Table 1.3, go to the third row, which has “L ≻ G ≻ H” as player 1’s preference. The only outcomes are L and G:
| 1 \ 2 | H ≻ L ≻ G | H ≻ G ≻ L | L ≻ G ≻ H | L ≻ H ≻ G | G ≻ L ≻ H | G ≻ H ≻ L |
| L ≻ G ≻ H | L | G | L | L | G | G |
Which one it is depends entirely on whether player 2 prefers L ≻ G or G ≻ L, regardless of how they rank H.
The big question is why the player preferring L to H might rank L ≻ G ≻ H. If this player ranks L ≻ H strictly because of costs of cutting CO2 emissions, L ≻ G ≻ H will be a very real possibility. G, after all, is cheap. We are immediately back to the moratorium, assuming the world doesn’t want G to win it all. Ban it, and hope to guide climate policy in a productive direction – toward H, that is.
If that player, however, ranks L ≻ H because they do not believe climate change is a problem worth addressing with aggressive action, L ≻ G ≻ H will be less likely. Why risk G if climate change isn’t all that bad to begin with?
Now we are in the third scenario: L ≻ H ≻ G. Zoom into the fourth row of Table 1.3 to see where this might lead:
| 1 \ 2 | H ≻ L ≻ G | H ≻ G ≻ L | L ≻ G ≻ H | L ≻ H ≻ G | G ≻ L ≻ H | G ≻ H ≻ L |
| L ≻ H ≻ G | L | H | L | L | G | G |
The most frequent outcomes are still L and G. If the other player ranks G on top, G wins. Not if, but when. What’s striking, then, is when G does not win. That seemingly goes counter to the “not if, but when” logic.
Let’s simplify the table a bit more to see this logic. We can drop the two columns where G is ranked first, and compare the first four columns for when L ≻ H ≻ G (row four of Table 1.3) to the ones when L ≻ G ≻ H (row three):
| 1 \ 2 | H ≻ L ≻ G | H ≻ G ≻ L | L ≻ G ≻ H | L ≻ H ≻ G |
| L ≻ G ≻ H | L | G | L | L |
| L ≻ H ≻ G | L | H | L | L |
If both players rank L on top, L wins. G doesn’t add much to this calculus. Let’s drop two more columns, to compare players ranking L first to those ranking H first. Now we’re left with exactly four cases:
| 1 \ 2 | H ≻ L ≻ G | H ≻ G ≻ L |
| L ≻ G ≻ H | L | G |
| L ≻ H ≻ G | L | H |
The first column has two cases leading to L as the outcome. That’s when the player ranking H ≻ L also ranks G last. The game essentially collapses to the prisoner’s dilemma of yore. G doesn’t influence the decision. L wins.
Almost there. We’re left with two cases.
With G wedged between L and H for both players, G wins. In some sense, the logic here is simply that the two players can’t agree on how much CO2 to cut, so they would rather settle on G than give the other player what they want in terms of CO2 cuts. That’s a disheartening solution. It’s also the one that calls for strong solar geoengineering governance. But it’s not the only solution.
If G is ranked below H for those preferring L to H, suddenly, H emerges as the winner. That’s true, even though one player still ranks L ≻ H. Here the “availability of risky [solar] geoengineering can make an ambitious climate mitigation agreement more likely.” That, in fact, is the title of the paper I wrote with then-Ph.D. student Adrien Fabre, arguing just that.36 The title of that paper is worth restating: it’s the mere availability of solar geoengineering that leads to this outcome. Another key word: “risky.” In fact, the riskier is solar geoengineering, the more likely is this outcome.
That mere availability helps break the prisoner’s dilemma, the free-rider problem. It isn’t a guarantee. But the mere possibility is worth pointing out: If G ranks just below H for either player, G might indeed help induce H. That’s true even though one player still ranks L ≻ H. Assuming G is not just fast and cheap but also highly imperfect – even those ranking L ≻ H still prefer H to G, putting it last – the mere availability of G might prompt otherwise quarreling parties to opt for H.
All of that is true despite our setup that rigged things against H in the first place. Recall how the weakest-link game setup in Table 1.1 all but guaranteed that L would win.
Enter G, and L is no longer a given. The most likely case with G as an option might still be for G to take all: Somebody, somewhere, will opt to use G, and it will dominate the final outcome. All of that seems to put the burden squarely on governments to rein in tendencies to do too much, too soon – in less-than-ideal ways. (Part III will explore the urgent need for governance in more detail.)
Meanwhile,
