Hi folks, On Thu, Jul 13, 2017 at 5:28 AM, Susan Payne <susan.payne@valideus.com> wrote:
Correct, the data for two registrars was excluded because of concerns about mining, but Analysis Group clearly were not confident that fully addressed the concern:
Yes, but remember those 2 registrars already accounted for 62.2% of the data. So, we're left now with only 37.8% of the entire data set at this point. How much more data should be "thrown out"? Presumably that 62.2% of the data had an "abandonment rate" of close to 100%, as it would have moved the total abandonment rate from the 93.7% figure (of the subset) to 99% (of the entire universe of observations).
We are wasting our time with this. What would not be a waste of time would be to identify the data that we wish we had and make recommendations such that any future review would not be similarly hampered.
We can do things fast, or we can do things right. I think it's our duty to do things right, and use the data we have (and it's clear that The Analysis Group did have lots of data; e.g. they would have access to *time-stamps*, because they were able to detect "bulk downloads" that happened to have the same time-stamp; that means one could, as I suggested, filter out the earliest part of the claims period, where any additional undetected "mining" would have been greatest). And obviously the registrar-by-registrar data exists, too (which can be anonymized), since they excluded 2 of them. Some folks seem to be working on an assumption that if we don't look for data or don't use the data we know exists (e.g. the top 500 strings requested via the TMC, instead of just the top 10; a longstanding request for that one), that supports the status quo re: RPMs, or it might also expedite a Round 2 (if future rounds of new gTLDs are even justifiable). I disagree with that assumption. The new gTLDs were an "experiment" that seems to have failed, and the burden should be upon those that pushed for new gTLDs and pushed for these RPMs in the first place to overcome the objections of those who disagreed with them back when they were proposed. To overcome those objections, data is required. The data that we've already seen analyzed by others (the 93.7% abandonment rate, gaming of the sunrise, top 10 most requested TMCH strings turning out to commonly used dictionary words like "ONE" as opposed to strong fanciful marks like EXXON or VERIZON, etc.) already point against the continuation of these RPMs. I think the fear by some is that what's already a strong case against new gTLDs and the RPMs will be made even stronger once we've analyzed more of that data (which has been withheld from our group and/or the public). Sincerely, George Kirikos 416-588-0269 http://www.leap.com/