One interesting result is that PVA glued joints seem to actually be stronger if there’s a small gap. This would suggest carpenters should be less obsessive about making joints perfect, and in fact should intentionally introduce some imperfection (in hidden faces, of course), to produce a stronger joint.
Why do you suppose glue filling a gap would be stronger than a “proper” fit?
Here is Fine Woodworking’s results on the strength of glue. I think what we need to take from these tests is that glue is pretty darn strong, so we shouldn’t beat ourselves up if a joint is a little looser that what we were expecting it to be.
In the Mathias video he is testing the glue not the joint, that is why his gapped joint was stronger. His failure mode was the same every time meaning he was only stressing the glue and relying on the wood strength not at all. The gapped joint acted as a plastic hinge which allowed him to put more force in as the joint went plastic past the yield point of the glue. All His other tests failed at the approximate yield strength of the glue he was using. His test is flawed and should be taken with a grain of salt. The Fine woodworking tests are much better and more accurate.