Michael Kofman

I now have the data to reevaluate my previous estimates, so why not? Until 2024, Mediazona and the leaked Russian death register match almost month by month. From 2025, Mediazona starts falling behind. By late 2025, it has recorded only about a third of the deaths already in the register. By the end of 2025, the leaked register itself also starts falling behind, as the confirmation lag catches up with it. The names aren't necessarily missing. They're just slow to appear. So I revised my GSUA ratios to fit the register: from 1:4.5 in 2022–23 down to 1:2 since February 2025. On the chart, the black line is GSUA adjusted by those ratios, and the dashed line is the register. They track each other closely until the register's recent months start filling in. At these ratios, Mediazona's list should eventually reach around 607,000 names. There is still room for refinement. Early Mediazona figures exceed the projected values, while SBS statistics suggest the KIA ratio may fall further in 2026. So I expect the final estimate to land somewhere between 580,000 and 620,000 dead. Even if the leaked dataset is fake, it's still useful to know what the other side estimates. You are free to disagree with those numbers, but they're worth examining.

🇮🇷🇷🇺 Despite Washington's recent assertions that Moscow is not backing Tehran, confidential Russian defence records tell a different story. Dallas Analytics publishes primary industrial documents proving conclusively what the export code "K-10" designates: Iran. 🕵️ State plants built specialised missile-launch hardware – arming the forces now strangling global energy transit in the Strait of Hormuz – alongside a wider pipeline spanning Yak-130 trainers and helicopter upgrades. Read our investigation and examine the original schedules: dallas-analytics.com/russia-…