Summary

  • RIPE NCC's NRO 2 November conflict file contains seven paired groups. The 3 November file contains fourteen. All seven additions pair an ARIN IPv4 line with a RIPE NCC IPv4 line, including two entries for the exact 187.78.192.0/18 interval.
  • Every RIPE NCC interval in those seven new groups appears as an exact row in its 2 November extended delegation original. Neither ARIN's 2 November nor 3 November complete extended original has an IPv4 row overlapping any of them. The report and the current input surface tell different stories.

The word conflicts sits on the public file like a red stamp. On 2 November it separated seven pairs of records; on 3 November it separated fourteen. The seven newcomers are all ARIN against RIPE NCC, all IPv4. A reader skimming the file could reasonably think that two registries currently claim the same address space. That reading fails when the daily source files are checked.

Take the cleanest example. The 3 November conflict output places an ARIN US line for 187.78.192.0, 16,384 addresses, beside a RIPE NCC RO line of exactly the same size and start address. It marks both assigned, dates the ARIN line 20 February 2025 and the RIPE NCC line 25 February 2025. That is one /18, not two neighbouring pieces. The apparent duplication has all the ingredients of a headline.

But the ARIN original dated 2 November contains no IPv4 row overlapping 187.78.192.0/18. Neither does its 3 November successor. The RIPE NCC side is present in its 2 November original, which calls that /18 allocated, and RIPE NCC RDAP returns RO-M247RO-20250225 for the same start and end addresses. Even the status label differs between the NRO conflict output and the RIPE NCC daily original. A file named for conflicts is not a shortcut around reading the records it compares.

The other six additions follow the same pattern. They identify RIPE NCC intervals beginning 64.254.88.0, 140.184.8.0, 140.184.28.0, 140.184.96.0, 153.52.128.0 and 208.90.252.0. The RIPE NCC original contains each exact interval. Across the complete ARIN 2 November and 3 November originals, none has an overlapping ARIN row. Some conflict-output ARIN lines are larger parents than the RIPE NCC child; the current ARIN files do not contain those overlapping parents either. Seven checks, one result.

The older seven conflict groups remained in the output. Among them are IANA-reserved documentation and protocol ranges; they are a separate population from the new ARIN–RIPE NCC set. Counting fourteen without separating the seven additions from the inherited seven turns a file-size change into a claim about current duplicate holdings. The issue is narrower and sharper: a daily aggregation has published ARIN lines that cannot be found in either of two current ARIN daily originals.

The public output does not identify the ARIN input revision or transformation step that produced those seven lines. A dated input manifest and row-level transformation trace would locate the divergence. Until those are available, the defensible finding is the mismatch in published evidence, not a theory about what broke inside the NRO job.