The transformer on the workbench is a Jefferson Electric, a ten-pound slab of iron and tar-poured housing that smells faintly of ozone and ancient dust when the current finally hits the coils. It is the hidden heart of a neon sign, the part that steps three-digit house current up to the required to make neon gas dance in a glass tube, yet the customer never asks about the transformer.
They ask about the shade of “Horizon Blue,” they ask about the flicker of the “E” in the “OPEN” sign, they ask about the price of the hand-bent glass. They sign the check because the glow is beautiful, never realizing that if the transformer is under-specced or incorrectly phased, the beautiful glow will turn into a rhythmic buzzing that eventually melts the wiring inside their walls.
The approval email from Dana arrived at , a Tuesday afternoon where the humidity in the office felt heavy enough to stall a cooling fan, and the text was as brief as a weather report. “Total looks reasonable, approved,” she wrote, and with those four words, she green-lit a purchase order that would effectively paralyze the warehouse’s new scanning system within forty-eight hours.
Kwame sat at his desk, staring at the digital invoice for 50 User CALs. He had requested Device CALs. He had explained, in a three-paragraph memo that was likely skimmed for dollar signs and nothing else, that the warehouse operated on three shifts with 142 different workers sharing 38 ruggedized tablets.
The math of the engineer is precise; the math of the approver is comparative. Dana saw a number that sat comfortably within the quarterly buffer, she saw a vendor name she recognized, she saw a product description that contained the words “Remote Desktop Services,” and she closed the loop. She approved the price, but she bought the wrong license.
The Invisible Chemistry of Failure
The wrong license is a quiet poison in a network. It doesn’t trigger an immediate crash, it doesn’t smoke like a blown capacitor, and it doesn’t send out a frantic alert to the sysadmin’s phone in the middle of the night.
Instead, the wrong license waits. It sits in the licensing manager console, looking compliant and tidy, until the moment the audit begins or the hardware limit is reached, at which point the “reasonable” price paid months ago becomes the down payment on a much larger disaster.
In the world of vintage sign restoration, we call this the “mounting bolt fallacy.” A client will pay four thousand dollars for a restored 1950s porcelain enamel sign, but they will balk at the sixty-dollar cost of high-grade galvanized mounting bolts, preferring instead the cheap zinc ones from the local hardware bin.
The zinc bolts look the same. They have the same threads. They hold the sign up on Tuesday. But zinc reacts with the acidity in the rain, it creates a galvanic cell against the iron of the sign, and five years later, the four-thousand-dollar investment is lying shattered on the sidewalk because the buyer only scrutinized the legible cost of the art, not the invisible cost of the chemistry.
When an IT team looks at the RDS CAL Store, they are often navigating a sea of technical nuances that the procurement department is fundamentally unequipped to judge.
An approver can tell you if $2,840 is more or less than $3,100, but they cannot tell you why a User CAL is a liability in a hospital ward where nurses rotate through a single terminal every forty minutes. They see the “User” label and assume it refers to the people, which is a logical linguistic leap, but in the architecture of Windows Server 2022 or 2025, that leap is a cliff.
The procurement officer looks at the line item, he sees the vendor name, he notes the subtotal matches the quote, he clicks the “Approve” button with a sense of duty done. The wrong license remains.
Judging What We Can Measure
This happens because oversight naturally concentrates on what it can judge and abandons what it can’t. This is a phenomenon I’ve seen in every trade, from sign hanging to bridge building.
In the late , during the initial craze for neon advertising in Los Angeles, the city’s building inspectors were overwhelmed. They knew how to check if a sign was hanging too low over a sidewalk, they knew how to measure if it was too wide for the storefront, and they knew how to check the permit fees.
They did not, however, understand the high-voltage secondary wiring. They waved through signs that were essentially lightning bolts waiting to happen, simply because the paperwork was in order and the dimensions were legal. The dimension that mattered most-the electrical load-was the one the inspector couldn’t see.
Kwame’s frustration wasn’t about the money; it was about the misalignment of reality. By approving the User CALs, Dana had technically followed the rules of fiscal responsibility while simultaneously ensuring that the warehouse would be “under-licensed” the moment the second shift clocked in.
If you have 50 User CALs and 142 users, you are out of compliance, even if you only have 38 devices. If you have 38 Device CALs, you are covered for all 142 users. The difference in price between the two options is often negligible, sometimes only a few hundred dollars, which makes the “savings” of the wrong license even more bitter.
The wrong license is a perpetual debt. Because these are perpetual licenses, once they are injected into the licensing server, they become part of the infrastructure’s permanent record. You cannot simply “swap” the DNA of a User CAL for a Device CAL once the key has been activated and the grace period has expired. You are left with a correctly-priced, incorrectly-typed asset that serves as a monument to a hollow approval process.
We often talk about “sizing help” as a luxury, a bit of hand-holding for the unsure. In reality, sizing help is the bridge. When a specialized vendor provides a CAL calculator or a pre-sales consultation, they aren’t just selling software; they are providing a translation layer. They are telling the Danas of the world that the “reasonable” price is only reasonable if it actually covers the 142 people standing on the warehouse floor.
I remember a project in involving a massive rooftop sign for a local bottling plant. The owner wanted the cheapest possible solution for the structural steel. He looked at the engineering quote, he saw a line for “wind load bracing,” and he crossed it out, noting that the building had stood for forty years without a sign, so it didn’t need extra bracing now.
He approved the cheaper version. He felt like a hero of the bottom line. Three months later, a Santa Ana wind gusting at turned that sign into a sail, which turned the roof into a lid, which turned the bottling plant into an open-air ruins. He had scrutinized the dollar amount he understood and ignored the physics he didn’t.
The wrong license is the wind load bracing of the IT world.
From Price to Purpose
To break this cycle, the internal conversation has to change from “Is this the best price?” to “Is this the correct unit of measure?” The engineer must become a storyteller, and the approver must become a student of the invisible.
If Kwame had presented the quote not as “50 licenses” but as “coverage for 38 terminals regardless of who touches them,” the legible reality might have shifted. But as long as the process is reduced to a “Total looks reasonable” email, the technical debt will continue to accumulate.
We are living in an era where the delivery of these licenses happens in about . That speed is a marvel of modern distribution, but it also means that the window for catching a mistake is smaller than it has ever been.
In the old days, you ordered a box, the box sat on the loading dock, someone opened the box, realized it was the wrong thing, and sent it back. Now, the wrong license is delivered, decrypted, and deployed before the coffee in Dana’s mug has even gone cold.
The efficiency of the transaction masks the deficiency of the choice.
When you stand back from the workbench and look at a restored sign, you want to see the glow. You want to see the vibrant, humming life of the neon. But if you’re the one holding the warranty, you should be looking at the transformer.
You should be looking at the mounting bolts. You should be looking at the things that don’t glow, because those are the things that determine how long the light will last. The price of the CALs is the least interesting thing about them; the type of the CALs is the only thing that will matter when the auditors or the server limits come knocking at the door.
We buy what we can see, we approve what we can calculate, and we ignore the rest until the buzzing starts.
The budget serves the price, but the price cannot serve the device.