Here’s a strange thing about modern cosmology: a growing number of serious physicists think our universe might be one of many, and that most of the others are duds. Empty bubbles, universes that collapsed in a blink, universes whose physics couldn’t hold a single atom together.
If that’s true, the interesting question isn’t whether failed universes exist. It’s whether we could ever find evidence of one.
Some researchers are trying. They don’t use telescopes pointed at another dimension, because nothing like that exists. They use something subtler: careful statistical searches of the oldest light in the sky, looking for faint scars that a neighboring bubble might have left behind. So far nobody has found anything conclusive. But how the search works, and why it’s taken seriously, says a lot about where physics is heading.
A quick note on terms before we go further. “Failed universe” isn’t an official label in any textbook. It’s shorthand, used loosely in papers and popular writing, for a universe that began but never became a place where complex structure could form. I’ll use it that way here.
So what counts as a “failed” universe?
There’s no single definition, which is part of the fun and part of the problem.
In some theoretical models, a universe fails because it lacks the right ingredients. One cyclic-universe paper on arXiv, Cyclic Universe and Infinite Past, uses the term quite literally. In that model, new universes spawn at each turnaround, and a tiny fraction fail to cycle because a stray particle in the patch triggers a premature bounce. The author then works out whether failed universes would outnumber successful ones, because if they did, the model would make our own existence wildly improbable.
In other frameworks, failure means something more physical. A universe might expand so fast that matter never clumps into stars. It might have a gravity so strong that it collapses before anything interesting happens. Or its fundamental forces might be tuned so that atoms can’t exist. In all these cases, the universe “works” in a mathematical sense but never produces anything like the cosmos we live in.
So when a physicist says failed universes, they’re usually talking about the losers in a cosmic lottery.
Where the idea comes from
The notion of a crowd of universes didn’t come from science fiction. It fell out of two serious lines of theory.
The first is eternal inflation. The standard picture of the early universe says space went through an ultra-fast burst of expansion just after the Big Bang. In many versions of this theory, inflation never fully stops everywhere. Instead, it ends in patches, and each patch becomes a “bubble” that settles into a normal expanding universe, while the space between bubbles keeps inflating forever. Our observable universe would be the inside of one such bubble. That’s how the setup was described in a 2010 study searching for bubble collisions, whose authors note that our universe would sit inside a single bubble within a much larger inflating multiverse.
The second is string theory. It doesn’t pick out one unique universe. Instead it allows a huge number of possible configurations, each with different physical constants. Writing in Harper’s, the physicist and novelist Alan Lightman described this “landscape” and noted estimates that it contains something like 10^500 different possible universes. That number is so large it’s hard to say what it means. It’s essentially “more than we could ever count.”
Put those ideas together and you get a sketch of the cosmos where bubbles form with different laws and different fates. Most might be sterile. A few, like ours, hit a combination that allows stars, planets, and people to exist. We shouldn’t be surprised to find ourselves in one of the lucky ones, because only lucky universes contain anyone to notice.
How do you hunt for something you can’t visit?
This is the part that sounds impossible, and for direct observation it is. You can’t fly to another bubble. Nothing we can send or receive crosses the gap.
But eternal inflation makes one intriguing prediction. Bubbles don’t just sit politely apart. In a space that’s inflating forever, they nucleate at random, and some of them could run into each other. A collision between our bubble and a neighbor might leave a mark on our side.
That mark, if it exists, would show up in the cosmic microwave background, the faint afterglow of the early universe that fills the sky at microwave wavelengths. It’s the most detailed baby picture of the cosmos we have. A collision would imprint something like a disk-shaped patch: a region with a slightly different temperature, with a specific profile around its edge.
The point is that this is a testable prediction. It’s not “we might see a portal to another universe.” It’s “we might see a statistical anomaly of a particular shape in a map we already have.”
What the first real search found
The first serious attempt used data from NASA’s WMAP satellite. A team led by Stephen Feeney at University College London, with collaborators from the Perimeter Institute and Imperial College, built an algorithm to hunt for bubble-collision signatures. Their companion paper on the analysis methods describes how they tested the method on simulated skies, with and without collisions, before running it on real data.
What did they find? In short: nothing you’d call a detection, and a little bit of tantalizing noise.
The team ruled out a range of collision scenarios, since those would have shown up and didn’t. They found four features in the WMAP data that were consistent with being bubble collisions. But after weighing the statistics, they concluded the data did not justify adding bubble collisions to the standard cosmological model. Their analysis put the average number of detectable collisions across the whole sky at under 1.6, at the 68 percent confidence level.
Cosmologist Matthew Johnson, one of the authors, wrote a guest post on Sean Carroll’s blog explaining the work in plain language. He was upfront that there were no clear detections, but that those four features were better explained by the collision hypothesis than by ordinary random fluctuations. That’s a careful way of saying: interesting, not convincing.
The authors also pointed out that data from the Planck satellite could settle the question more decisively. That’s how this kind of science works. A weak hint is a reason to look harder, not a reason to celebrate.
Another trail: universes before ours
Not everyone looking for traces of other universes is thinking about bubbles. Another group asks whether our universe is one in a sequence.
The best-known example is conformal cyclic cosmology, proposed by the Oxford mathematician and Nobel laureate Roger Penrose. In his picture, the far future of one universe becomes the Big Bang of the next, in a series of “aeons.” If that’s right, some evidence from the previous aeon might survive into ours. Penrose and collaborators have argued that certain circular patterns in the microwave background, which they call Hawking points, could be remnants of evaporated black holes from an earlier universe. Forbes covered the claim and noted that it’s not definitive proof, only the kind of signature that would support the idea if it held up.
And it hasn’t gained wide acceptance. Many cosmologists question whether the patterns are statistically significant or whether they would appear in random skies too. This is a recurring theme in the whole field: when you search a noisy map for rare shapes, you’ll often find some. The hard part is proving they mean something.
Still, it’s a nice example of the same strategy. Whether the thing you’re looking for is a neighboring bubble or a predecessor universe, you look for fingerprints in the oldest light we can see.
Our own universe as evidence
There’s a quieter way to hunt for failed universes: study the one we’re in, and ask whether it looks like a lucky draw.
Physicists have long noticed that several basic numbers in nature seem finely balanced. If the strength of gravity, the mass of the electron, or the energy of empty space were noticeably different, stars might not form, or chemistry might not work. That observation motivated an influential argument in the 1980s: if a multiverse exists with varying constants, then the value of dark energy we observe should be roughly as large as it can be while still allowing galaxies to form. The measured value turned out to be in that general neighborhood, which many people took as a point in the idea’s favor.
Others aren’t impressed. A prediction that’s “roughly in the right range” is easy to claim and hard to test sharply.
Even basic questions about the shape of our own universe remain open. A 2019 analysis by Eleonora Di Valentino, Alessandro Melchiorri, and Joseph Silk argued that Planck’s data prefer a slightly closed universe and warned that assuming flatness could be hiding tensions in cosmology. That finding is debated, and I wouldn’t lean on it, but it shows that measuring our own universe’s geometry carefully could eventually inform which kinds of origin stories are plausible.
Why skeptics push back
If you’re thinking this all sounds a bit too convenient, you’re in good company.
The strongest objection is about testability. A theory that predicts an unobservable multiverse can be hard to falsify, because almost any outcome can be fit into some bubble somewhere. Critics argue that if a hypothesis can explain everything, it risks explaining nothing.
Defenders reply that eternal inflation isn’t unfalsifiable in principle. The bubble-collision search is exactly an attempt to test it, and it could have turned up a clear signal. The trouble is that a null result doesn’t kill the idea. Collisions might simply be rare, or too faint to see, or might have happened beyond our horizon.
Lightman, in that Harper’s essay, is realistic about this. He suggests the best we can hope for is that the theories predicting a multiverse also make other predictions we can test within our own universe, while the other universes themselves likely stay in the realm of conjecture. That strikes me as the right level of humility. You can build confidence in a framework through its testable consequences, without ever seeing the thing it describes.
There’s also a more philosophical worry. Some scientists dislike leaning on “we only see this universe because it’s the one that allows observers” as an explanation. It can feel like giving up on finding a deeper reason. Others think it’s simply the correct reasoning if the multiverse is real.
Why it’s worth looking anyway
Even if the search comes up empty forever, it isn’t wasted.
First, the tools are good for other things. The statistical methods built to scan the microwave background for collision signatures also sharpen how we test the standard model of cosmology, and they help rule out or constrain exotic ideas.
Second, null results are information. Each time a search fails to find a bubble collision, the allowed range of theories shrinks a little. That’s real progress, even if it doesn’t make headlines.
Third, the question forces cosmologists to be precise about what “our universe” even means, and what would count as evidence of something beyond it. Those are hard conceptual questions, and answering them carefully is a kind of science in itself.
And finally, there’s the plain curiosity. People have always wanted to know whether we’re alone, and “alone” can be asked on larger scales than other planets. Whether ours is the only universe, or one of a vast and mostly lifeless crowd, is among the biggest questions anyone can ask. It seems worth poking at even if the poking is slow.
Frequently asked questions
Have scientists actually found a failed universe?
No. There’s no confirmed evidence of any other universe, failed or otherwise. The searches so far have turned up ambiguous features at most.
Can we ever observe another universe directly?
Almost certainly not. The realistic hope is indirect evidence, such as a collision signature in the microwave background or other predictions from the theories that imply a multiverse.
What’s the cosmic microwave background?
It’s the oldest light in the universe, released roughly 380,000 years after the Big Bang. Tiny temperature variations across the sky encode information about the early universe, which is why it’s the main hunting ground for these searches.
Is the multiverse science or philosophy?
Both, depending on who you ask. It becomes science to the extent it makes testable predictions, like the bubble-collision signatures described above.
The bottom line
The hunt for failed universes isn’t a hunt for a place. It’s a hunt for statistical hints in data we already have, guided by theories that say our cosmos may be one small success among a sea of misfires. So far, the sky hasn’t given up any clear answers. That’s okay. A good search is one that could have found something, and these could have.
If the evidence ever does turn up, it’ll probably look unglamorous at first: a faint, odd patch in a map, argued over by people with spreadsheets. That’s how most big discoveries begin.
