I sort of laid out the logic for inclusion of human factors considerations in the question. Three observations that would support limiting them on SU: 1. They might be a magnet for attracting low quality answers or discussion even though they are not inherently so. 2. The knowledge base. Suppose someone asks a question like, "My computer was working fine last night when I backed up everything to a DVD. This morning, I can't get it to boot up." And someone posts an answer like, "You probably left the DVD in the machine. The DVD drive is the first boot device and your backup disk isn't bootable." That would probably be better as a comment, but nobody would blink at that as an answer. It is based on common knowledge. Nobody would expect the answerer to cite a reference source.<br><br> In the "teh" question, my answer is common knowledge to anyone with human factors training. It's pretty basic stuff--it's the reason you can blink when something approaches your eye faster than you can consciously make yourself blink. However, it was seen by some people as an opinion-based answer, or at least a low-quality answer because it lacked citations.<br><br>There is an important curation function performed on the site, policing answers to ensure quality. That is difficult to do when questions or answers are in a subject area for which the majority of the high-rep members are not familiar. That is a reason to limit topics to a defined range of subject areas for which the community has expertise. 3. Existing capabilities vs. new capabilities. Shopping questions are off-topic. They deal with adding or improving capabilities rather than solving problems with what is there. Almost by definition, solving problems with existing hardware or software is a "silicon-based life form", to use Mokubai's terminology. Questions that involve improving the user interface (as opposed to fixing it), include the human component, even if that is not explicitly raised. If the goal is to limit the site to silicon, consider defining the site as dealing with the existing hardware and software. Or, define user interface questions (which would include ergonomics), as off-topic.