the weakening effect of knots ...just doesn't usually matter ...
An absolutely wrong statement ! Do not tell it to anybody that has something hung over his head,
and was forced to use a knot to tie it somewhere...
There are a lot more at stake in the strength of knots, than a released fish ! 
I hope we can move past this debate, with this
understanding : the state-of-the-practice, in various
fields, uses knots without any really good basis for
knowing their strength --as Xarax seems to demand--
other than that
this is how things are done .
We might learn, through hands-on, out-in-the-field
surveys, that in fact there are here and there some
wells of experiential knowledge of a *strength* that
belies that got from the test device --one born of
wear & tear and repeated loading cycles rather than
a single slow-pull to rupture!?
.::. In short, measurements of strength, to my awareness,
have been well short of the degree of thoroughness and
detail and breadth to provide much guidance for use.
Moreover, a review of the history of usage will show that
in many / most cases,
knots do not break in normal use--that just isn't a problem crying out for a solution. (I'm reminded
of one sailor's remark about docklines surviving a hurricane when
tied in
bowlines vs. those w/eye splices --knots didn't break,
though some lines did break, elsewhere (likely over an edge).)
And to hypothesize degrading conditions or extraordinary
loads that
might fall
just in the small range of forces
that could make a difference between different knots
is to indulge in statistical insignificance & speculation!
Another aspect of knot strength might be that of cumulative
effects on the material : that using a weaker knot will lead
to a greater material weakening over time than using a
stronger knot. But this is mere speculation, and it might
be that the difference between knots of reasonably good
strength/efficiency (say, 60% vs. 80%)
at normal loadswill be so small that one should expect other effects of
material deterioration to be still what determine when
to retire or downgrade that cordage to lesser uses.
(Rockclimbers, e.g., sometimes cut of ends of their
long lines and use the remainder; testing has shown that
the ends of such lines will be more weakened than the
central parts (not necessarily from the knot tying and
compressing, but also the rope being drawn through
a 'biner on each fall --coming more at a small distance
from the end ("run out")). )
Furthermore, I must emphasize my notation " *knot* "
in this : if we are talking purely of the thing
representedby, e.g., ink on paper --schemas for formation--, the general
geometrical form, it is unlikely that such entities are best
for associating with a *strength* value
--in contrast to
some particular material so-knotted.
I.e., it might be the case that a
bowline is stronger in
some material vis-a-vis some other knot than in another
material. (I've speculated that HMPE cordage might be much
susceptible to rope movement and generated heat, and so
do better when knotted in ways that restrict movement;
another material --an aramid copolymer, Technora, e.g.--
might do much better when knotted where movement occurs,
vs. HMPE?!)
Then you might ask yourself why the most common bend ever is the [Offset] Overhand Bend (EDK).
Even rock climbers seem to prefer this bend over bends with much higher knot efficiency.
Yes, because strength of any knot here is more than adequate.
(Counting "most common" is tricky : is it what gets tied
the most? --probably the climber's tie-in knot.
Or what appears in the most cases along a climb?
--maybe the
grapevine bend in formed slings
(or the
water knot with tape), even though these
knots are pre-tied to the climb. Even in a multi-pitch
abseil, I suspect the ARJoining bend is tied once and
re-used, not re-tied each pitch.)
--dl*
====
