Summary

The extra eleven are cuts, not gifts

The arithmetic is unusually clean. On one day AFRINIC carries a single available /37. On the next it assigns the first /48 of that span to F36F3EF6 and writes down every leftover piece separately. The first remnant is the adjacent 2001:43fc:9801::/48; the next is 2001:43fc:9802::/47. The sequence doubles the size of each free piece through 2001:43fc:9c00::/38. Taken together, the assigned /48 and eleven free pieces are disjoint and fill the original /37 exactly. No new IPv6 space arrived. The ledger acquired more edges.

AFRINIC’s own guide warns that its summary line counts records. Here the warning has a concrete face. The old one-line /37 becomes twelve IPv6 lines. Twelve minus one is eleven, precisely the movement from 9,455 to 9,466 in the file’s IPv6 summary. Reading that rise as eleven new awards would turn a formatting consequence into a resource story. The award inside this /37 is one /48.

The recipient is more than an opaque statistical code. In the same daily comparison, AS329844 moves out of available ASN stock into an allocated Kenyan row dated 2 November, while 102.200.180.0/23 moves from available IPv4 stock into an assigned Kenyan row dated 3 November. The IPv6 /48 is dated 2 November. All three carry F36F3EF6. The holder code alone names no company; AFRINIC’s exact RDAP makes the join. The three resource objects name ORG-WL14-AFRINIC, and the entity record calls it WHIZMO LIMITED. The ASN was registered on 2 November at 09:08 UTC, the IPv6 /48 at 08:58, and the IPv4 /23 on 3 November at 06:55. Those timestamps show a staged administrative arrival across resource classes.

The network view remains separate. At midnight UTC on 4 November, RIS’s exact-prefix queries show 0/322 for the IPv4 /23 and 0/313 for the IPv6 /48. This is a dated view from its collector set. The registry transaction is concrete whether or not the routes are visible at those observation points. A later exact or more-specific announcement, or an operator activation record, would date the operational step. It would not change the fact that the apparent eleven-record jump came from subdividing one available /37.

There is another change in the 4 November file: AS328472 and 102.67.172.0/22 appear as reserved after prior allocated rows. That separate status conflict already has its own article. It is excluded from this count. The Whizmo case turns on a simpler, harsher question for anyone quoting registry statistics: did the number of things change, or did the number of boxes used to describe them change?