RE: [gnso-dt-wg] Collecting Facts
- To: "Bruce Tonkin" <Bruce.Tonkin@xxxxxxxxxxxxxxxxxx>, <gnso-dt-wg@xxxxxxxxx>
- Subject: RE: [gnso-dt-wg] Collecting Facts
- From: "Mike Rodenbaugh" <mxr@xxxxxxxxxxxxx>
- Date: Mon, 30 Jul 2007 16:54:18 -0700
Bruce, thanks. I have discussed with Olof and he will engage ICANN
Staff in hopes of gathering data along these lines. Others on the list
have already begun some work in regards to your scenarios #2 and #3, at
least, but you lay out very specific data that seem achievable and
certainly would be useful.
From: owner-gnso-dt-wg@xxxxxxxxx [mailto:owner-gnso-dt-wg@xxxxxxxxx] On
Behalf Of Bruce Tonkin
Sent: Thursday, July 26, 2007 4:54 AM
Subject: RE: [gnso-dt-wg] Collecting Facts
> If a name is "re-registered" in one month, 3 months, 6 months
> or one year
> after the initial registration (and deletion after 5 days),
> how can you call
> that "kiting"? I think Jothan's comment earlier today was
> very helpful in
> drawing the distinction between "kiting" and tasting. See
> http://forum.icann.org/lists/gnso-dt-wg/msg00031.html. We should be
> distinguishing between the "intent not to pay" ("kiting") vs
> "reviewing the
> suitability of a domain name" ("tasting"), and then looking
> for specific
> facts and research on how the five day add grace period may
> or may not be
> contributing to abuse.
Note on the topic of "tasting" - I can imagine that you could "sample" a
domain name at a certain "sample interval", and then keep the name if it
has some value at the time of sampling.
E.g I could create the domain name "icannhasgreatpolicy.com". I could
then cancel the name if it doesn't get any traffic within 5 days. I
could then decide to re-register it every day, week, month, year etc
depending on what sampling interval I choose to use. Hopefully one day
the name will have value.
Note that a particular domain name may be sampled by more than one
party. And the sampling interval could be quite short - in fact as the
zonefile is published - some actors in the market could have a business
model to simply sample everything another party rejects - on the basis
that the first party generally has good judgement in picking good names,
and the second party is better at monetising a particular name.
The number of variables becomes quite complex to determine the
motivations of the registrant. You could certainly analyse a set of
names over a period of time - e.g a sample size of several thousand
randomly selected names that were registered and deleted within 5 days,
and then track how often each of these names is registered over a period
of time e.g 3 months. You might find that in a particular sample of a
thousand names, 10 names are registered more than once over a 6 month
period, and 2 name were registered up to ten times. I don't know what
the stats actually are - but at least some basic statistical analysis
could inform the GNSO on the size of the problem.
What is also not clear is what the problem actually is.
Possible problems are:
(1) As so many names are being sampled, potential registrants are
finding it hard to register their preferred name. This could be
determined by selecting a sample size of names that were registered and
deleted within the 5 day period, and then correlating this data with the
number of check operations done on the names within the 5 day period.
(2) Many of the names being registered are being used for possible
trademark infringement. This could be determined by selecting a sample
size of names that were registered and deleted within the 5 day period,
and then comparing the names with a database of trademarks (e.g USA
trademark office). You might want to identify direct matches (e.g
check for icann), and also potentially common misspellings (e.g check
(3) Many names are being repeatedly registered and deleted - leading to
possible stability issues, and potentially escaping payment (e.g
kiting). This could be determined by selecting a sample size of names
that were registered and deleted within the 5 day period, and monitoring
those names over a longer period - e.g 30 days, 60 days, 90 days etc.
Identify how many names are re-registered and at what frequency.
Rather than focus on the motivations of the registrant - it is probably
better to focus on the actual statistics of the activity, which may help
identify the problem, and then determine what policy actions may be
So far the only real statistics I have seen relate to aggregate numbers
of names that are deleted within 5 days, but not the next level of data
as described in the three potential scenarios above.