I wonder if what proportion of Russia's nukes actually work...
Probably a fair question to ask of the arsenals of any of the 'original' five - it
could be possible that any (or all) of them are not even close to 100% reliable.
For example, US stockpile stewards and expert observers have certainly expressed concern about the US nuclear stockpile ageing, raising question marks concerning reliability over time. The absence of full-scale and more extensive testing could miss potential common-mode failures (complex equipment doesn't tend to age gracefully and the fissile components, in particular, can degrade in a number of weird and wonderful ways leading to unexpected or unintended effects when the pin is pulled).
The US/UK/France have (according to official statements) adhered to a strict "zero yield" hydrodynamic/subcritical testing regime within the spirit of the CTBT (essentially yields of grammes of TNT equivalent), whereas the Russians have indicated that they consider the treaty as permitting yields up to one tonne (they have reportedly declared hydronuclear test yields of up to a few hundred kg). This interpretation certainly provides for scope in verifying confidence at near-full-scale in low-yield weapons (where the Russians
might have an advantage* and could easily have continued covert testing**), facilitating development of novel designs in the sub-kiloton class with (eg) high fusion-fraction yield, low contamination and specialised outputs for addressing particular scenarios (for example: hardened shelter penetration, biological effects, electronics suppression). Those same low-yield devices also potentially serving a key role in the Russian escalate-to-deescalate doctrine, perhaps providing them with a means to draw a line in the sand and leave the strategic escalation ball in their opponent's court.
That is, of course, mainly focussed on the physics package/warhead. The reliability of the delivery mechanism is an entirely separate factor. Likely their rocketry is still largely up to spec (the nuclear forces are a whole separate division again from the rest of their military, with their own budget). The latest Sarmat, SS-27/R-S28, test suggests they can still [potentially] deliver the necessary engineering (somewhat ironically that is of course largely a wholly home-produced one-for-one intended replacement of the in-service Satan, SS-18/R-36M, delivery system, the missile, motors and control systems of which were designed and produced at Yuzhmash in Dnipro, from where they are obviously no longer getting any of the spare parts that they've ordered). That is 6+ years behind schedule and still awaiting entry to service. Then there is the state of the signalling/comms infrastructure to consider...
* Arising from the Soviet's Programme 7 which was far more extensive than Plowshare.
** Any fairly competent nuclear power, certainly one with many decades of testing experience, could perhaps test up to 1kt whilst minimising risk of detection. That would suffice to validate criticality, safety, fissile materials behaviour, pure fission designs, fusion effects and boosting, all with greatly reduced uncertainties (insufficient for tests of multi-stage thermonuclear designs).
www.wsj.com
www.theguardian.com
