Tiny structural errors in proteins may have been responsible for changes that sparked complex life, researchers say.
A comparison of proteins across 36 modern species suggests that protein flaws called "dehydrons" may have made proteins less stable in water.
This would have made them more adhesive and more likely to end up working together, building up complex function.
The Nature study adds weight to the idea that natural selection is not the only means by which complexity rises.
Natural selection is a theory with no equal in terms of its power to explain how organisms and populations survive through the ages; random mutations that are helpful to an organism are maintained while harmful ones are bred out.
But the study provides evidence that the "adaptive" nature of the changes it wreaks may not be the only way that complexity grew.
Single-celled life gave rise to more complex organisms, and with them came ever-more complicated networks of gene and protein interactions.
Michael Lynch, an evolutionary theorist at Indiana University, teamed up with Ariel Fernandez of the University of Chicago, both in the US, to look specifically at protein structure.
They considered 106 proteins shared among 36 modern-day organisms of widely varying complexity, from single-celled protozoa up to humans.
The pair were studying "dehydrons" - regions of proteins that make them more unstable in watery environments.
These dehydrons - first discovered by Dr Fernandez - make the proteins more sticky in water, thereby raising the probability that they will adhere to other such proteins.
The analysis showed that organisms with smaller populations - such as humans - had accumulated more of these defects than simpler organisms with vastly higher population numbers.
The suggestion is that it is the acquisition of these defects, with sticky proteins more likely to work together in ever-more complex protein-protein interactions, that nudged cellular complexity upward.
"We've tried to bridge the gap between protein structure and evolution and believe we've uncovered evidence that proteins develop mild defects in organisms with smaller population sizes, over the great divide from bacteria to unicellular eukaryotes to invertebrates up to us vertebrates," said Professor Lynch.
These slight defects may decrease protein function even as they increase protein cooperation.
The authors suggest then that other adaptations occur that "undo" the deleterious effects of the sticky proteins.
For example, the protein haemoglobin that carries oxygen in our blood, is made of four identical subunits, each with a range of dehydron flaws; simpler organisms have globin molecules that accomplish the same job with just one subunit.
But the overlap of the four subunits actually masks the flaws in each one.
The authors stress that they are not arguing against natural selection as a process; they say rather that it can be aided by "non-adaptive" mechanisms.
"There's been this general feeling that complexity is a good thing and evolves for complexity's sake - that it's adaptive," Professor Lynch told BBC News.
"We've opened up the idea that the roots of complexity don't have to reside in purely adaptational arguments.
"It's opening up a new evolutionary pathway that didn't exist before."
Ford Doolittle of Dalhousie University agrees that this mechanism, separate from Darwin's vision of natural selection, is an important consideration.
"Darwinists are a little bit like the pre-Darwinists before them, who would have marveled at the perfection of God's creation," he told BBC News.
"We tend to marvel at the Darwinian perfection of organisms now, saying 'this must have been highly selected for, it's a tuned and sophisticated machine'.
"In fact, it's a mess - there's so much unnecessary complexity."
While he called the Nature study "important and interesting", he disagrees with the mechanism that allows organisms to recover from the protein flaws.
He has long argued for a "presuppression" mechanism, in which some organisms may have a way to overcome the limited functionality of the slightly damaged proteins, and those that do survive best.
"He's putting the cart before the horse," Professor Doolittle said of Professor Lynch's idea that subsequent mutations solve the problems raised by the protein changes.
"But we both agree that much of complexity does not have an adaptive explanation."