Keynes was talking of a mostly closed economy with wages set by something approaching a market mechanism. This made sense as, well, that's what was around at the time.
By mostly closed I don't mean that there was no trading at all -there was, indeed, much less than there had been before world war one, but still quite a bit. But crucial economic inputs were typically available locally, or had substitutes, or several sources.

In other words, Keynes had little reason to try and think of a society that had been entirely built on the basis that it would always get plentiful cheap oil even though it did not produce any. It had no room for the providers of a key input to suddenly wonder why on earth they should keep selling their depleting reserves at extraction cost now they had considerable market power. In other words, it had no room for something like an oil shock, because it was irrelevant to the situation the model was trying to describe. The mechanics of inflation transmission in Keynes's work (or in its "bastardisation by Samuelson", if you must) were different and indeed stagflation following a couple of oil shocks was not a natural conclusion -simply because the possibility of such a shock had not been included in the model.

It's more than a bit disingenuous to fault a model for failing to predict a situation it was not designed to assess. I don't read expiry dates to decide how long to cook something in the oven. Who knows what Keynes would have said about the economics of external shocks? And how could the events following an external shock refute his thoughts on an internal depression? Still, it gets worse.

Apparently, the suggestion that inflation could be maintained following an external shock, in a sort of self-sustaining phenomenon, struck economists as an almost unbelievable result, impressive enough that we should give a major benefit of the doubt to any other idea coming from Chicago. Inflation expectations ruled the world, and extremely strongly weighed on inflation.
Now, I don't want to imply that any effect of inflation expectation is an illusion necessarily -I understand the phenomenon of price changes being done infrequently, and therefore trying to incorporate some of the future inflation (although I think that idea can be overplayed). However, we don't need to look as far as inflation expectations to explain a self-fulfilling phenomenon: back then, many (and I do mean many) employment contracts contained a clause guaranteeing that they would at least rise in line with inflation. Pay rises were in real, not nominal terms. So, give the economy an external shock on something as big as oil, and you would see wages rise quite a bit, even as the economy slows down.

But wait, some (not, let's be fair, all) economists will say. That does not lead to an inflation spiral, because the prices of final goods will remained determined by the laws of supply and demand.
At this point, if you happen to live on Earth rather than model-land, you should probably just laugh. Whatever can be construed as an extremely long-run equilibrium, most prices are not fixed by a constant supply and demand market clearing. Companies charge cost plus. While it may be true that an individual company may have to absorb a cost shock it alone experiences (but even that is far from always. I remember a hairdresser explaining that he had raised his prices because his son had started attending a private school. 20 years later, he's still in business), remember that the oil prices affected everyone.
And most of the inflation indexing of wages were sectorial agreements. All companies would have had the same impact (that is, both the impact from oil itself, and then the impact from wages rising with inflation, year after year). All companies would have passed it onto the customer, with no impact on their market share.

And that's not even a huge insight: companies tell you they do that! Price rises are typically explained by costs having increased. Yes, they may not be able to do that in some simplistic market models, but on the planet we happen to share, that's how things go.
Of course, there are a few markets where companies are absolute price takers. Typically, grain. Well, oil is a major input for them so, if the price is too low, they simply won't produce. Guess what happens to the price of grain when there is a shortage? There are few really good substitute for food over the medium term.

So guess what: if you increase the price of a component, you get inflation. And if you have a system that increases the price of the main component (labour is an even bigger part of the price than oil, would you believe it?) in line with inflation, then you get a spiral. This seems somewhat underwhelming as a paradigm-changing epiphany.

Still, it clearly has not sunk in, even 4 decades later. As I mentioned earlier, you get suggestions that the UK inflation of over 2% proves that more stimulus would have been in vain (yes, I know, suggesting that the UK was at full employment just because its inflation was not lower than 2%, and not collapsing, strikes me as using the wrong indicator when you have direct ones showing that there were about 10 people looking for a job for every job opening). But look at the news reports and you'll find that, just before, the UK government had increased VAT twice!
Guess what? the VAT hikes showed up in inflation figures. And when they had been absorbed, inflation started falling. But you see, in a pure form clearing market, this is not supposed to be possible. So I guess it must not be happening...

Models are useful. I will keep using them and defend their sensible use. But constant reliance of a model, beyond its scope, with no understanding of what it really says is a recipe for disaster. Refusing to inform your analysis with context is not objectivity, it's hermeticism.
Bayes would have wept.

(cross-posted on my blog Anachronicles) (edit: changed the clumsy title)