A study by scientist David Soergel to the failure of scientific software shows that there are approximately 15 to 50 errors per 1000 lines of code. Note that code written by professionals. In the (university) science is also once common for students to go during their studies to work themselves with their own programmed software, which is even more unreliable
.
scientific research make use of more analogue instruments must to take into account defects, but which are predictable and measurable. With the use of software, you know that there are differences, just do not know where these deviations occur and how big they are.
Soergel says that the software errors scientific results are actually worthless. “If you know that one or two bugs in the code output already 100 percent maim, every scientific conclusion on the basis of the output is in advance all worthless.”
He points out that any scientific research by many common causes has been running a high risk of incorrect output and certainly erroneous conclusions. If that is still invisible maimed by bugs in the software, science has a big problem.
Soergel is proposing to do more research on this issue. The well-known journalist Glyn Moody is advising him not to use any software.
No comments:
Post a Comment