[Dan Lehman]

[snipped from Reply, over quota ...]

This one, sadly, resonates as true. It's a patronizing, even contemptuous motivation, like teaching people pidgin English, because they aren't intelligent enough to handle real English. Those climbers, by and large, are not stupid; they just aren't familiar with knots. And even really simple knots can be --and are, sometimes-- tied incorrectly. So climbing instructors are perpetuating inferior knots for everyone, for no good reason.
Rising nobly above this situation, I would stick to the question, of what is a good bend, and this can be answered with hard data. I'm a bit overextended on tests at the moment, but if anyone out there wants to pester someone into doing tests, you are looking for both strength and security.

It's a tough question, at times : to choose the *safer* path and guard against possible
mistake can be seen to protect ignorance from being overcome. OTOH, one can point
to mistakes happening. (I recall being amazed at RC.com folks not comprehending the
diff.s between square & granny & thief & whatnot/grief !) Some SAR (et al.) folks insist
on a "back-up"/"safety" knot --to make any failure overcome two tyings.

But it's also arguably presumptuous to claim to know better than what has been used
now for many decades by thousands and thousands of rockclimbers. How much testing
are you going to do, to achieve that frequency? --though done w/o special notice to
the particular form/geometry of the knot, still, with such numbers, it's hard to think that
varieties escaped some use.

For the latter, note that many knots can be "dressed" in more than one way, most often in how the ends lay inside the knot. With the Strait Bend, for instance, they lie alongside each other, and either one might be "on top."

Indeed, as my RC.com reference at the top shows, even the TIED offset water knot
can be *dialed* into differing orienations, where at one extreme the thin line loops back,
and the other extreme it arcs forwards with the thicker rope looping : does that matter?
(unlikely, re security & flyping, at expected loads). The butterfly --known earlier as
the "lineman's loop"-- was specified to have its eye legs (tails, were it end-2-end)
crossed a particular way, by discovers Wright & Magowan (1928); but it is seldom
presented in this way, usually with the simpler ends/legs-abutting orientation. Similar
variations exist for Ashey's bend (#1452) & #1408 & the zeppelin.

Back to those usually urged "ridiculously long tails" of the infamous "EDK" : yes, that
has the likeness of saying "oh, that street's perfectly safe at night --just wear a flak vest
and carry an AK-47". Rather than leave such material in case ..., my urging is to DO
something with it --and tying off the thinner (if ...) tail around the other, with an overhand
snugged to the main knot, puts use in the structure, preventing the feared rolling,
rather than being there (at some remove) to somehow nip it in the bud should it occur.

But back to my early point : for all the loading this knot will see in practice, each person
has the ability to do meaningful testing, loading, bouncing, knocking about their own
knotted ropes, variously tied. And I think that just focusing on the "EDK" and tying it
purposefully will be the right course --advance past superstition, and decline the novelty
knots conjectured as somehow better.

--dl*
====

ps: Re Tom Moyer's testing, one can see that even with the offset fig.8, the more
risky knot --more vulnerable to flyping--, it took some load to flype it.
(I see his note

Dan Lehman has also proposed some variations of the overhand to me that look very promising. They keep the asymmetry and are all probably much harder to flip than the overhand. If I get any spare time I will test these and post the results.

Spare time has been too scarce, in a decade --and counting !