Kostiantynivka is being ground down by slow infiltration and is at real risk of
falling by the end of summer 2026 โ but its probable loss is occurring INSIDE a
broader Russian OPERATIONAL FAILURE, not a Russian breakthrough. The tactical
vector (locally negative for Ukraine) and the operational vector (theater-wide
negative for Russia) point in OPPOSITE directions at the same time. That
divergence is the central fact; "the city may fall" must not be read as "the
front is collapsing." (Confidence: HIGH on the divergence; MODERATE on city loss.)
KEY JUDGMENTS
1. The city is CONTESTED, not taken. Russian control ~13% on 10 Jun, up from <=5%
in Dec 2025 โ roughly eight points in six months. The 11-12 Jun "flags in the
western city / brigades withdrawing" claim is milblogger-sourced and sits
against ISW's geolocated negatives (railway station held, Dovha Balka cleared,
Ukrainian local advances). Weight the geolocation over the footage. (MODERATE)
2. Russian operational tempo is at a multi-year low. All-front gain in May was
14 km2 โ lowest in three years and the first net-negative month since 2023
(March was 133 km2). Effort is concentrating as yield collapses. (HIGH)
3. Capture does not equal breakthrough. Pokrovsk Myrnohrad fell early 2026 yet
produced no exploitation westward since Dec 2025. The strongest evidence
against an automatic Fortress-Belt cascade. (HIGH)
4. Mass without momentum. ~18 brigades 8 divisions are committed to the belt;
the Slovyansk opening failed, Chasiv Yar is uncleared (blocking northern
envelopment), and the Siversk-Kramatorsk axis is stalled. (MODERATE-HIGH)
5. Command is optimizing politically, not militarily. Two infeasible deadlines
set and missed (all-Donetsk by 1 Apr; the city by May). (HIGH)
ANALYTIC READ
[Systems] The belt is one networked logistics organism (Kramatorsk-Druzhkivka-
Kostiantynivka-Sloviansk). Russia is not storming it; it is fragmenting it โ
seizing peripheral nodes (Berestok, Illinivka, Novodmytrivka) to sever intra-city
movement and render positions untenable. The dominant feedback loop is the
drone-saturated airspace: whoever wins FPV/logistics interdiction owns the tempo,
and Russia's own resupply now runs through the ruins of cities it captured โ a
negative loop on its sustainment. [Game theory] Russia is paying maximal cost
(heavy losses, ~80% unit-replenishment cycles) for minimal territorial yield and
continuing anyway; the deadlines and the "whole of Donetsk" demand in talks are
costly signals anchored to aspiration, not to the ~20% of Ukraine actually held
or to a belt that remains unbreached. Urban attrition favors the defender's payoff.
[Information theory] The leading edge is a low-SNR environment: curated capture
clips are high-volume/low-information; multi-day geolocated control deltas are the
high-information signal. Bayesian discipline says update on encirclement geometry
(Chasiv Yar, the in-city river line, the ~2 km pincer gap), not on flags.
[Intel/counter-intel] Flag-raising is textbook denial-and-deception โ manufacture
the perception of control to induce premature withdrawal or Western fatigue.
Source motive ranks the evidence: ISW/DeepState are least motive-conflicted (top
weight); Ukrainian brigades and Russian milbloggers each carry information-
operations incentives (discount accordingly).