If there’s anything in the cosmos more mysterious than dark matter, it’s probably dark energy. It makes up more than two-thirds the energy in the universe and encourages the universe’s accelerating expansion. Yet no one has the slightest clue what it is. Some undiscovered particle? A cosmological constant? Or perhaps dark energy emerges as a sign that space and time aren’t what they seem.

Every graduate student who studies particle theory suspects at one time or another that space-time might be discrete. There's a surprising reason behind that: the sums at the heart of quantum field theory are hopelessly, woefully infinite, primarily because the theory misbehaves at very short distances — ironically, this is exactly the reason it's so hard to figure out a quantum theory of gravity. If we just made it so there was a smallest possible distance, a tiniest possible step you could take, that problem would evaporate. But theorists figured out another way to deal with most of the infinities — one that does a remarkable job of explaining how particles behave — and anyway, how do you define a tiniest step? What should it look like?

Well, it should look like a pyramid. Space and time really might not be what they seem.

That view, recently proposed in the journal

*PLOS One*, stems*from a controversial idea called dynamical triangulation. First proposed as a theory of quantum gravity, DT holds that minute tetrahedrons—triangle-bottomed pyramids—are the fundamental units of space and time. Viewed from a distance, these units disappear into the smooth, continuous world of everyday life. Up close, churning stacks of tetrahedrons bend left, right, front, and back—and that’s where dark energy might lie. Calculations suggest that energy stored in the bends is distributed almost uniformly, just as observations suggest dark energy should be. And while the model’s predictions for dark energy’s density are rough at best, they manage to come within a factor of ten of the measured value, something quantum field theory ignominiously fails to do—it misses the mark by a factor of 10^107, or 1 followed by 107 zeros.*