...
Merali, "Computational science: ... Error" in Nature. "As a general rule, researchers do not test or document their programs rigorously, and they rarely release their codes, making it almost impossible to reproduce and verity published results generated by scientific software, say computer scientists."
...