6:17 p.m. on June 24, 2003 (EDT)
I didn't see your post, tuclimber, because you posted it as a response to your original. And sorry if my comments came off as a tirade.
Still, it amazes mee that this discussion goes on and on. Many people don't bother to do a google here or on rec.climbing to see what the facts are and make up their minds, they just say something like "You must always back up a fig-8 with an overhand knot; no, with a double overhand; no with a TRIPLE overhand!" The only evidence they offer is "Because some old fart from the Sierra Club said so." Hey, wait, I'm an fart from the Sierra Club--never mind.
There is one basic fact: just about any knot is strong enough without a backup (or finish knot or dress knot or whatever you want to call that useless extra knot). Clyde Soles knows it, Chris Harmston knows it, Brian and Macca know it. Heck, even I know it. BELIEVE IT. The contest is not to the strongest knot; if it were, the Yosemite bowline would win after a few simple tests.
Beyond that, your choice depends on your selection of criteria. Among the objective criteria might be: ease of tying, ability to be tied with one hand, ability to be tied without taking one hand off the rope (or a hold), ease of tying with gloves on, minimum amount of rope consumed, ease of being untied after loading or when frozen, speed of tying and untying, ability to withstand "ring loading" (the loading you would get if you clipped something into the loop formed through your harness). In all of these, the Yosemite bowline wins, too, as test data and my experience have convinced me. If someone disagrees, let them offer the data to support their view.
Then there might be subjective criteria: If you are already familiar with some knot, should you change? If more people are familiar with another, less satisfactory knot, should you use it for that reason alone? Are you more comfortable emotionally with a particular knot because it seems to have passed the test of time when used by others? If others are more familiar with a particular knot, does that make it better because it is more readily inspected (or, on the other hand, does an unusual knot get inspected more, while familiar knots get less notice)?
Then there might be statistical criteria: In actual use, how many failures have there been? So far as I have been able to research, no properly tied knot of any kind has failed when used as a tie in. Certainly not the common knots. The failures I have been able to dig up were either due to improper tying (Lynn Hill), or to ring loading (rethreaded fig-8 and rethreaded overhand are alone in this problem).
Then there are statutory criteria: Does your organization require a particular knot be used, even if they have no objective reasons for doing so? (Sierra Club, for example).
The discussion would be a lot shorter if people would begin by announcing their main criteria up front. Instead, most posters simply repeat what they have read and agree with for subjective reasons, as in "I was always taught that the best (or only) knot is a figure-8 with a double overhand finish" Then people dive in with discussions of what knot they think is best. No one offers a list of their criteria or test data, or even original reference. Thus grows the folklore of climbing.
For me, the most important criteria (several decades ago, when I first made an informed decision on this issue) were objective. I figured I could deal with the subjective criteria as separate issues. I chose the Yosemite bowline (back then it didn't have a name). If someone else begins with different criteria, they may reach a different conclusion. BTW, in a photo in the National Geographic of May 2003 (the one with the famous photo of Ed Hillary on the cover), you can clearly see the tie-in knot in the rope at Hillary's waist that binds him to Tenzing Norgay: it's a bowline!