Putin Watch
What it measures. Observable pressure on Vladimir Putin’s regime: elite purges and reshuffles, economic stress, war-front losses and deep strikes on Russian territory, diplomatic isolation, domestic dissent, succession signals, and search demand for exit terms inside Russia.
How it’s scored. Seven inputs, each rated 0 to 10 daily by a human editor from dated, linked sources. The weighted average is multiplied by 10 for a 0 to 100 index; higher = more stress on the Kremlin.
What it is not. Not a prediction of any specific event: not a coup, not a collapse, not a market. A high PWI does not mean regime change is imminent; it is a public read on pressure. Full methodology →
Why we track this
Authoritarian regimes rarely announce their own weakness. They announce strength on the day they are weakest, and the record of what actually happened is assembled afterwards, usually by people outside the country. The Putin Watch Index is an attempt to keep that record in public and in real time instead: to write down, each day, what is observably true about the pressure on the Kremlin, with a date and a link attached to every judgement, so the assessment can be argued with rather than merely believed.
The index exists because the coverage around it tends to swing between two errors. One is breathless collapse-watching, where a single purge or a bad week on the front is read as the beginning of the end. The other is flat denial, where a regime is treated as permanent because it has not fallen yet. Both are unfalsifiable. A number that moves on stated inputs, published daily, is falsifiable: if the score is wrong, the inputs that produced it are listed, and the sources under them are dated and linked.
What that means in practice is that this page is deliberately boring on most days. Pressure on an entrenched regime is not a story that breaks; it accumulates, and most of the accumulation is unremarkable. Elite reshuffles that look decisive in isolation are routine. Sanctions bite on a lag measured in quarters. Battlefield losses matter less for territory than for what they force the state to say at home. The index is built to register that slow accumulation rather than to react to whichever development is loudest in a given news cycle.
On the human scoring. Every one of the seven inputs is rated by an editor, not by a model or a keyword count. That is a deliberate trade. An automated score would be more consistent and less honest: the underlying signals are things like the tone of a state television segment or the significance of a particular general's reassignment, and those do not survive being counted. A human score is contestable, which is the point — the methodology page sets out the weights, the scoring rules, and the criteria for each input, so a reader who disagrees can see precisely where the disagreement lies and check the sources for themselves.
How to read a move. Single-day changes are usually noise. The index is most useful as a trend across weeks, and most informative when a component diverges from the others — when, for example, elite-signal stress climbs while economic stress is flat. Those divergences are what the component breakdown below is for. A high reading is not a prediction that anything in particular will happen, and this page will never make one; it is a statement about observable strain, and nothing more.