A clip going around makes a chain of claims: a hard drive gets heavier when you fill it with data, all the world’s stored data currently weighs about a kilogram, data is growing about 25 percent a year, and so in 340 years the data on Earth will outweigh the Moon. From there it leaps to dark matter and human consciousness, and it credits the whole chain to a physicist who never said any of it.

The physics errors are worth cataloguing, and we’ll get to them. But the useful lesson has nothing to do with physics. Two of those numbers cannot both be true, and working out which one is wrong tells you exactly where the reasoning came from. That mismatch is a general-purpose tell, and once you can see it you’ll catch it everywhere.

The idea being garbled

Start with the name attached to the claim. Landauer’s principle, published in 1961, says that erasing one bit of information must dissipate at least kT·ln2 of energy as heat into the surroundings. It’s a real result, experimentally confirmed in 2012. It concerns the energy cost of erasure. It says nothing about mass, nothing about writing, and Landauer died in 1999.

The idea that a full drive outweighs an empty one belongs to Melvin Vopson, who proposed in 2019 that each stored bit carries a rest mass of kT·ln2/c², about 3.19 × 10⁻³⁸ kg at room temperature. That’s a conjecture. It’s untested, it’s disputed in the peer-reviewed literature, and the viral version presents it as an established finding by the wrong man.

The arithmetic that gives it away

Take Vopson’s own bit mass and apply it to the world’s stored data, call it 200 zettabytes. The total comes to 5.1 × 10⁻¹⁴ kg. About fifty picograms. Not a kilogram, and not close to one: the viral figure overshoots by thirteen orders of magnitude.

Now compound that real figure at 25 percent a year. It reaches the mass of the Moon in roughly 373 years. The clip says 340. Close enough to be the same calculation.

But run the same growth from the kilogram the clip actually gives you, and you reach the Moon in 236 years. Not 340. The two numbers offered in the same breath are incompatible. The 340-year figure was computed from a starting mass the speaker never had, and doesn’t appear to know.

The final leap abandons physics entirely. There’s no hidden mass inside a drive waiting to be found. A stored bit is a magnetic domain or a trapped charge, which is to say it’s electromagnetic through and through. Too small to weigh is not the same thing as invisible to light. And the jump to dark matter fails on size before it fails on logic: dark matter in the observable universe runs to about 2.4 × 10⁵² kg, some sixty-six orders of magnitude more than every bit humanity has ever stored, and it was bending starlight for billions of years before anyone built a hard drive.

Why the mismatch is the tell

  • Bad reasoning usually fails uniformly. Here a correct extrapolation sits on top of a premise that’s wrong by a factor of ten trillion. That combination doesn’t come from one confused person. It comes from a calculation and a premise that were picked up in different places, with nobody checking whether they still fit together.
  • A number you worked out yourself stays tethered to its inputs. A number you inherited floats free of them. Asking whether the conclusion still follows from the stated premise costs a minute and catches almost every secondhand retelling.
  • The one piece of the story that’s genuinely Landauer, that erasing data pushes energy out of a drive, argues against the story’s own thesis. Erasure costs energy. Writing doesn’t have to. Bennett’s work on reversible computing shows that storing and computing can in principle be free.
  • Publication located this claim; it didn’t certify it. Vopson’s paper is real and peer-reviewed, and it has drawn published rebuttals arguing the principle generates unacceptable paradoxes, along with sustained criticism of the work that followed.

The honest caveats

  • Landauer’s principle is correct and experimentally verified. Nothing above disputes it. The problem is what got attached to his name.
  • The 340-year figure is very nearly right given its premise. Whoever first ran it did the work. That’s precisely the point of this piece, not an aside.
  • Vopson is a working physicist publishing in a peer-reviewed journal. Probably wrong and unserious are different accusations, and only the first is on the table.
  • An exponential extrapolated across 340 years is meaningless in any case. Vopson’s own paper concedes that the power required to sustain the growth exceeds the planet’s total energy consumption long before the mass becomes interesting.

Bottom line: when someone hands you a premise and a conclusion together, check that the conclusion still follows from that premise. If the number only works with inputs they never gave you, they’re repeating someone. And if they had checked, they would have caught it, which means nobody in the chain did.