The Shape of Salvation History

the case for things getting better

Most religions give history a definite shape: first, things go bad; then, they get permanently better.

If we’re interested in which religion is most probable from the armchair (i.e., without God telling us, which is cheating), this claim is prima facie puzzling: if we are living through salvation history, and some times are better than other times, then salvation history must take some sort of shape. But why should we expect it to take this shape (worse → better forever), conditional on God existing?

After all, we can imagine history going from better to worse, or ping-ponging between good times and bad times forever. Does theism make the <worse → better forever> shape more likely? If not, then most religions have a target on their back: an auxiliary assumption that we’d have no reason to expect on theism.

In “The Shape of History”—which is not about religion, per se, but the axiology of history in general—Michal Masny asks whether the shape of history matters in itself. Consider two ways our history could go:

Improvement. Our history begins in the depths but has an upward trend: hardship in the early years, mixed fortune in the middle period, followed by flourishing at the end of our time.

Decline. Our history begins at the heights but has a downward trend: an early period of flourishing, mixed fortune in the middle period, followed by hardship at the end of our time.¹

I think if I did a poll, most people would report that Improvement seems better than Decline.

If you judge that Improvement is better than Decline (or, for that matter, that Decline seems better than Improvement), that judgement is puzzling on its face: after all, we can stipulate that for every good or bad moment in one history, the other history has it too. We can even suppose that both timelines contain the same people, and those people’s lives contain the same level of wellbeing and illbeing.

As Masny notes, on the view that Improvement trumps decline, “if a god were deciding which of these two worlds or divine plans to set in motion, they should opt for Improvement”.²

Assuming God is good and that his goodness should lead us to expect him to bring about good states of affairs (as most Bayesian arguments for God presuppose), then—prima facie—we should only expect God to bring about a history with an upward-slopping shape (rather than a downward-sloping history, or one that’s a random mix up of good and bad with no conspicuous trajectory) if Improvement really is the better shape. (A history where everything is just always good might be what theism should lead us to initially expect, but theism is a hypothesis about the world we live in, and our world is incompatible with such a history.)

An implausible theory

Can we justify the intuition that Improvement is better than Decline, all else equal?

One view is that time itself makes the difference. According to Michael Slote: “…what happens late in life is naturally and automatically invested with greater significance and weight in determining the goodness of lives.” Applied to the shape of history, rather than individual lives, you might think that what happens later in history is “naturally and automatically invested with greater significance and weight” than what happens earlier.

There are at least three intuition pumps that caution against this idea.

Problem #1 (due to Masny): imagine a world where, as Johann Frick puts it, “each generation of humans dies and vanishes without trace before the next one is born (perhaps, like mayflies, each generation of humans lays eggs before its death, but disappears before their offspring hatched).” In this world, each generation of mayflies is independent of one another: the actions of one generation has no bearing on the generation after it.

Now consider two shapes the history of this world might take: in one timeline, each generation is better off than the one before it, meaning the generations improve from worse → better. In the other, each generation is worse off than the one before it, meaning the generations decline from better → worse.

Bearing in mind that in the egg example, the generations are causally discontinuous, it’s a lot less intuitive—to me, on reflection—that Improvement is better than Decline. Since Slote’s account only cares about when things happen, and nothing more, it lacks the resources to explain why even if Improvement is generally better than Decline, it is no better in Frick’s case of the human eggs.

Problem #2 (due to Masny):

“[C]onsider a world much more like ours: socially integrated and with a rich array of traditions, languages, works of art, and scientific projects. Suppose, however, that these bonds and engagements remain stable over time, and the only thing that changes is the hedonic aspect of people’s lives. In the first course of history, early generations experience only a modest amount of pleasure, but each successive generation enjoys a bit more. To make things concrete, we could perhaps imagine that in the first course of history, fruits and vegetables that make up people’s diet are initially quite bland, but then become more delicious, so there is a discernible pattern of improvement. In the second course of history, the opposite happens, so there is a pattern of deterioration.”

If taste pleasure at later times is worth more than taste pleasure at earlier times, the course of history where the veggies get yummier is better than the course of history where they get blander.

But by Masny’s lights—and mine—this judgement seems incorrect.

Problem #3 (due to me): imagine a very short-lived universe, with a shelf life of 24 hours. Now consider two timelines: in one, an extremely happy Boltzmann Brain appears at the beginning of the day, and vanishes twelve hours in—simultaneously, a second, extremely sad Boltzmann Brain appears, and lives till the universe dies. In the other, an extremely sad Boltzmann Brain appears at the start of the universe, and vanishes 12 hours in—simultaneously, at the other end of the universe, an extremely happy Boltzmann Brain appears 12 hours in, and lives until the universe dies (which the Brain is completely fine with).

My judgement is that all else equal, both timelines are equally good. But Slote’s view, applied to the shape of history, predicts the opposite: the timeline where the happy Brain comes second should be better (which seems implausible).³

A better theory

Masny suggests a different theory, one that fares much better.

G.A. Cohen—the great Analytic Marxist—famously defended the view that there is a truth in conservatism: when a valuable thing exists—e.g., the Golden Gate Bridge—it is better to conserve it than to destroy it and replace it with something of equal value.

Cohen’s conservatism is liberal and undogmatic. If the replacement is drastically more valuable than what it’s replacing, we shouldn’t stomp our foot and conserve. Still, all else equal, it is better to conserve an already existing value than to destroy it and replace it with something of equal value.⁴

Applied to the shape of history, Cohen’s conservatism explains why Improvement is typically—though not necessarily—better than Decline. There are lots of valuable things about humanity: our projects, our achievements, our artworks, our strivings. (From a God’s eye point of view, there may be additional values—our struggle against Satan, our steps towards enlightenment, our progress towards the sort of theological knowledge that Swami Vivekanda thought was the “goal of all mankind”, and so on.)

On average, if history has a downward slope or ping-pongs from good to bad, fewer of these values will be conserved over time. As a result, by Cohen’s conservatism, Improvement is—on average—axiologically superior to Decline.

What’s neat about Cohen’s view is that, applied to the shape of history, it doesn’t end up proving too much: in the Frick’s example of the human eggs, the shape of history didn’t seem axiologically important. This is neatly predicted by Cohen’s view, since—in the egg example—nothing is conserved from generation to generation. It is only when conservatism is under threat—when things get too fucking woke—that the shape of history starts seeming relevant.

Likewise, in Masny’s case where the food changes in yumminess over time—but nothing else changes along with it—Cohen’s view predicts, intuitively, that either timeline is equally good, since nothing worth conserving—projects, relationships, aesthetic achievements, etc.—are any less likely to be conserved in either timeline.

Finally, in my case of the Boltzmann brains, Cohen’s conservatism predicts, plausibly, that neither timeline is more valuable than the other, since the brains are causally isolated from one another.

Why endorse Cohen’s conservatism? Well,

1. It gets intuitive results about the shape of history;

2. It seems right in individual cases, too. (E.g., Imagine we’ve preserved a gorgeous historical painting in an art gallery, and learned all we need to learn from it; if an artist could somehow wet the old paint and re-scramble it into a painting of equal—or even slightly greater—aesthetic value, it seems like there’d be something tragic about her doing so.)

Applied to the shape of salvation history, Cohen’s view predicts—to the extent that God is responsive to intrinsic value—that He would tend to ensure history slopes up instead of down, or instead of ping-ponging back and forth: after all, that approach would conserve more value, and God, if he is good, is conservative.

---

1. Masny, M. (2025). The Shape of History. Journal of Moral Philosophy, 1, 3.

2. Ibid., p. 4.

3. Masny also considers a marginally different account—due to Frances Kamm—that is similar enough to Slote’s that the bullets it must bite are the same; he then considers a different account—due to David Velleman—which he rejects on different grounds. Read Masny’s paper for a discussion of these views; I omit them here for the sake of brevity.

4. For a wonderful discussion of Cohen’s conservatism, see this piece on Substack by philosopher

Richard Pettigrew

.