On May 27, 2009, at 14:29 PM, average wrote:
BTW, I know the topic's pretty dead by now, but there is a pretty conclusive argument on this topic (which I bring back up only because I find myself continually annoyed at most IDE's).
If the medium *emits* light, it's significantly better to have a dark background (the reverse being true if the medium is reflective--like the surface of a book or Kindle). Several arguments *support* this and none *contradict* it:
Ah yes, there are arguments and there are arguments. The ones you posted sound very persuasive. But in several controlled studies, the test subjects detected errors at a significantly better rate when the background was light than when it was dark. They did not report greater fatigue than the subjects who had a dark background. I try to rely on arguments only when I can't find controlled studies to rely on. Oh, thanks to denis.spir I now see that the explanation is probably due to overall luminance rather than to polarity per se: http://www.math-nat-fak.uni-duesseldorf.de/WE/Psychologie/abteilungen/ aap/Dokumente/Buchner-et-al.-in-press-Ergonomics.pdf Although this probably doesn't change the fact that you'll find errors better if you set your background to be light and your text to be dark. Regards, Zooko P.S. The tenuous link to the topic of this list is that if there *were* any controlled studies about Python programmers using different line lengths, then they could give us something to go on. Maybe somebody out there knows how to get research funding for experimenting on programmers? That would be wonderful.