can't prove.



Patrick's take


This part is opinion. Mine.


I was always a build-versus-buy guy, and I always came down on build, for exactly this reason. The journey and the learning were the point. When you build it, you know where the bodies are buried, and so do the people next to you. When you buy it and fire the people who built the last one, you get a vendor, a contract and a support portal. What you lose is the person who remembers why the router config looks weird.


That's the missing layer: the people who actually care whether it stays up. You can't see them on an org chart. They're the engineer who says "we tried that in 2009, it took down 129 facilities." When you replace them with whoever is cheapest this quarter, the lesson doesn't disappear right away. It just stops being enforced. Then one Friday afternoon it comes back.


Measles is making a comeback for the same reason. The vaccine still works. The people who remembered why it mattered stopped being in the room.


The FAA numbers say this is real in at least one place that matters. I'd bet it's real in more. I can't prove that yet, so I'm telling you it's a bet.



What to do this week


  1. List your carriers. For each critical system, name the person who knows why it is configured the way it is. If the answer is a vendor, or nobody, write that down.

  2. Count the retirement cliff. How many of those people could leave in five years? Who are they teaching?

  3. Pair before you cut. If a role is being outsourced or eliminated, require a written handover and a shadowing period first. It's cheap insurance.

  4. Make lessons executable. A lesson that lives in someone's head leaves with them. One that lives in a pre-change checklist, an automated config test or a change freeze rule stays. The FAA accepted that in 2010. The question is whether it was built.

  5. Check your contractors' changes with your own eyes. Outsource the work if you must, but keep someone in-house who can say no to a change.


Sources


  • Testimony of Dave Spero, Professional Aviation Safety Specialists (PASS), House Transportation and Infrastructure Committee, Subcommittee on Aviation, March 4, 2025.

  • DOT Office of Inspector General, "Opportunities Exist for FAA to Strengthen Its Workforce Planning and Training Processes for Maintenance Technicians," AV2023027, May 2, 2023 (as summarized; we did not read the full report).

  • DOT Office of Inspector General, letter CC-2010-012 on the November 19, 2009 FTI outage, June 17, 2010.

  • GAO-26-107992, "Air Traffic Control Systems: Ambitious New Modernization Effort Needs to Improve Cost and Schedule Planning," September 2026. GAO-24-107001, September 2024.

  • FAA statement on the NOTAM outage, January 19, 2023.

  • GovInfoSecurity, "Anatomy of FAA ground stop," November 2026.

  • BleepingComputer, "Microsoft says months-old bug caused global Microsoft 365 outage," November 16, 2026.

  • Nextgov, on NSA staffing cuts, November 4, 2025.

  • Google, The Site Reliability Workbook, Appendix C, "Results of Postmortem Analysis."

  • Uptime Institute, Annual Outage Analysis 2026 (public summary); survey findings via trade-press summaries.

  • Audris Mockus, "Organizational volatility and its effects on software defects," FSE 2010 (via secondary review).

  • Our series: The week the apocalypse didn't show up, Forgot about 2009, and our FAA scorecard.



How do AI models see YOUR brand?

AIPM has audited 250+ domains. 15 seconds. Free while still in beta.


Comments

Rated 0 out of 5 stars.
No ratings yet

Add a rating
bottom of page