Nature 's roundup of the papers and issues gaining traction on social media.

Ancient hominin bones made good fodder for debate on social media of late, when researchers suggested a theory about the identity of the Indonesian 'Hobbit'. Scientists also took note of a fast, efficient computer chip that replaces binary processing with a much more brain-like approach.

Based on data from Altmetric is supported by Macmillan Science and Education, which owns Nature Publishing Group.

Two papers stirred up the palaeoanthropology world by suggesting that Homo floresiensisa putative human relative discovered on the Indonesian island of Flores in 2003 — was instead an example of Homo sapiens with Down's syndrome. The theory, published in the Proceedings of the National Academy of Sciences (PNAS), was greeted with much scepticism. As part of a string of tweets, anthropologist Holly Dunsworth at the University of Rhode Island in Kingston said: “Conclusion [is] based seemingly on zilch.” Co-author Robert Eckhardt, a geneticist and evolutionary morphologist at Pennsylvania State University in University Park, defended the diagnosis in a comment posted on a blog of the Natural History Museum in London, saying that his group and others have spent the past decade “trying to turn the 'Hobbit' circus into science”.

The specimen, known as LB1, is an odd character. Estimated to be 18,000 years old, she (or possibly he, which is another side to the controversy) was a full-grown adult standing just over 1 metre tall with a skull about one-third the size of a human's. She reportedly had relatively large feet and short legs, which, along with her short stature, earned her the nickname Hobbit. The skull also has a small chin and a receding brow.

The PNAS papers argue that the supposed anomalies of LB1 “are highly diagnostic of Down syndrome”. Reached for further comment, Eckhardt said that the team had considered many other syndromes to explain LB1's characteristics. After initially rejecting the idea of Down's syndrome several years ago because the skull seemed too small, the authors were “forced” back to the hypothesis by evidence such as the short femur, he said.

In a follow-up interview, Dunsworth said that the researchers took an odd and unconvincing path to make their case. “I would expect them to first show that it's possible to diagnose Down syndrome from a skeleton,” she said. “That's not what they did.” Dunsworth was also critical of the overall tone of the papers. “The picture they paint of palaeoanthropology is that it's nothing but conjecture,” she said.

Eckhardt admitted in his blog comment that the Down's diagnosis is not certain because there's no way to confirm the Hobbit's chromosomal makeup. “So far recovery of even reliable fragmentary DNA has been elusive,” he wrote.

Adam van Arsdale, an anthropologist at Wellesley College in Massachusetts, used Twitter to share his big-picture perspective on the matter. “Whatever Flores is, it is not the Rosetta Stone of human evolution. It is a strange side story, like an episode of Lost.”

Henneberg, M. Eckhardt, R. B., Chavanaves, S., & Hsü, K. J. Proc. Natl Acad. Sci. USA (2014); Eckhardt, R., Henneberg, M., Weller, A. S., & Hsü, K. J. Proc. Natl Acad. Sci. USA (2014)

A report of a computer chip that mimics the architecture — if not the thinking power — of the human brain grabbed the attention of both technophiles and neurophiles on social media. The TrueNorth chip, created by researchers at IBM and Cornell University, combines 5.4 billion transistors to create 1 million interlinked nodes that communicate with each other in much the same way as neurons. The researchers say that the brain-like approach and better chip-production technology make TrueNorth 1,000 times more efficient than conventional chips.

Whereas many observers were excited by the advance, some people still needed convincing. Yann LeCun, director of artificial-intelligence research at Facebook, says on Google+ that “this type of neural net ... has never been shown to yield accuracy anywhere close to state of the art on any task of interest.”

Writing on his blog NeuroLogica, neurologist Steven Novella at Yale University School of Medicine in New Haven, Connecticut, said that although this sort of chip can potentially speed up computing, it will be difficult for personal computers to make the shift from binary to brain-like chips because software would probably have to be rewritten from scratch.

Merolla, P. A. et al. Science 345, 668–673 (2014)