This support belongs to the Starlink assessment published on 18 October 2026. Its evidence cutoff is 17 October 2026 at 20:49 UTC. Public references and qualified reporting are distinguished from this publication’s interpretation. The six estimates and two withheld estimates are shown with their reasons. Each confidence rating belongs to its specified judgement.
Assessment · Evidence · Full comparison
J01
A service-control or administrative restriction was the principal cause of the reported movement-linked disruption, rather than Russian hostile interference.
Likelihood: Likely.
Confidence: Moderate.
Reasoning: The conjunction of a reported motion-linked boundary, different service effects and a named account separating provider repair from a retained cap fits service administration better than a hostile attack as the whole explanation. This is an inference from qualified testimony, not authenticated telemetry. The precise control, entitlement and intent remain separate.
Evidence sufficiency: Sufficient for a qualified principal-mechanism estimate; insufficient for an exact technical diagnosis or attribution of every outage.
Confidence basis: Named technical testimony and several operator/ministry accounts provide a coherent mechanism-level picture, but source overlap, inconsistent thresholds, imprecise onset and absent logs prevent High confidence.
What could change it: Matched logs or credible corrections showing no service change, or incident-specific hostile interference reproducing the reported pattern, would materially weaken this judgement.
Evidence: E01, E06, E07, E08, E09, E13, E29, E97, E98, E99, E101.
J02
US government direction or authorisation likely contributed to the principal restriction.
Likelihood: Likely.
Confidence: Low.
Reasoning: Reuters reports a White House message reaching defence leaders; a separately bylined Telegraph account alleges a Musk proposal and Trump agreement. General service-governance testimony makes such a pathway coherent. Denials are material counterevidence. The balance favours involvement over a wholly provider-only event, without assuming these reports have independent informants or proving the identity of the author.
Evidence sufficiency: Adequate for a low-confidence involvement estimate; not sufficient for an authenticated order, exact command chain or a personal presidential finding.
Confidence basis: The decisive evidence is anonymous, may share sources, and conflicts with public/official denials. Direct knowledge, source overlap and execution records cannot be checked.
What could change it: A source retraction, demonstrated common-origin error, matched order/provenance, or a documented provider-only change would change this estimate.
Evidence: E91, E92, E93, E94, E96, E102, E105, E106, E107.
J03
Political coercion of Ukraine was a material purpose of the restriction.
Likelihood: Not assessable.
Confidence: No substantive rating is attached to an unmade estimate.
Reasoning: The reported pressure context makes coercion a plausible explanation. Government-origin reporting does not establish purpose, and a non-political government entitlement/policy decision remains a live rival. The affected service reportedly was not used for refinery strikes, weakening direct mission-denial claims but not excluding indirect leverage. There is no sufficiently discriminating purpose-bearing evidence to assign a defensible likelihood.
Evidence sufficiency: Insufficient for a reliable motive estimate; enough to retain political explanations prominently and specify what would discriminate them.
Confidence basis: No substantive confidence rating is manufactured for a withheld likelihood. Source access and causal discrimination are inadequate.
What could change it: An authenticated purpose-bearing instruction, credible direct testimony of deliberation, or a well-matched routine policy/entitlement record would permit estimation.
Evidence: E67, E70, E76, E77, E81, E91, E102, E104, E105.
J04
Trump personally ordered or approved the restriction.
Likelihood: Not assessable.
Confidence: No substantive rating is attached to an unmade estimate.
Reasoning: The Telegraph account alleges agreement while Trump and an attributed official deny responsibility/request. Reuters leaves the actual White House decision-maker unidentified. General authority does not settle personal participation. A single unverified initiation chain and overlapping unknowns do not support a reliable personal estimate.
Evidence sufficiency: Insufficient to identify the personal decision-maker.
Confidence basis: No likelihood or confidence label is invented; this withholding is compatible with the broader low-confidence government-involvement estimate.
What could change it: Direct order/approval evidence, credible independent first-hand testimony, or substantiated correction of the alleged discussion would change the result.
Evidence: E94, E102, E105, E106.
J05
A wholly accidental change, without deliberate direction or policy action, explains the principal restriction.
Likelihood: Unlikely.
Confidence: Low.
Reasoning: Two newsroom attribution accounts and a reported persistent cap during negotiation weigh against the whole event being an accident. Their credibility is conditional and their informants may overlap. Accidental scope, outages or entitlement errors within a deliberate decision remain credible and are not included in this narrowly wholly-accidental proposition.
Evidence sufficiency: Adequate only for a qualified comparison with the wholly accidental explanation, not exclusion of error contributions.
Confidence basis: This estimate depends materially on the attribution accounts and the reported distinction between repaired outages and a retained cap.
What could change it: Removing both attribution chains, or finding a matched accidental configuration/entitlement record, would require reassessment; technical-error contributions remain open already.
Evidence: E91, E92, E97, E98, E99, E102, E105.
J06
Russian hostile interference was the principal cause of the reported restriction.
Likelihood: Unlikely.
Confidence: Low.
Reasoning: Positive service-administration evidence, the reported repair/retained-cap distinction and instruction reporting fit the principal pattern better. Known EW, sabotage and orbital capabilities establish possible threats, not this cause. The low likelihood is a comparative judgement using positive rival evidence, not a deduction from the missing incident link; local interference or a mixed contribution is not excluded.
Evidence sufficiency: Enough for a cautious relative principal-cause judgement; insufficient to rule out selective interference or secondary effects.
Confidence basis: No technical logs, tested scope or authenticated hostile activity match are available.
What could change it: Matched interference evidence, provider confirmation excluding a service change, or refutation of the reported repair/cap sequence would materially raise this alternative.
Evidence: E07, E13, E85, E86, E87, E88, E89, E98, E99, E107.
J07
The restrictions imposed meaningful costs on at least some Ukrainian drone operations.
Likelihood: Likely.
Confidence: Moderate.
Reasoning: Attributed frontline/unit accounts and later named technical testimony describe constrained missions and capability. Meaningful local costs are a narrower claim than national paralysis, quantified losses or particular casualties. No casualty or battle outcome is attributed to the cap by this assessment.
Evidence sufficiency: Sufficient for a bounded some-operations effect; insufficient for total coverage, casualty estimates or counterfactual battle outcomes.
Confidence basis: Different roles provide a coherent reported effect picture, but officials are interested, some accounts are relayed/shared and operational records are absent.
What could change it: Unit corrections, representative mission data or evidence showing unaffected alternatives fully substituted for the reported missions would alter the assessed cost.
Evidence: E04, E05, E08, E09, E11, E12, E14, E19, E97, E101.
J08
Material high-mobility limitations remained unresolved in the latest public accounts at the cutoff.
Likelihood: Likely.
Confidence: Moderate.
Reasoning: The strongest restoration claims refer to recovering connections or limited movement capability, while later public accounts still describe restrictions. Different times, cohorts and functions can explain part of the apparent conflict. It is more reasonable to treat complete recovery as unestablished than to turn a restoration headline into a universal service result.
Evidence sufficiency: Sufficient for a public-information operational-status judgement; not a live service measurement.
Confidence basis: Later named/public accounts add useful temporal detail, but common official origins and missing per-terminal tests constrain confidence.
What could change it: A time- and cohort-specific notice plus independent operational confirmation of removed limits would supersede the judgement; this is frozen analysis, not continuous monitoring.