It's years now (15?) since I was reading valuations of BRK on fool.com and understood that book value was gradually getting worse and worse as an indicator...

I think a more delicate wording would be this:
As time goes on, there is less and less reason to believe that book per share should remain a meaningful yardstick.
This remains a true observation.

However, more or less entirely by coincidence, so far it's about as good a yardstick as ever.
Even if you use a very fancy valuation method, you still get a number that rises at around the same rate and by the same overall amount.
(certainly within the error bars of any such valuation method, anyway)

This is because the big moving parts that drive down fair P/B, and the ones that drive up fair P/B, have roughly balanced.
There is no reason to assume that this balance will continue. But so far, all the fancy valuation work I've done has been a waste of time.
Except to the extent that, had I not done it, I wouldn't KNOW it had been a waste of time.

The main thing about book as a metric: I think it's fair to assume that future dips in book, like past dips in book, will all be transient.
By extension, when book drops a bit it's most sensible to assume that value hasn't dropped.
I can't think of any plausible way to come to the conclusion that a share isn't in reality worth *more* than it was a couple/few quarters ago.
Cash stacks up at around a billion per fortnight, for one thing.
We certainly haven't had any permanent value impairments of that magnitude.

Jim