Sign in or Join FriendFeed
FriendFeed is the easiest way to share online. Learn more »
Cameron Neylon
Query come to me this morning: do you have any dramatic examples of when an error in coding has lead to an important scientific error that was either caught just in time, or only caught after the results had been published?
There was quite a big story in crystallography in 2007 - 5 papers retracted due to small software error (covered by half of science bloggers here, for example Bosco http://boscoh.com/protein... or Hari http://harijay.wordpress.com/2007... ). People suspected for years that something is wrong with these structures, although it took quite some time to find the error. - Pawel Szczesny
Neil was faster. Again. :) - Pawel Szczesny
Here's a retraction "As a result of a bug in the Perl script used to compare estimated trees with true trees, the clade confidence measures were sometimes associated with the incorrect clades." http://www.ploscompbiol.org/article... - Dave Lunt
I think the Mars Climate Orbiter is winning the thread here but loving the other examples. Keep em coming! - Cameron Neylon
There's also cool site about famous software errors: http://www5.in.tum.de/~huckle... - Pawel Szczesny
The Talk page at the Wikipedia article on the MCO is good as well: http://en.wikipedia.org/wiki... - Cameron Neylon
Software bug reported for artificial heart:http://www.ncbi.nlm.nih.gov/pubmed... and defibrillator http://www.ncbi.nlm.nih.gov/pubmed... - Pawel Szczesny
Study designed to determine whether natural disasters affect suicide rates retracted due to software error http://www.nejm.org/doi... - Pawel Szczesny
@dgmacarthur mentions the Therac-25 case http://en.m.wikipedia.org/wiki... on twitter: http://twitter.com/dgmacar... - Cameron Neylon
Does this count? An arithmetic overflow caused by a bad cast caused an Ariane flight to explode, costing $370m -- http://en.wikipedia.org/wiki... - Andrew Clegg
in math circles, this is a favourite: http://www.emery.com/1e... - Andrew Lang
Not so much a software error as a procedural error: Edward Lorenz printed out his intermediate results only to three decimal places whereas his machine internally stored them to six decimal places. Upon restarting from a given check point the results quickly diverged from the original. Rather than simply being an annoying error this incident gave insight into the problem of sensitivity of initial conditions within the new emerging field of Chaos. (Best online account I could find of this is: http://www.aip.org/history... but there's also one in James Gleick's book "Chaos") - Dan Hagon
Just ran across a list of errors, some likely due to software. I love this concluding sentence "It remains to be seen whether such mistakes are isolated problems or nuggets broken loose from a veritable mother lode of error." see "Two infamous examples of discovered errors are cited by Dewald and Thursby: a pension study by Martin Feldstein [6],...." in Kane EJ. Why journal editors would encourage the replication of applied econometric research. Quarterly Journal of Business and Economics. 2003 ;23(1): - Heather Piwowar
Doing a bit of thread necromancy here, but has everyone seen this just in from Nature News: http://www.nature.com/news... - Andrew Clegg
Haven't read yet but its in my inbox. - Cameron Neylon from twhirl
Hi Cameron, haven't had time to read the past comments. This is an example (http://egybioworkshops.wordpress.com/2007...) that may have already been suggested above - Ramy Karam Aziz
From the Nature News piece Andrew mentioned above: "There needs to be a real shift in mindset away from worrying about how to get published in Nature and towards thinking about how to reward work that will be useful to the wider community." - Daniel Mietchen
RT Would it be fair to say that the greatest service Nature could do the scientific community might be to simply shut down? - Marius Kempe - Daniel Mietchen
I'm now imagining an April Fools joke from Nature, announcing that in the service of science, they are closing down, and all submissions should henceforth be put on the arXiv, and submitted to PLoS One :-) - Michael Nielsen