Every skilled trade that survived the twentieth century survived it because it fought for a floor. Electricians, plumbers, pipefitters, machinists. Apprenticeship, journeyman, master. Licensure that made it illegal for an employer to direct you outside code without exposing themselves legally. A union that organized around the work itself, not around any single employer, and that outlasted depressions, world wars, and the collapse of entire industries around it.
IT never did that. It skipped straight from hobbyist to knowledge worker without ever building the structure underneath. No licensing body. No apprenticeship system that gives a twenty-five-year veteran and a six-week bootcamp graduate different standing. No collective floor beneath which an employer cannot push you. Just job titles, vibes, and whoever is willing to answer the phone at 11pm during the pledge drive without comp time.
IT got the twenty-first century’s demands on the nineteenth century’s protections. Which is to say, none.
The frameworks arrived stripped of what made them survivable
ITIL, Kaizen, Lean, Scrum. These didn’t originate in a vacuum. American manufacturing absorbed Lean decades after its unions had already won the fights that gave the shop floor leverage. The continuous improvement culture arrived on top of protections that were already load-bearing.
Import that same framework into IT and you get the demand without the floor. Continuous improvement, but nothing improves your standing. Kaizen boards and sprint retros measuring throughput on a workforce with no grievance procedure, no seniority protection, no external body that can say a directive violates standard practice and must be refused. The machinery of operational excellence, bolted onto a trade with no organized labor underneath it to absorb the pressure.
Kaizen without a union is just surveillance with a whiteboard.
The fork happened in the same war
IT didn’t skip the fight because the opportunity never came. It skipped the fight because the people doing the founding technical work were classified out of eligibility for it before the war even ended.
World War II produced two technical workforces that came out of the same conflict, trained on the same wartime urgency, and landed in the postwar labor market within a few years of each other. Signal Corps radio and radar technicians came home to an industry ready to absorb them. RCA and the IBEW built a formal apprenticeship program together, television servicing, that took wartime electronics training and turned it into a licensed trade with a union standing behind it. That’s the direct technical lineage between a war and a floor.
The other workforce was the first programmers. ENIAC’s Army-funded team at the University of Pennsylvania drew from a pool of nearly a hundred women already doing ballistics calculations by hand. Six of them were selected to program the machine. The Army’s own classification for this work was sub-professional. Not a trade. Not even a profession. Foundational technical labor, coded clerical from the start, because the people doing it were women and the category “skilled tradesman” was never built to include them.
Radio and radar came out of the war as a trade. Computing came out of the war as a clerical function that later got promoted into knowledge work without ever passing through the stage where a trade organizes and wins a floor. IT didn’t miss its 1937. It was defined out of the room where 1937 could happen.
One classification. That’s the whole difference.
Professionalization was chosen instead of organization
The fork didn’t close after the war. It widened, on purpose, over the next two decades, and the choice made in that widening is the actual reason IT ended up where it did.
Programming in the early 1950s was still coded as clerical labor, a continuation of the ENIAC classification, and it was still substantially staffed by women. Historian Nathan Ensmenger has documented what happened next in The Computer Boys Take Over: rather than organize the way the radio and radar technicians organized, the emerging management layer around programming pursued professional legitimacy instead. New credentialing pathways, university degree programs in computer science, personality tests built to screen for the kind of person who supposedly preferred machines to people. The tests didn’t just filter candidates. They reshaped who the field imagined itself as employing, and the field’s self-image shifted from clerical and female to technical and male over the course of roughly two decades.
This is the second half of the fork, and it matters because it shows the choice was made twice. The first time, at the level of the Army’s own job classifications, computing labor was defined out of trade status. The second time, at the level of the industry building its own professional identity, computing management chose credentialing and personality-testing over unionizing, chose to chase status as a profession rather than leverage as a trade. Professions confer prestige. Trades confer floors. IT ended up with neither, because it spent the years when it could have built a floor building a resume instead.
A profession is a floor you rent. A trade is a floor you own.
The generational blind spot is structural, not attitudinal
Younger IT workers came up entirely inside the framework-of-the-month era. Sprint stand-ups, OKRs, Kaizen boards, ITIL change tickets. They have never seen what a trade with actual leverage looks like, so there is nothing to compare it against. It reads as normal because it is the only water they have ever swum in. This isn’t a knock on them. It’s what happens to any workforce that never had its own 1937, its own Flint sit-down strike, its own moment where the trade organized and made continuing to operate without a floor more expensive than granting one.
Normal is just whatever nobody organized against.
Happy is not the same as protected
Here’s the objection, and it’s a fair one: most IT workers aren’t miserable. Plenty of people in this field like the work, make good money, have a manager who treats them like an adult, and would tell you the whole framing above is overwrought. A job is a job. Nobody’s suffering. What’s actually being asked for here.
That’s true for a lot of people, most of the time, and it’s beside the point. A floor isn’t insurance against a bad mood. It’s insurance against the one bad year. The manager who gets replaced by someone worse. The diagnosis that means you need an accommodation your employer would rather not grant. The economic downturn that turns “we value your contributions” into a layoff list with your name on it and no severance formula you had any say in. Nobody needs a floor on the good days. The entire function of a floor is what happens on the day it isn’t a good day, and that day is not optional. It arrives for almost everyone eventually, on someone else’s schedule, not yours.
Ask an electrician why the license matters and they won’t describe their best day at work either. They’ll describe the day a foreman told them to skip a step to hit a deadline, and the fact that refusing didn’t cost them their job, because refusing was protected. That protection isn’t there because electricians are unhappier than programmers. It’s there because someone decided, decades ago, that a good day at work shouldn’t be the only thing standing between a worker and getting run over. IT decided the opposite, or rather, had it decided for it, twice, in the 1940s and the 1960s. The people who are fine right now are fine because they haven’t hit their bad year yet. That’s not a rebuttal to the argument. That’s the argument, restated as a countdown.
What a license would actually do
A licensed electrician can refuse a directive that violates code and the refusal itself is protected. The license is not a credential for a resume, it’s a legal backstop. It converts “I told you to do it” from a shield for the person giving the order into evidence against them. IT has no equivalent. When a director edits a document shortly before claiming not to have reviewed it, there’s no licensing board that cares. When a final written warning lands in the same window as a protected accommodation conversation, there’s no journeyman-level standing that makes the timing itself actionable on its face. The credentials in the audit log attach to whoever executed the action, and the license that would separate “I was directed to do this” from “I am personally exposed for doing this” simply doesn’t exist.
Code and license aren’t the same thing, and IT is missing both. The code is the technical standard, the National Electrical Code, a plumbing code, the correct and legally adopted way to do the work. The license is the legal standing to invoke it, the thing that turns “this violates code” from an opinion into a protected professional judgment an employer can’t retaliate against. IT has fragments of the first: ITIL, NIST, SOC 2, ISO 27001, a pile of competing frameworks that function more like rival style guides than a single body of standard anyone adopted into law. It has none of the second. There’s no governing code an IT worker could even point to with the standing of a building inspector, and no license that would matter if one existed. Two missing halves, not one.
A ticket documents. A license protects. IT built the first and called it the second.
Who controls the license is the whole point
Here’s where this stops being a metaphor and has to get specific: IT already has credentials. CompTIA, Cisco, Microsoft, AWS, a dozen others, an entire industry of study guides and renewal fees built around them. None of them function as a floor, and the reason isn’t that they’re not rigorous enough. It’s that the vendor controls them.
A vendor certification can be redesigned, devalued, or discontinued by the same company that sells it, the moment doing so serves the product roadmap. It has no standing to tell an employer that a directive violates practice and must be refused, because the body issuing it isn’t accountable to the worker holding it, it’s accountable to the company selling the platform the certification is stamped on. That’s not a license. That’s a subscription with an exam attached.
The distinction that matters isn’t whether IT has credentials. It’s who sits on the board that issues them. Electricians don’t get licensed by whoever manufactures the wire. A license that means anything answers to the trade, to the people doing the work, not to whichever vendor has the most market share this decade. Put the license in industry hands, not market hands, and it starts doing what a license is supposed to do. Leave it where it is now, and it’s just another certificate that expires the moment it’s no longer profitable for someone else to keep issuing it.
Certifications aren’t licensure, and the field already knows it
The proof that IT understands the difference is sitting in its own vocabulary. Ask anyone who was hiring sysadmins in the early 2000s about the “paper MCSE” and watch the face change. Microsoft’s certification program was so heavily gamed by brain-dump sites, memorized answer keys sold to anyone with a credit card, that the industry coined its own insult for the result: a person with the credential and none of the skill. Resumes wallpapered HR offices with a piece of paper that had stopped meaning anything the moment enough people learned to pass the test without learning the job. Microsoft eventually responded by adding performance-based simulation questions in 2005, forcing test-takers to actually configure the system instead of answering multiple choice about it. That fix came from the vendor, on the vendor’s timeline, a decade after the damage to the credential’s reputation was already done.
Notice what the industry did with that lesson, because it’s the same lesson as the license question and it got answered the same wrong way twice. Faced with a credential that had been hollowed out, IT didn’t build a body that could vouch for competence and stand behind that judgment. It built more gatekeeping into the hiring process instead: live coding tests, take-home projects, whiteboard interviews, portfolio reviews, the informal and often unpaid burden of proving yourself all over again at every single job change because no institution exists whose word anyone trusts. A bootcamp graduate with three solid GitHub repositories will beat a four-year computer science degree in plenty of interviews, and a computer science degree will beat both against a hiring manager who just wants someone who already worked at a company they’ve heard of. None of that is irrational. It’s what happens when a field has no accountable credentialing body and every employer has to independently reinvent the wheel of figuring out who’s actually competent. The paper MCSE didn’t get fixed. The cost of the fix just got pushed onto every worker, at every interview, forever, instead of being solved once by an institution built to solve it.
What documentation actually replaces
What documentation, minimum disclosure, and a formal complaint filed through a state civil rights agency accomplish, for one person, after the fact, at real cost, is what a grievance procedure would have handled automatically, in real time, as a matter of course, for a licensed trade with an organized floor beneath it. The absence of a functioning HR structure during an active employment matter isn’t a personal failure to escalate. It’s what it looks like when the informal, ad hoc substitute for a real protection structure runs out of people to escalate to, because the org never had a formal one to begin with.
This isn’t a story about one bad manager or one dysfunctional nonprofit. It’s what the entire field looks like without a floor. Every layer above the technical worker optimizes for its own liability exposure, not for good technical or ethical decisions, and the person actually executing the work absorbs both the operational risk and the fallout when that optimization fails. That’s not a bug in this particular organization. That’s the default condition of a trade that never organized.
Documentation is what you build when nobody built you a floor.
Four objections worth taking seriously
Before closing, four pushbacks earn a direct answer rather than a dismissal.
The first is that this isn’t an IT problem, it’s an American labor problem. Taft-Hartley, at-will employment doctrine, and forty years of union decline flattened the floor under most of the American economy, not just IT. Marketing has no floor either. Retail has no floor. That’s true, and it doesn’t erase the specific IT story, it explains the ceiling everyone’s working under while leaving open why IT ended up at the bottom of it instead of somewhere higher, the way nursing did, the way teaching did in states that kept collective bargaining. The general decline set the ceiling low for everyone. The WWII fork and the professionalization choice explain why IT sits at the floor of that low ceiling instead of a few rungs up it. Both things are true at once.
The second is offshoring. An electrician’s work is bolted to a building. IT’s isn’t, and a licensing requirement that raises the cost of hiring domestic labor hands an employer a lever no electrician ever had to worry about, move the work somewhere the license doesn’t apply. This one doesn’t have a clean answer, because it’s a real structural difference between the trades and knowledge work, not a rhetorical trick. What can be said is that the absence of a floor hasn’t stopped offshoring so far, so treating a floor’s presence as the thing that would finally trigger it gives the current arrangement credit it hasn’t earned. And regulated corners of IT already prove partial versions of this work: healthcare IT under HIPAA and financial IT under SOX and PCI both require accountable, often domestic, staff for exactly the highest-risk work, without those industries collapsing into total offshoring. A floor doesn’t have to cover every seat in the building to be worth building.
The third is the one this essay has to answer honestly, because it already handed a critic the ammunition: licensure is what pushed women out of programming the first time. The Cannon-Perry personality test and the IBM Programmer Aptitude Test were both credentialing systems, and both were used to reshape who got to call themselves qualified. Someone could read every section above this one and conclude that what’s being proposed is a new gatekeeping apparatus wearing a union’s clothes. The answer isn’t that credentialing is safe. It’s that those two tests were built and administered by an employer and a vendor, answerable to nobody but themselves, which is exactly the failure mode the “who controls the license” section above is arguing against, not for. A licensing board seated by working IT professionals, accountable to the trade rather than to whoever’s selling the exam, is a structurally different animal from a personality test IBM wrote to filter candidates on IBM’s terms. That doesn’t make the new version incorruptible. Any body that gates entry to a trade can be captured and turned into exclusion again, and that risk is real, not hypothetical, given the field’s own history. The alternative to that risk isn’t no floor. It’s a floor built with worker control over its own gate as the design requirement from day one, not an afterthought bolted on once the exclusion problem shows up again.
The fourth is that IT isn’t one trade, and licensing something that broad is a category error. Fair, and worth taking apart by discipline instead of waving off. A sysadmin or sysops worker carries direct operational responsibility, patching, backups, uptime, the person who gets the 2am call and the one whose name is on the change ticket when a maintenance window got skipped to hit a deadline. A devops worker sits at the seam between writing the code and running it in production, which means absorbing blame from both directions, “you broke prod” from ops and “you’re slowing down the roadmap” from engineering, without the testing safeguards developers get or the maintenance windows sysops gets. An infosec analyst does the clearest version of the electrician’s refusal from earlier in this essay: flagging that a deployment violates security policy, getting overridden by a director who wants to ship anyway, and then being the name in the audit log if the vulnerability gets exploited later. Three different jobs, three different daily realities, and the same exposure running under all of them, directed to work outside of what the job actually calls for, and absorbing the liability alone when that direction turns out to be wrong. That’s not a case for one undifferentiated IT license any more than the building trades run on one undifferentiated construction license. Electricians, plumbers, and pipefitters are three separate licensed trades under one broader umbrella, not one trade wearing three hats. IT not being one trade doesn’t undercut the argument. It’s the argument for building several floors instead of pretending one flat certificate could ever cover a sysadmin, a devops engineer, and a security analyst equally.
IT is a tradecraft. It has the complexity, the consequence, and the specialization to justify the same floor every other skilled trade fought for and won. It just never had the fight.
It had two chances to have it, and both times something else was chosen instead. The Army classified the ENIAC programmers as sub-professional in 1945, choosing a clerical designation over a trade designation for the workforce that built the first computer. The industry that grew up around that workforce spent the following two decades choosing credentialing and prestige over a union. Neither choice was accidental and neither was inevitable. A different set of decisions, made in the years right after the war and in the years the aptitude tests were being written, produces an IT worker today who carries a license instead of a job title, and whose employer needs a reason before they can be directed outside of it. That worker doesn’t exist. The choices that would have produced them were available, and were declined, twice, by people who are no longer around to answer for it. The people who are around are the ones still doing the work without the floor.
Two chances. Zero floors.
Further reading: the IBEW’s 1950 profile of RCA’s jointly-run television servicing apprenticeship program is documented in the IBEW’s own historical archive at ibew.org. The classification of ENIAC’s original programmers as “sub-professional,” and the pool of nearly 100 women they were drawn from, is documented by the U.S. Army’s history office. The professionalization and masculinization of programming in the 1950s and 60s, including the Cannon-Perry personality profile and the IBM Programmer Aptitude Test, is documented at length in Nathan Ensmenger’s The Computer Boys Take Over (MIT Press, 2010). The “paper MCSE” problem and Microsoft’s 2005 addition of performance-based exam questions is documented in trade press coverage from Redmond magazine and Certification Magazine dating to that period.