It's easy to agree with "yes, significant claims should be backed up", but it's a vain statement as long as there's no common agreement of what "significant" means.
The problem is: what is a significant claim? What's significant to you may not be significant to someone else, and vice versa. Is it reasonably possible to back up everything that someone out there may consider significant? If not, then who decides what's significant and what's not?
Are the following claims significant or not? Or is some of them significant, but so obvious that it doesn't need to be backed up because of that?
- Developers are notoriously anti-management.
- Managers love processes.
- Projects tend to go overtime.
- Startups can beat big, established corporations.
- Big, established corporations can beat startups.
- Stackexchange sites are popular.
- Beginners tend to make mistakes.
- Big corporations tend to have heavy processes.
- Lighter processes have become more and more popular during the last 10 years.
- Formal processes become a requirement when company grows to > 10 people.
- Climate change is a major issue.
- Java is one of the most used languages out there.
- Smoking kills.
- I feel that a couple of cups coffee does good to my programming efficiency.
- Too much coffee is bad.
- Lines Of Code Per Day (LOCPD) is a lousy measure of developer productivity.
Do all of these need to be backed up? If not, which ones, and why? If yes, how would you back up e.g. the first one ("Developers are notoriously anti-management.")?
Personally I think that since the definition of significance and obviousness depends totally on who you ask, these sites should be kept quite lax - much more lax than some extremists suggest. Backing one's claims should of course be encouraged, but it shouldn't be required. Attempting to suppress noise may improve signal-noise ratio, but it will also more or less distort the signal itself. The fallacy is that one starts to consider his own (or his community's) judgement as the only valid measure of significance or obviousness, and dismisses anything that deviates from it. How is that better than opinions without evidence? Questions with tens of upvotes getting closed is a sad example of this. Lots of people would be interested in the question (that's why they upvoted it), lots of interesting answers have already been written, and then... some police closes the question as "Not a real question" or something. It's really sad.