Researchers have taken a second look at the NSA SKYNET leaks, as well as the GCHQ data-mining problem book first published on Boing Boing, and concluded that the spy agencies have made elementary errors in their machine-learning techniques, which are used to identify candidates for remote assassination by drone.
Read the rest
UPDATE: After you read this story, make sure you check out the follow up piece. Editors at Embargo Watch have found evidence that The Sustainable Food Trust manipulated the media to prevent public criticism of this paper.
Yesterday, in an aside to a post criticizing an astroturf political campaign in California, Mark mentioned a new study that supposedly found GM corn causes tumors in rats. As Mark said in an update to that post, this study is severely flawed, but I wanted to follow-up on that with some discussion about why it's flawed.
After all, the study was peer-reviewed, right? Doesn't that mean we can trust it?
Here's the thing. Peer review is not perfect. It's not a panacea. It's simply the basic level of due diligence. By submitting work for peer review, a scientist has allowed people outside her own team to critique her work. And the journal might require some changes to the paper based on the critique — anything from edits for clarity to requesting that the scientist perform another experiment in a different way. If a paper hasn't gone through peer review, you should be more skeptical of it. Avoiding peer review means that the researcher decided to show the public her results before allowing those results to be critiqued by independent experts.
But, at the same time, just because something has gone through peer review doesn't mean it's been certified to be accurate. It just means that roughly three other experts have looked at the paper before publication. Read the rest
The two papers documenting evidence that CERN has found a particle matching the description of the Higgs Boson have cleared peer review and are now published in the journal Physics Letters B
. Ironically, that journal is the offspring of Physics Letters
, the journalwhich rejected Peter Higgs' 1964 paper
that first hypothesized the existence of the Higgs Boson. Higgs' paper was eventually published by a different journal, Physical Review Letters
. You can read it online
. (Many thanks to Rachel Courtland for the history, and to Jennifer Ouellette for the Higgs paper link.) Read the rest
Remember how scientists discovered alien-esque life forms in California and the Internet was all, "Oh, sheeeet!" But then other scientists started critiquing the research and there was a giant debate about whether one scientist could call out another scientist for bad data on a blog, rather than in a peer-reviewed journal, except that the peer reviewed critiques basically said the same thing and the "discovery" turned out to be totally incorrect? I'm making light of arsenic life here just a bit, but this story of de-discovery continues to be interesting and important. Today, on NPR's Science Friday, science journalist extraordinaire Carl Zimmer will explain why, and will talk about what happens when scientists are wrong
. Read the rest
Anything that inspires a good angry rant in real life can be turned into a Downfall video.
Getting a peer reviewed research paper through the aforementioned review process can be a stressful, rant-inducing experience. Remember, in order to be published, the paper is read by three (usually anonymous) reviewers who work in the same field of science. They judge things like whether the experiments described in the paper were done well enough, whether the work is original, and whether the take-away conclusions the scientist is presenting match up with the results of the experiments.
Last year, I wrote up a longer piece explaining peer review in more depth. Give it a read, and then see if you're surprised that there are multiple versions of peer review Hitler.
Above, Hitler is having problems with the third reviewer on his peer review board. Below the cut, Hitler's grant proposal is rejected by the National Institutes of Health. Read the rest
Peer-review does many things, but it isn't built to weed out fraud. In the wake of large scandals like the expose of Andrew Wakefield's fraudulent autism study, the British government is starting to consider regulating science for fraud the same way it regulates restaurants for public health. Brian Deer, the journalist who helped expose Wakefield, supports the idea. What do you think? (Via Ivan Oransky) Read the rest