So, requests 3 and 4 on our list were:

3 - Can providers give the WG some general information about the percentage of requests for disclosure that are successful

4 - For Q4, do providers also have information about the type of claims those relate to e.g. If they are from LEA, 3P IP claim etc.?

 

And I agreed to provide some colour as to why these are probably difficult questions to answer.  The below should be true for most providers (and really, for any support system), but should any Registrar colleagues disagree, feel free to chime in.

 

The concise reason is that compliance desks are optimized for throughput of issues, and not for policy related reporting.

 

In more detail -

 

Compliance (or whatever dept that deals with issues around domains) may not use a ticketing system.  If the provider is small enough, it may just be an email address.   There is no obligation for anyone to use a system that is even capable of producing data, or from which data may easily be extracted.  

 

Even if they do use a system robust enough to allow for reporting, the data we're requesting may not typically be captured in the course of an investigation, or even useful for the running of the business. For question 3, for example, 'successful disclosure' is mostly relevant to the requester, but is not an attribute required to close the ticket.  That information may be captured as text inside a description, but if it's not a specific flag on the ticket, it's very very difficult to report on accurately.  We at Tucows occasionally mine the text of the tickets related to our retail operations, and the results are always considerably noisy.

 

Question 4 presumes that requester type as well as as issue type are ticket attributes.  I would think that some such systems do capture the issue type (IP, defamation, offensive, etc etc) but each service provider is going to have differently defined categories that may cause difficulty in the aggregation.  As well, just because a claim is made, doesn't mean it has any particular validity which impacts categorization.  If someone claims IP infringement, but our investigation yields no such issue, does the category still stand?  As well, because requests are usually dealt with manually, categorizing requesters may be unlikely, aside from a distinction between LEA and 'other'.

 

Lastly, as should be clear from previous discussions, there are very few black and white issues that we see.  Given the variety of issues and the multitude of interest areas that they cross, straightforward classification is exceptionally difficult, and analysis that attempts to place requests into distinct buckets is going to lose a substantial amount of detail.  Nonetheless, I'm in the process of looking into our system at the moment to see if possible to sketch out some trends.  Unfortunately, we migrated to a new platform earlier this year, so there is not much to draw from.


Graeme