It would be better than anything else we have at the moment

It really isn't? There is no specificity, it also doesn't state that earthquakes _only_ happen after solar activity. So this is equivalent to saying "There could be an earthquake somewhere at any time in the future".

With distributed ray detectors and suitable modeling of inner earth processes, assuming that the premise is correct - seems it may have potential to work?

Ie, maybe able to generate the level of specificity required.

EDIT: Also there aren't that many places on earth at high risk of earthquake that also have poor construction, etc. Meaning any advance warning, that a significant quake may hit somewhere, can trigger "battening down the hatches".

You'd have to get a theory of why the rays correlate to earthquakes to have a usable model.

Realistically, if you live in a place with poor construction and relatively large earthquake risk, the sensible thing is to always have your emergency kit ready to go. Chances are, you have other infrastructure issues anyway, so keeping a week or so of emergency rations and water available might come in handy more often than somewhere that has tigher building codes (and enforcement). It would likely be hard for everyone to prep within the same two week window anyway. But, if it was quite specific and accurate, relief organizations could begin staging and start traveling to be closer and quicker to respond.

Doesn't seem like it's anywhere close to that from descriptions in this thread.

Personally, I prefer waiting for all of the animals to seek higher ground as my seismic activity indicator.

Seems figuring out what it is they're doing and duplicating it would be a straightforward thing to do. Surely there's a research group working on it?

From https://en.wikipedia.org/wiki/List_of_earthquakes_2021%E2%80... there were 44 earthquakes with a magnitude >= 7.0 in the last 2.5 years. That is one every 17 days. The GP is almost correct: There is a high chance [1] of an earthquake somewhere in the world, including oceans, in 17 days.

[1] If you want to be fancy with probabilities, 64% of at least one earthquake. In some 17 days periods you will get no earthquake and in others you will get more than one earthquake. In average you will get one.

Too late to edit: The correct number is 21 instead of 17.

The 64% is still 64% because if N is big enough it almost doesn't depend if you consider something with a probability of 1/17 in 17 days or something with a probability of 1/21 in 21 days.

If the increase in cosmic ray detections is due to some change in Earth's magnetic field, presumably where and how the magnetic field changes would in some way correlate with the activity.

If the magnitude of the quake can be estimated from it, it can be somewhat useful for response agencies in earthquake and tsunami prone areas to make sure their ducks are in a row. Kind of like how storm season tends to make people make sure their emergency supplies are in order, even if their area doesn't tend to get hit by big storms.