[rpd] Hierarchical AS-SETs objector analysis -> astroturfing

[hvisage at hevis.co.za]

I gave Claude Cavemen the mailing list archive, and requested it to 
analyze the objectors’ question and answers and why this could be 
classified as Astroturfing… the humans with brains in here can stop 
reading as they had already seen all the signs, but for the AI writing 
tools, the summarised version… the name references at the astroturfing 
section was Fable 5 that made/found/linked it, not me… just… 
revealing

---

## 1. Questions the objectors POSED — and the answers they received

Every substantive question they asked **was answered on the record**:

| Their question | Answer given | Where |
|---|---|---|
| "Does hierarchical naming validate AS-SET *contents*?" (the 
attribution vs content distinction) | **Conceded**: no, and the proposal 
never claims it — scope is name uniqueness + creation attribution 
only. Content = RPKI/ASPA/bilateral territory. | Hendrik 015120, Frank 
015123 ("it doesn't claim to… It creates an improvement") |
| "Show the exact operational harm" | MANRS AS-AMAZON incident (2022); 
live empty AS-GOOGLE in AFRINIC DB vs populated RADB one, pasted 
side-by-side; Jaco Kroon's first-person upstream-collision testimony | 
Frank's two-object post; 015120; Jaco 20 Jul |
| "Why is this the least restrictive solution?" | Deployment asymmetry 
(one naming rule at one registry vs universal consumer reform); 
squatting has no consumer-side remedy; ~34 voluntary hierarchical sets 
already existed and collisions persisted anyway | 015120 |
| "Where is the proportionality/impact analysis?" | Published staff 
Impact Assessment: minimal member impact, no legal/finance issues; scope 
limits written into 7.8.3–7.8.6 | 015120 + proposal text |
| "Why must AFRINIC follow other RIRs?" | It needn't — reframed: with 
the other four closed, AFRINIC's namespace is the **last open one**; 
residual global squatting surface concentrates here | 015120 |
| "Why the policy route instead of operational?" (Kone — genuine 
dissent) | Community may prefer/enforce via policy; that's its 
prerogative over the RIR | Jordi Palet 015195 |

**None of the answers received a counter-engagement.** The objections 
were then restated as if unanswered.

## 2. Questions ASKED OF the objectors — and how they responded

| Question | Response across 23 accounts / 161 messages |
|---|---|
| What should filter software do with the empty AFRINIC AS-GOOGLE vs 
populated RADB one? (Frank) | **Zero answers.** ≥4 direct replies to 
Frank skip it. Only cohort message containing "AS-GOOGLE" quotes Frank's 
own words via digest. |
| What remedy does a squatting victim have under your alternatives? 
(Hendrik 015120) | **Zero answers.** Alternatives lists kept being 
reposted without touching it. |
| "Do you agree it provides improvements?" (Frank) | **Conceded by 
presupposition** — "An improvement is not automatically a 
justification for mandatory policy" (Thulisile 015130). Never a plain 
yes, never a no. |
| What evidence threshold would satisfy "proven enough / proportionate 
enough"? | **Never named.** The bar stays undefined — unfalsifiable by 
construction. |
| Which specific Impact Assessment finding do you contest? | **IA never 
once mentioned** by any objector, while "the analysis has not been 
completed" kept being asserted. |
| Would new participants introduce themselves? (Ben — invoking the 
objectors' *own* "real person who stands behind it" test) | **Refused** 
("not a membership interview"); three new accounts appeared within 22 
minutes instead. Sole compliance: Mandla Matshika — "Policy insights 
contributor at NRS". |

Their only direct "answers" = concessions: mechanism works, it's an 
improvement, harm evidence is "useful evidence," and their stated 
technical case = the already-conceded scope point.

## 3. Why this classifies as astroturfing

Astroturfing = manufacturing the *appearance* of independent grassroots 
opinion. The behavioral evidence (no stylometry needed):

1. **Manufactured breadth**: 23 accounts first-seen in the weeks 
before/during Last Call, mostly `name+digits at gmail` + student mailboxes, 
producing 161 messages — vs 2 established participants opposing 
anything all year. 72 of 74 oppose stances in the Last Call window came 
from these accounts.
2. **One voice, many names**: five distinct arguments total; everything 
else restatement. Same vocabulary kit ("mandate laundering", 
"invariant", "gatekeeping", "Participation is not mandate") crossing 
accounts; the same new talking point delivered by two personas within 
one exchange; cross-account signature slip (gugudhlamini343 message 
signed "Nonhlanhla"); verbatim-copied objection between two senders 
(0.985 similarity, per our dossier).
3. **Coordination artifacts**: literal `[Name]` template placeholder 
sent as a signature (015134); contradictory explanations of the same 
double-post 8 minutes apart; two display names each operating two 
mailboxes; batch digest-reply workflow; three fresh accounts 
materializing in 22 minutes when introductions were requested.
4. **Asymmetric responsiveness**: supporter missteps drew replies in ~20 
minutes; the five concrete questions stood unanswered >1 day across the 
whole roster — the signature of a template pipeline that can restate 
but not adapt.
5. **Attention exhaustion as the mechanism**: volume aimed at swamping 
Last Call and manufacturing "unresolved objections," not persuading — 
what Mike Silber called an "AI generated DDoS on the PDP" (015179).
6. **Link to a known astroturf actor, self-declared**: Mandla Matshika 
introduces as "Policy insights contributor at **NRS**" (015177); 
Nia/Nonhlanhla cites **Lu Heng's** "Policy Mirror" as framework (015011) 
— NRS/Larus/Lu Heng being the named actors in AFRINIC's own 
2021–2023 misinformation warnings.
7. **Independent recognition**: not one observer's read — Omo Oaiya 
("coordinated or based on common templates"), Silber, Sowe, Noah, Ben, 
and sitting co-chair El-Nakhal himself: *"astroturfing is the best term 
to define such emails"* (015029).

The rough-consensus consequence (RFC 7282, per Alston 015166): 
objections count when grounded in evidence and *sustained through 
engagement* — this corpus reduces to two issues, both answered, one 
mutually agreed out of scope, echoed by many mouths. That's the 
definition of astroturf: headcount pretending to be argument.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.afrinic.net/pipermail/rpd/attachments/20260721/c1ca762f/attachment.html>