Summary
- RIPE Policy Proposal 2026-02 is Denis Walker's single-author draft, version 1.0 published 13 October 2026: a trigger of 14 cumulative days inside any rolling 30-day window, then a strict cure of seven business days.
- Its rationale justifies uniform operation "across all 70+ jurisdictions" while no reviewed page lists those jurisdictions, names who counted them, or prints the population against which the count is a fraction.
The document is short and unusually confident. Its enforcement clock is written to the day and then to the business day. Its justification is written to the nearest round number with a plus sign bolted on. Both sentences sit on the same page, and the gap between them is the whole story.
Denis Walker's proposal is titled "Contractual Accountability Framework for Infrastructure-Level Network Abuse". The RIPE NCC page for it prints version 1.0, publication date 13 October 2026, the Security Working Group as its home, and one author with no co-author beside him; the annotation under his name reads "Retired RIPE Database Senior Analyst and Engineer" Proposal 2026-02, RIPE NCC. Nothing about the authorship is ambiguous, which matters for a proposal whose subject is who answers for what.
The meter
Here is the trigger, quoted from the policy text as the page prints it:
Upon receipt of an authenticated notification demonstrating that an IP address within a RIPE NCC allocated or assigned block has engaged in Network Abuse, as defined in Article 1.1, for a cumulative total of 14 days within any rolling 30 day window, as documented on an Approved Infrastructure Blocklist, the RIPE NCC shall initiate an administrative delivery protocol. Proposal 2026-02
Read that as arithmetic rather than policy. The unit is a day. The count is cumulative, so fourteen scattered days accumulate; they need not be consecutive. The container is a rolling window, so the count is always being re-measured against the last thirty days rather than against a fixed calendar. The evidentiary condition is a notification that is authenticated and documented, and the clock starts from a fact that a third party recorded, not from the registry's own observation.
Then the cure:
A resource holder is granted a strict grace period of seven business days from the confirmed timestamp of electronic delivery to achieve compliance. Compliance is achieved when all identified abusive prefixes successfully drop off the Approved Infrastructure Blocklists, or when the resource holder takes verified administrative action to isolate the abusive prefixes. Proposal 2026-02
Two clocks, two units, and it is worth naming the difference because a compliance officer will meet it at two in the morning. The trigger counts calendar days. The cure counts business days. Both are anchored to a timestamp rather than to a letter arriving, which is the sort of detail a lawyer writes and a slogan cannot fake: the delivery protocol produces a confirmed electronic time, and the seven business days run from that instant.
That is a meter a member can actually argue with. A holder facing it can ask which notification, authenticated by whom, documenting which prefix, and on which day inside which rolling window. Those are answerable questions, which is exactly what precision buys.
The round number
The rationale opens its supporting case with this sentence:
It utilises the official registry contacts that every member is contractually required to maintain, operating uniformly across all 70+ jurisdictions. Proposal 2026-02
Everything load-bearing in that sentence hangs on a noun nobody defines. Jurisdictions, plural, more than seventy. Which ones? A jurisdiction here could mean a country with a national regulator, a legal system in which a contract is enforceable, a tax territory, or a service region that the registry happens to name on a different page. The reviewed page prints the count and does not print the list.
It also does not print the counting hand, and it does not print the denominator. Seventy-plus is a floor, so a reader cannot tell whether the intended population is eighty, a hundred and twenty, or the full set of territories reachable from a RIPE NCC service relationship. A figure written as an open floor with no population behind it cannot be wrong, and that is the problem: as printed, it is datable — 13 October 2026 — and not reproducible. Nobody can take that sentence away and check it.
Set the two surfaces beside each other. Where the clause constrains a member, the drafting is exact to the business day, because the author knows the clause will be disputed and wants it to survive the argument. Where the prose persuades a working group, a floor with a plus sign will do, because it will be read once inside a rationale and never audited. Precision tracks who bears the cost.
What the same registry prints elsewhere
This is not a claim about what a template permits. The same publisher, in the same season, prints the note that answers the question on a sibling page:
Impact Analysis Note: In order to provide additional information related to the proposal, details of an impact analysis carried out by the RIPE NCC are documented below. Proposal 2026-01, RIPE NCC
That is the 2026-01 page, whose proposal is in a concluding phase and which carries the RIPE NCC's own analysis on the page. On the current-proposals index as reviewed on 22 November 2026, an impact-analysis link appears against 2026-01 and does not appear against 2026-02 Current Policy Proposals, RIPE NCC. The reviewed text of the 2026-02 page carries no such section.
Keep the boundary tight. This is an absence on the surfaces reviewed on 22 November 2026, not a statement that no analysis exists anywhere. What can be said is narrower and still useful: one proposal in this batch carries an institutional reading of its own cost beside the author's case, and the other — the one asking for an enforcement path — carries the author's case alone. A member deciding how much weight to give the cost framing has one page more to read on 2026-01 than on 2026-02.
The status line does the work
The page dates itself in two places. State: "Discussion Phase: Awaiting decision". State description: "Awaiting new version to start new Discussion Phase" Proposal 2026-02. Those two lines keep everything above hypothetical. The days are counted; the counting binds nobody yet.
The chairs' own record agrees, and adds the objection that matters most to a clause this exact:
The discussion phase for 2026-02: "Contractual Accountability Framework for Infrastructure-Level Network Abuse" ended on 11 November 2026. [...] Concerns were also raised about the lack of clear due process and appeal mechanisms. Security Working Group chairs, RIPE NCC list, 13 November 2026
Read that sentence for what it is: the chairs reporting that participants raised an objection. It is not the chairs' verdict on the draft, and it is not a provision of the draft. Both distinctions matter, because a meter with a seven-business-day cure and a rounded justification is precisely the kind of instrument that invites a due-process question, and the record shows the question was put rather than answered.
The same message announces the revision without dating it: "Denis is working on an updated version taking in consideration the feedback received. A new discussion phase will start when the new version of the proposal is published" Security Working Group chairs. No publication date for that version appears in the message or on the reviewed page.
My own subtraction, offered as mine and not as a printed figure: 13 October 2026 to 11 November 2026 is 29 days, and the status line still read "Awaiting new version" when I reviewed it on 22 November 2026, 40 days after publication. A month and a half is not a scandal. It is context for a reader who needs to know that the version with the days in it is still a draft.
The stop, named once
Two limits hold for everything above. The clauses governing what happens after the seven business days were outside the text examined for this piece, so their terms are not characterised here. And the missing impact analysis is missing from the four surfaces reviewed on 22 November 2026; whether one exists elsewhere was not established. The identity behind the count — who counted, which jurisdictions, against what population — is printed on no reviewed page.
Those limits leave the finding intact. A proposal is a machine for turning a member's behaviour into a number, and the number that would let a reader check its own justification is the one number the page withholds. The next version can settle it cheaply: print the list, or drop the count. If it prints the list, "70+ jurisdictions" becomes a checkable claim about where this rule would run. If it prints a different round number, the arithmetic will tell us which side of the page the author was watching.
Member Briefing
Read the full trend briefing
Members read the full briefing with the source notes behind every finding.
Strategic Circle
A daily briefing in your language, built from changes we have verified.
Join for USD 20/monthCancel anytime.
Members also get the full research library, a daily briefing on the networks they follow and the private forum.






