The speed of light in a vacuum is one of those immutable rules of existence. With his theory of relativity, Albert Einstein waved away the ether and waved into being an unchanging constant. But at one point, I was not so sure. A little over four years ago, I learned of two incidents that gave me pause.

Back then, for navigating unfamiliar roads and places, I still preferred to use my old GPS unit. And every so often, that meant I had to install the many updates required by the unit’s software. While trying to determine where to download the new bits of roadmap and firmware—for through abandonment I was several generations behind —I stumbled across several GPS forums. One thread in particular appeared in my online searches, noting several “navigation bugs” that persisted from one generation of the hardware and software to the next.

As I delved deeper, I learned that there were certain situations, mainly related to overpasses and feeder lanes on highways, where the unit I owned would get confused and jump the user’s position from one street to another. The unit would periodically assume that the driver was on the feeder lane rather than the highway, or driving on the incorrect side of the road, or sometimes even on someone’s lawn, and provide the appropriate—albeit incorrect—directions or warnings. Several intrepid users had begun trying to measure the amount of positional error, and while it was small, it seemed to be real. They had contacted the manufacturer numerous times, and while a continuing attempt had been made to correct for the problem, it kept on recurring.

Around the same time, I happened to read several articles in the popular press about high-speed trading, with stock purchases transacted in fractions of eyeblinks. Conducted by sophisticated algorithms, we now had trades precisely timed together, all at the speed of light. Yet recently, the algorithms had been blundering, their timing just inaccurate enough to send the price of hard red winter wheat on a high-speed oscillation, and the S&P 500 into a stall. Punditry blamed a similarity in trading strategies, with each proprietary method in lockstep, allowing any single choice to become magnified a million-fold and causing feedback and collapse. We had seen this several years ago, and it was only getting worse.

But I was armed with some additional information: my GPS unit didn’t work. The Global Positioning System requires knowledge of relativity in order to be accurate. Ignoring the adage that hoofbeats means horses, not zebras, I went full-on fringe. Perhaps, I wildly hypothesized, the speed of light itself was changing? It would account for both the navigation errors and the transactional ones. In the case of the former, a change in the speed of light would render the GPS unit’s calculations incorrect, and yield a wrong position for the driver. And in the case of the latter, the trades wouldn’t be synchronizing properly due to the speed of light’s change, causing all sorts of inadvertent financial cascades. In a short speculative article, I made just such a suggestion.

While tongue-in-cheek—and mainly an excuse to discuss some seemingly unrelated topics—I inoculated myself against crankery by providing some inexact precedents. There was the Pioneer anomaly—the observation that the Pioneer probes were slowing down more than expected as they journeyed beyond the gas giants of our solar system—which some speculated might have been due to unknown or changing properties of the laws of physics. Other scientists have even considered whether something known as the fine-structure constant—a combination of a slew of primal physical constants—might vary, both over time and even from place to place in our vast universe. In other words, that which we imagine to be constant could simply be due to our provincial perspective of the cosmos.

So what would it mean if the speed of light were itself truly changing? It would affect everything from our fundamental theories of physics to how we define our units of measurement. Photons hurtling through the heavens are actually the basic means for defining distance. The definition of the meter, once based on an underground metal bar, is intimately connected with the speed of light, and is in fact now linked by its definition: 299,792,458 meters per second describes both light and length.

When I wrote this little article about the speed of light changing, I didn’t expect any sort of corroboration; market collapses and a manufacturing defect were just the means for having a discussion of the anomalous and unexplained in physics.

However, I began to receive letters from readers, noting that their GPS units—made by different companies—also suffered from the same quirks as my own. They malfunctioned in the same way, and their respective corporations had continued to patch them, only to continue to be vexed by the same problems. Nevertheless, from the perspective of these GPS manufacturers, the matter was trivial: as long as the problems were corrected periodically, the issue never compounded. Yes, they noted, this was duct tape for technology, but until they had further insight, it was a resolved matter. And while certainly an unsatisfied curiosity, I relegated it to a similar category in my own mind and moved on.

But two months later I received this email:

Your fun little article about the changes in the speed of light reminded me of something that I noticed in one of the textbooks I use in my math classes. I teach third grade and in one of the books that I use to teach arithmetic, I noticed a mistake in the online answer key:

164,059,930 + 135,732,528 = 299,792,457.8

I see errata all the time, but this one made no sense! These kids are not supposed to know about decimals for another year or two, so where did that decimal point come from?

I didn’t think about this again until I read your article and realized that the correct answer to the problem should have been the speed of light in meters. Not sure what to make of this, but thought you might get a kick out of it.

My interest in these anomalies was renewed. Rather than Nature itself rewriting its book and pulling the rug of invariable speed out from beneath us, maybe it was Humanity? Perhaps some anthropic cause was the source of these errors: the actual speed of light wasn’t changing, but it only seemed to be.

Errors can propagate for a long time. In fact, even errors about our physical world can last for years. From the mythical mountains of Kong—etched upon the maps of western Africa for years despite their nonexistence—to the erroneous island of California, our world is a mix of the real and the rapidly dwindling fabulous, as we root out the mistakes of history.

Just a few years ago, an island in the Pacific that had been passed along from one cartographic generation to the next, even unto Google Maps, was determined to be a fantasy. Known as Sandy Island, it was the size of Manhattan and completely nonexistent.

Was this then a deliberate inaccuracy? Were there computers somewhere out in the nebulous cyberspace rewriting any instance of 299,792,458, slowly altering our perception of reality?

Was this due to a computer virus, a computational creature burrowing its way into machines? Or even a massive army of industrious fabricators, working around the clock to manually modify texts related to the speed of light?

I began to explore the extent of these changes. I found real, though subtle, evidence on physics websites, digitally archived articles in the journals of Elsevier and Springer, online copies of textbooks at Google Books, even the manuals available for download at the National Institute of Standards and Technology, the very holders of metric invariance. While print material remained constant, the inflexible speed of light had drifted in every document I searched for in the online world. But how? And why?

Some professors periodically assign their students sophisticated Wikipedia vandalism to teach them the nature of the truth; was this a postmodern exercise in the nature of physical relativism? Was the one truly unchanging constant of the universe now subject to wiki-style revision?

Or perhaps this wasn’t a commentary on permanence and facts; perhaps it was a delicate argument against digitization? If Amazon can modify, or even repossess, the electronic books you own, maybe that which is fixed can only be found in printed matter? And by vandalizing the physical constants of the universe we would finally recognize that. That printed knowledge is neither gossamer nor a fact on temporary loan. It alone is permanent.

Combing through the source code in the basement of the Internet to solve this mystery seemed to be an overwhelming task. But I knew that there was at least one way of narrowing down the possibilities. I contacted my GPS manufacturer, describing the textbook email and my suppositions from it. I asked, Would they be able to examine their servers? Might there be some sort of virus, or forced informational entry, all in order to rewrite our perception of physics?

Several days later, they informed that they had found something.

It was a worm. A tiny bit of code that embedded itself deep within the background of a computer—invisible beneath the algorithmic hum of the everyday behavior of the machine—and designed to do two things: replicate and change physics.

Like all living things, with their need for reproducing and spreading instructions for its further spread, the worm was designed to make copies of itself (though not too many) and periodically mutate, in order to evade detection.

But it was also designed for scientific and engineering mischief. Any time the speed of light was detected in a file on the host machine—in meters per second, miles per second, kilometers per hour, and many other units of measurement—it would modify its value by a tiny amount. And as time elapsed, lightspeed slowly decayed, bit by bit.

The GPS team was astonished at the craftsmanship of the tiny program. But they were also scared beyond imagining when it came to what the worm could do to the world of global positioning.

Who uses GPS? Those who are lost, new to town, and the military. For the most part—soldiers aside—if you’re not sure where you’re going, incorrect directions are a nuisance, but really, you’re already confused, so what’s a little bit more befuddlement?

But if you belong to the armed forces and your missiles are consistently missing their mark, that matters. My visions of the English comparative literature coder began to dissipate. Bringing down directions, not to mention the financial system, began to seem far more sinister than simply teaching us a lesson on the permanence of knowledge.

This worm was carefully crafted, and clearly the work of professionals. But who was responsible for it? We pored over the code, looking for tiny clues. Perhaps something in the comments of the program might point us to its source. Even the language of the comments would be enough for a lead. Unfortunately, we never found anything.

The virus is still a mystery. Perhaps it was something designed by a government, along the lines of Stuxnet, which destroyed the centrifuges of Iran. Here though, nothing physical was destroyed. Instead it was a more existential form of assault. Why attack the infrastructure of a country when you can disrupt the informational bedrock of a society? A nefarious infection that leaves misinformation in its wake, misinformation of a fundamental sort. There was something poetic in how it worked, though the poetry was reduced to numbers slowly unspooled from reality. It is said that Persian fables open with the formulaic “It went like this. But wasn’t.” This worm had brought this turn of phrase from storytelling to engineering reality.

Perhaps this virus is the harbinger of things to come, or of things already here. We are now living on constantly shifting ground, bystanders in a battle for truth and reality. Maybe I’m being sensational and this is a minor problem. And I hope that it is. But in the meantime, there is a simple solution: look to primary sources, the physical texts that Luddites praise for their feel and their smell, but are actually important not for these reasons, but because they are unchangeable.

My bookshelves are my bulwark against error. I have become a collector of antiques: I focus on books published prior to 1969. And the only relevant calendrical designation: Before ARPANET.