[OT] trademarks (was: Re: Bundling)

John C Klensin klensin at jck.com
Sat Dec 12 17:45:40 CET 2009

--On Saturday, December 12, 2009 10:21 -0500 Vint Cerf
<vint at google.com> wrote:

> Mark, et al,
> this may not reach you if you are off on holiday but I would
> point out to others on the distribution that ICANN's Uniform
> Dispute Resolution Process does take into account trademark
> arguments and these often factor into the recommendations made
> in this non-mandatory arbitration method. Trademarks are not
> uniquely assigned, as I think we all know. Trademarks for
> Mumble-frotz skis and Mumble-frotz butter can be held by
> distinct parties in most trademark jurisdictions. Things get
> more complicated in the case of so-called "famous" marks, I
> believe.


In case it wasn't clear, my question to Mark was trying to
explore some issues wrt the discussion in this WG.  I am
certainly not a trademark lawyer, nor trying to second-guess his
remarks or those of his wife, just trying to understand
relevancy to the WG and its work.  Specifically:

	(1) Were whatever conventions about matching of
	characters for trademark purposes in fact universal
	--any two strings that differed by only that one
	character would be treated as identical everywhere and
	for all purposes-- or were they subject to different
	interpretations depending on jurisdiction and use of the
	strings?  The answer to that appears to be that there
	may be more uniformity than I expected, but there are
	still no universal rules, it is still a matter of local
	judgment.  Moreover, the conflict between the ability to
	use the same (exact) string as a trademark for different
	purposes in different jurisdictions and the global
	uniqueness of DNS nodes, and hence the difficult of
	drawing analogies from one system to the other, is one
	that has been understood for two decades, discussed with
	and in WIPO, etc.
	(2) If we tried to make decisions based on trademark law
	or practice, would they be particularly helpful with,
	say, Sharp-S and "ss"?   I think Mark's response is
	consistent with what I had understood from previous
	conversations, particularly those about the UDRP:  for
	trademark purposes, the criteria for confusability are
	associated with use and extend well beyond either the
	visual or character substitutions within a given script.
	Those criteria extend to "sound alike" as well as "looks
	alike" and might make Eszett confusingly similar, not
	only to "ss", but also to characters that produce more
	or less the same phoneme ("more or less" because "bad
	accent" doesn't typically solve trademark or other
	confusability problems).   So, to the extent to which
	trademark conventions could be used to demonstrate that
	"ß" and "ss" are really the same, they could be used to
	demonstrate that any character, in any script, with
	approximately the same phonemic properties is also the
	same.  As you know, that criterion as been discussed in
	ICANN or elsewhere as a requirement on registries to
	prohibit registration but finally rejected, in part
	because it would lead to madness and/or to making IDN
	use in gTLDs nearly impossible.  Instead, situations
	involving similar phonemes are resolved on a
	case-by-case basis that includes examination of intent
	and the likelihood of actual confusion in context.
	While the UDRP applies only to gTLDs subject to ICANN
	oversight and to those ccTLDs and other registries that
	have chosen to adopt it or use it as a model, its
	development and application history (in which WIPO, as
	well as ICANN's intellectual property constituency, had
	significant roles) are, I believe, entirely consistent
	with the above.

My conclusion is that trademark practices are very interesting
but are not very helpful to us in determining whether "ß" and
"ss" are somehow the same, whether one should be more or less
permanently mapped to the other, or whether one should be

YMMD, as usual.


More information about the Idna-update mailing list