As of July 6th until July 13th there are 5 rows that have a 6th column. There are 4 rows that have an extra row (I believe these are cloudflare IPs). 104.28.4.92/32,AU,,Newcastle, Nsw, 104.28.4.91/32,AU,,Newcastle, Nsw, 104.28.4.81/32,LB,,Beirut, Lebanon, 104.28.4.82/32,LB,,Beirut, Lebanon, vs. 146.75.228.14/31,BS,,, 146.75.228.16/31,BS,BS-NP,NASSAU, 146.75.228.18/31,BZ,,, Simple check to show commas per line. The 3 0s are blank lines which most CSV parsers handle gracefully. Most lines have 5 commas, but there are a few with 6. awk -F ',' '{print NF}' egress-ip-ranges.csv | sort | uniq -c 3 0 17675 5 4 6

It has been heavily implied but I can't find confirmation anywhere that the "Tracker Only" IP address hiding is actually using the same system. Can that be confirmed? Early testing with Beta2 in the US header bidding market "Tracker Only" doesn't appear to be active despite the sites in question being listed in the Privacy Report as trackers. Is there any clarity on when we would expect that feature to be enabled in the greater ecosystem?

Will there be more details on the egress IP configuration/policy? My team and I have a number of questions (focused mostly on US currently): How often will the egress IPs change? Will there be schedule of when changes will apply or will always be 'live' and it's to geo location providers to poll API frequently to notice changes that have already occured? I notice the mask-api response has cache control of 3600 seconds, does that imply an hourly poll for changes would be acceptable? Will the 'shapes' associated with these regions be published? It's not clear how the assignments might be made (see image below). Some areas have quite granular cities near to each other, others appear to be somewhat randomly selected cities in less dense regions. Is it as simple as Voronoi_diagram or something more complex? It could also be that there is an non-geographic atom underlying this mapping like ZIP but the assignment problem still exists. An explicit mapping would help the industry handle these changes while still preserving user privacy. Are all the egress providers using the same mapping rules and regions? I notice some differences in names between fastly/cloudflare which seems unlikely if they were being provided a list of shapes and names. Can anyone confirm that "Use broader location" will randomly select a random region somewhere in right timezone rather than a broader bucket like "US Eastcoast" (though I did see a blanket "US" IP range)?