[CPWG] Bulk Registrations, missing data and assumptions?

[John McCormac]

Just reading through the agenda for today's meeting and the section on 
DNS Abuse. DNSAI, in the letter to Sebastien Ducos, mentioned that it 
had no data on bulk registrations. This is not surprising as much of the 
bulk registration activity took place before the founding of the DNSAI. 
It still happens.

The problem with bulk registration is that it is intrinsically linked 
with certain types of websites and other activity such as spam and malware.

A few years ago, there was an effort by CENTR and a few European 
registries to try to classify web usage (I think that there was also an 
Registry Registrar Data Group effort (RRDG) which suffered from a 
similar lack of data.). Some of it followed the inaccurate attempt to 
measure usage that was naively initiated by the CCT as an input to its 
deliberations. Of the supposedly developed sites in the new gTLDs 
mentioned in that CCT document, approximately 80% of the domain names 
had been deleted by the time CCT published its final report.

The five year deletion rate for heavily discounted registrations is less 
than 1%. That means that just over 99% of bulk registrations will not be 
in the same gTLD's zone file five years later. As a numerical example, 
.LOAN had 1,866,032 registrations in February 2018. In February 2023, 
2,227 of these domain names were still in the zone. That's approximately 
0.12% of the February 2018 domain names.

Measuring web usage is a lot more complex than merely breaking down 
websites by whether they are "parking" pages or have "rich" content. 
Such crude methodologies don't work with the bulk registrations problem.

The ccTLDs are highly concentrated markets while gTLDs are often 
composite markets (registations from hundreds of countries) with a small 
global market. As a result of this, ccTLDs and gTLDs will have an 
overlap on some types of DNS abuse and will also have concentrations of 
some types.

With the data, bulk registrations are apparent as spikes in new 
registation volume followed approximately a year later by a spike in 
deletions. (I've run this analysis back to January 2004 on the legacy 
gTLDs and also on the new gTLDs back to 2014.)

The term "bulk registrations" is problematic because it ignored the 
primary economic factor: heavily discounted registrations. Without these 
heavily discounted registrations, the economic model of bulk 
registrations would be much less viable.

Some of the references quoted by the DNSAI in the letter mentioned 
above, such as the Interisle report, are quite good. But bulk 
registrations aren't typically a ccTLD issue (apart from repurposed 
ccTLDs). This means that much of the analysis based on ccTLDs is useless 
in this context. Bulk registrations are very much a shifting target and 
that makes it difficult to measure their effects on the webscape with a 
good methodology and impossible with a crude one.

The phrase "bulk registrations" does seem to be a problematic one. 
(There seems to be some pushback from the registries and registrars on 
this. Apart from measuring the effects with the data, it might be 
necessary to reframe the issue.

Regards...jmcc
-- 
**********************************************************
John McCormac  *  e-mail: jmcc at hosterstats.com
MC2            *  web: http://www.hosterstats.com/
22 Viewmount   *  Domain Registrations Statistics
Waterford      *  Domnomics - the business of domain names
Ireland        *  https://amzn.to/2OPtEIO
IE             *  Skype: hosterstats.com
**********************************************************

-- 
This email has been checked for viruses by Avast antivirus software.
www.avast.com