UPDATE: IBM "very much involved" with Blue Brain Project

On January 8, I wrote that IBM had pulled out of the collaboration with the Blue Brain project.

It's not true; but actually to appropriate Peter Woit's complaint about string theory, it's not even wrong. Let's unpack.

The reporting on the Blue Brain project so far has been heavy on hand-waving and light on technology and business details. That might have something to do with the fact that not even the people at IBM are clear on their relationship with the Blue Brain project. Even at IBM's T.J. Watson research campus in New York, where much of the Blue Brain effort was happening between 2005 and 2007, the wires got crossed, leading to the misapprehension that "IBM Research completed the first phase of the Blue Brain project and we are not involved with the second phase." A senior staff person for Blue Gene computing confirmed what the PR person had told me, saying that IBM and Ecole Polytechnique Federale de Lausanne, Blue Brain's home, had no formal ties and that Blue Brain project director Henry Markram could put his Blue Brain on the next generation of IBM's Blue Gene, or that he could put it on some other company's supercomputer. "Our architecture would be the path of least resistance," he shrugged.

IBM Zurich leaped to the rescue with a full correction on the status of the IBM/Blue Brain collaboration.

David Cremese, the manager of Deep Computing Programs at IBM Zurich, told me that the first phase of Markram's project is complete but that IBM intends very much to collaborate on future phases. It is very likely, he says, that the next generation of Blue Brain will be housed on the next generation of Blue Gene. And Markram confirms that the Blue Brain project is not going to continue without the Blue Gene/P. This is not quite like upgrading your laptop. A system this complex is riddled with programming specific to the architecture of the machine, so it would be very time consuming and expensive to switch horses mid-stream.

For Markram, though, this presents an opportunity to clear up some muddy reporting about his project. "There is a serious misconception that IBM somehow funded or donated to support the Blue Brain Project," he says. "The press through the last 3 years has always got this wrong." In many reports, the IBM tie was overemphasized. Blue Brain is funded primarily by the Swiss government and secondarily by Markram's grants and some donations from private individuals. As far as I can tell, the misapprehension that IBM funded Blue Brain started when IBM sent EPFL a Blue Gene/L below market price. Markram says that's because the Blue Gene/L was a prototype at the time and EPFL was a beta site.

IBM also paid for two postdoctoral fellows to the Blue Brain effort: one postdoc was sent to Switzerland from T.J. Watson to work at EPFL, and a second postdoc remained at Watson to work on computational neuroscience algorithms for Blue Brain. However, it turns out Watson had no intention to support (via post-doctoral fellows, discounted supercomputers, or any other kind of assistance) the project past 2007.

The same is not true of IBM Zurich. "IBM is very much involved and is a trusted technology partner of the project," IBM Zurich media analyst Susan Orozco says. "The team that developed the Blue Gene technology [at Watson] will continue to collaborate with the Blue Brain project," adds Cremese.

The takeaway: Blue Brain is just fine. IBM, well, they just need to get their stories straight.


Tech Talk

IEEE Spectrum’s general technology blog, featuring news, analysis, and opinions about engineering, consumer electronics, and technology and society, from the editorial staff and freelance contributors.

Newsletter Sign Up

Sign up for the Tech Alert newsletter and receive ground-breaking technology and science news from IEEE Spectrum every Thursday.