Handling of "no applicable method"

[fritzson@PRC.Unisys.COM (Richard Fritzson)]

When I mistakenly apply a slot accessor on an entirely inappropriate
object (such as NIL), I would expect to see something like "No
applicable method". Instead the method for "slot-value-using-class
(standard-class)" is invoked on the object producing very bad results.
(On the Xerox machines a lower level error,"ARG NOT IWMC-CLASS", is
produced; on some unix lisps, including kcl and franz, you get a
segmentation fault.)

I can see a use of slot-value being optimized into this situation, but
why do accessors produce it too?