by Henry Farrell on April 6, 2020

Attention conservation notice: below the fold is a lengthy and spoiler-filled response to William Gibson’s new book Agency. Probably best not to read unless you’ve already finished Agency, or have no intention of reading it and want to get some sense as to what the book is about. In either case, you’re likely better off reading Mike Harrison’s Guardian review, which covers much the same ground as below, but with more subtlety and fewer spoilers.

Agency is a sequel to Gibson’s The Peripheral, but if anything is even more pessimistic: more black and less comedy. The worlds it depicts are based, I think, on an organizing social-political-economic thesis. I don’t want to overclaim here. Agency, like any good sf book, is a novel, with negative capabilities. I don’t know whether Gibson himself believes in the thesis, or is pushing as far as it can go, to see how it works as a premiss for fiction. Your guess is as good as mine or better. However, I’m going to read it as a social scientist, trying to figure out what the thesis is and what its political implications are. That will highlight some aspects of the novel (which it may or may not understand correctly), while downplaying others. Equally, it may or may not help clarify the relationship between the world that we live in, and the world that Gibson invents.

The implicit thesis as I see it, is as follows. First, that the relationship between the natural world and human activity is producing an incredibly nasty set of self-and-mutually reinforcing positive feedback loops. Second, that democracy and multilateralist internationalism are completely incapable of damping these loops down, and indeed are likely to collapse themselves, as their internal flaws are exposed by brutal external circumstances. Third, that the best we can reasonably hope for is a quasi-benign soft autocracy, optimized for survival rather than for preserving (possibly illusory) liberal and leftwing values of human agency. The more plausible outcome, however, is ruthless oligarchy of a kind that is itself going to be unstable in the long run, leading to further crisis and disaster.

The first of these is the background to what Gibson describes in both books as the “jackpot,” a large scale human die-off. Notably, the jackpot isn’t a single event – it is a series of mutually reinforcing catastrophes that start in the very near future. As I read it, the implied future involves all of the shittier aspects of globalized capitalism feeding off each other and making each other worse. Thomas Friedman and others spun out elaborate fantasies of globalized markets where consumers got what they wanted, businesses competed with each other across a world smeared into a flat reflective plane by the invisible hand, and geopolitics evanesced into market-subordinated states and neoliberal awesome. Gibson instead presents a globalization where all the purported externalities, the unfortunate side effects, man-made Sargassoes of garbage, urban sinkpools of destitution and imposed squalor, plagues and industrial poisons feed on each other to create a complex self-ramifying system of their own, which is more or less inimical to human survival. He doesn’t dwell on the precise ways in which they reinforce each other. They aren’t particularly difficult to extrapolate.

This system not only is bad in itself, but reinforces the worst aspects of politics. The current global complex of multilateralism has a bright public aspect and a very dark shadow. The bright side is a variety of (mostly non-binding) efforts to cooperate to solve global problems. Here, examples might include the World Health Organization’s efforts to eradicate diseases (not so much state-to-state cooperation as the things that happen in the interstices such as the “Flu Network”). The dark side is the complex of convenient arrangements, technical agreements, and areas of induced non-agreement that reinforce problems rather than solving them. The shift of the investor-state dispute resolution system into a series of instruments for restraining governments at the behest of business. The tax arrangements identified by Saez and Zucman that allow rich individuals and corporations to stash their money. The transformation of the global refugee regime into a system of holding camps for people without rights.

Already, we can see how this may undermine the political accountability that liberal democracy is supposed to be founded on. The international freedom to move money was justified (as Quinn Slobodian’s book The Globalists describes) as a protection against the power of states to seize the property of the vulnerable. This was a plausible argument in a world that remembered how Nazi Germany had dispossessed Jews as well as murdering them en masse. Yet it also, as Gulnaz Sharafutdinova and Karen Dawisha have shown, may make it less likely that states will develop liberal institutions in the future. In the old dispensation, those who made a lot of money through fair means or foul, had self interested reasons to foster the rule of law in their home countries, since they didn’t want their wealth confiscated from them by arbitrary government power. Billionaire self interest had some general benefits. Now, however, oligarchs don’t have nearly the same need to develop independent institutions in their home jurisdictions, since they can just export their wealth to London or other havens of oligarchy, where it will be perfectly secure. They just don’t have the same need to strike domestic bargains with the less powerful.

Gibson’s future world is one in which these tendencies have reached their logical conclusion. The prettier facade of multilateralism has unraveled completely, while the malign aspects of globalization have come to dominate. The “klept” – the international kleptocracy – has joined hands with the ancien regime of the City of London, whose guilds and archaic practices do not seem particularly quaint when they have consequences. Both are already conjoined, and doing well in our present: Gibson’s novels suggest that they will do even better, and become more closely entwined, in the future

After the jackpot, everyone lives on the sufferance of the very rich and the very corrupt (there is no practical distinction between the two). Again, we aren’t told exactly how this has happened, but in a world of right populism and coronavirus, we can make some very good guesses. As global problems become ever worse, people’s instinct is not to cooperate to solve them, but to find strongmen who can protect you against the other strongmen.

Science fiction writers aren’t in the game of predicting the future, but of identifying tendencies, and turning them into engines of plot. Still, Gibson’s logic seems grimly similar to our current world where states are fighting each other over medical masks and vaccines, and where the intricate value chains that globalization depends on are being torn apart and rearranged as states battle over who gets access to which supplies, hoarding what few masks and pharmaceuticals they have for themselves.

In this world, democracy too has completely unraveled. Gibson suggests that in part this is because it is readily manipulated (more on this later). The only politics that remains possible is ruthless oligarchy – but ruthless oligarchy is itself unstable. As Lowbeer, the enforcer who keeps things ticking over, puts it when she is asked if everything would collapse if the oligarchs were left to their own devices:

But for the occasional pruning,” she said, “under the auspices of an impartial eye, yes. Their tedious ambition and contempt for rule of law would bring everything down, around their ears and ours. They managed to do that with the previous world order, after all, though then it was effectively their goal. They welcomed the jackpot, the chaos it brought. The results of our species’ insults to nature did much of their work for them. No brakes magically appeared then, and I don’t see them appearing now, absent someone free to act, with sufficient agency, against their worst impulses.

The outcome is not rule of law, but a system where Lowbeer and the oligarchs hold each other in check, providing a limited degree of dynamic stability that prevents the system from again spinning out of control. The oligarchs recognize (collectively) that they need an outsider like Lowbeer to prevent their individual selfishness from leading to disaster. Lowbeer, in turn recognizes that she will only be suffered to stay alive and powerful so long as she only uses her autonomy to maintain the stability of the system as a whole, rather than trying to make it better. She has “agency,” but only “sufficient agency” to do what she is required to do. Her role is a little like that of the podesta who enforced order among the feuding clans in early modern Genoa – an official empowered to keep faction in check, but no more.

Gibson hints heavily that this temporarily stable system is on the verge of going bad. Lowbeer is the last of the enforcers, and is very old. When she dies, there is unlikely to be any replacement (in Genoa too, the podesteria faded away and the clans went back to war). She is able to intervene in the “stubs” – the alternative pasts that drive the plot of both Gibson’s novels – to prevent them from going bad. However, she can only do this because the oligarchs consider them as toys. If the oligarchs thought of her hobby as a threat to them (and one does try to make that case before she squashes him), she’d be in trouble. Because the stubs appear to be economically useless, they are not worth bothering with (there is some tension between “Agency”‘s depiction of the stubs, and “The Peripheral,” where a war-torn stub comes up with ingeniously nasty weapons systems, which can be usefully deployed in the root reality).

What all this amounts to is a world, or worlds, where liberal notions of agency are an illusion. No-one is really in charge of their fate, and those who think that they are become the butt of grim comedy. Lowbeer has some situational agency, but only in narrowly defined areas, where others think that her actions don’t really matter. The focus of “The Peripheral” was the relationship between the powerful and the powerless. The focus of “Agency” is on how no-one seems to have very much power at all, and what capacity for agency there is is leaking over to non-human processes.

Specifically, in both books, Lowbeer is aided in her efforts to regulate the “uncles” of the kleptocracy by the “aunties,” a body of algorithmic processes that are capable both of monitoring vast amounts of information and of predicting, with some accuracy, the likely perturbations that would be caused by specific interventions. These are distant intellectual descendants of the self-aware surveillance state that emerges in the last pages of Gibson and Sterling’s The Difference Engine. The aunties play a more visible and important role in Agency than in The Peripheral, because they are grown up versions of a less developed machine intelligence, “Eunice,” which comes together at the beginning of Agency, is scattered into a series of autonomous processes in the early pages, and comes together at the end. Eunice (or the “branch plant” daughter-processes she gives rise to) are quite literally plot-engines. They manipulate everyone, especially including the main character Verity, into doing what needs to be done to stave off Armageddon. Verity rapidly becomes dependent on Eunice telling her what to do: “She caught herself waiting for Eunice’s instructive pictograph hands to appear.”

In Gibson’s world, staving off Armageddon is the best that we can reasonably hope for. Democracy is fragile, and open to manipulation. Agency suggests that the reason why Brexit happened and Trump won is because Russia manipulated social media. In another future stub (the one described in “The Peripheral”), democracy survives, sort of, but only thanks to Lowbeer and the aunties’ abilities to predict and manipulate voters into choosing the ‘right’ president. Eunice too is a manipulator, but like Lowbeer a probably benign one. Both of them are optimized to create what stability is possible in “competitive control areas,” or high risk environments. The book ends with Eunice revealing herself to the world. There’s the barest hint that there might be hope of something different emerging – she is a genuinely new experiment in a history that has been wound back to just before it becomes irredeemably awful – but it’s only only a hint, and perhaps not even that.

It’s a grim thesis for two reasons. First, because it suggests that the kinds of societies in which we think we might have agency depend on a contingent and self-undermining set of conditions, a moment when it seems that humans have the power to act upon the world, just before the world begins to act upon them. Second, because even within those conditions, agency might be an illusion. Real autonomy isn’t possible in a world where machine intelligences can nudge us to do things that we don’t realize we are doing. If The Peripheral’s world is a refracted version of Piketty, Agency’s world is the one of all those critiques of machine learning (the two worlds are aspects of each other, but aspects that accentuate different things).

Again, I don’t know whether Gibson actually believes we live in a world that is something like the one depicted in Agency, but that’s not particularly germane to the question of Agency’s relevance. There are obvious ways in which our world resembles Gibson’s – and may be coming to resemble it further. The logic of the Jackpot – interlocking catastrophes that stress our system past its breaking point, seems far more real now than when Gibson’s book was published, a couple of months ago. As Riana Pfefferkorn notes, the novel’s alternative version of the present is described as a moment “the drivers for the jackpot are still in place … They’re still a bit in advance of the pandemics at least.” Ouch.

As pandemic intersects with crashing economic systems, with environmental catastrophes caused by global warming, and with the “existential politics” of climate change, things are going to become less predictable. Fareed Zakaria, an establishment foreign policy thinker if there ever was one, warned a few days ago that “We are in the early stages of what is going to become a series of cascading crises, reverberating throughout the world,” and called for more global coordination to stop it. It’s safe to say that global coordination has not been the instinctive response of the great powers to the current situation.

Yet there are also ways in which the organizing thesis exaggerates the dire situation we’re in. It’s fair game for a science fiction novel to rest on the assumption that Russian bots manipulated us into a world where Trump became president. Yet there’s less supporting evidence for that claim in the real world than many people suppose. The problems of American politics in 2016 were mostly the problems of American politics in 2016. Most political scientists think that Russian influence efforts didn’t do very much to change things.

More generally, the world of Agency is one where people can be reasonably reliably manipulated into doing what others (whether malign actors, machine learning processes, nascent artificial intelligences, or mixtures of the above) want (or sometimes “want”) us to do. Shoshana Zuboff recently wrote an article suggesting that this is true today; that we are puppets of Facebook and Google’s algorithms. Yet again, the empirical evidence for these claims is scant. A recent survey article finds null effects for the vast and expensive efforts politicians run to change voters’ preferences about who to vote for. There is a whole lot of bullshit and snake oil about the power of algorithmic messaging to manipulate voters, but there isn’t any evidence that it’s actually effective. Cambridge Analytica seems to have run on the statistical equivalent of woo and pixie-dust. The wizards of Pung’s Corners aren’t any more effective than they were in the 1950s.

This is because human beings plausibly aren’t nearly as gullible as they are made out to be. Hugo Mercier’s new book presents a lot of very good evidence that human beings are more stubborn and better able to detect falsehoods than the commonly accepted myths would suggest. Lots of people subscribe – in a sense – to Q-Anon and to bullshit about pedophile rings operating under DC pizza parlors. But only one or two genuinely believe, so that they are prepared to go into Comet Pizza with a gun, or shoot a mafioso to help Q prevail.

The lesson to be drawn for this, for good and for bad is that the combination of social manipulation and machine learning is much less efficacious than it is made out to be. “Nudge politics” is less likely to result in people being manipulated for the general good or the general bad than Dominic Cumming’s Covid 19 clusterfuck, where over-optimistic social science and modeling help lead politicians to make terrible policy choices. Perhaps, with awesome new forms of general AI, we would be in a different place. But we don’t have those forms of AI, and can’t expect them anytime soon.

This isn’t to say that democracy may not be in trouble. But the trouble that it is in is arguably less the product of successful manipulation, than of the pervasive fear that it is being successfully manipulated, together with a variety of other more mundane threats (including the ferocious inequality that Gibson describes in the Peripheral). The frailty of democracy is that it is held together by broad social beliefs that when the other side loses the election, it will accept its loss and go into opposition. When those beliefs unravel, they may unravel quickly, as a side that fears the intentions of its opponents gets in its own retaliation first. Those are very serious problems – but they are less fundamental problems than the problems suggested by the thesis of Agency, which could be read to imply that democracy and liberalism are aberrations, and indeed errors that are liable to be corrected when history resumes its ordinary bloody course.

History is resuming that course, at least for a period, so that either way, we are likely to find out. One of the things we will discover over the next several months is how successful the manipulation of democracy can be in a world where people are dying around you. The incompetence and moral depravity of Trump, of Bolsonaro, of Orban and of other leaders is going to result in many deaths. Can they successfully persuade a majority of people that those deaths aren’t happening, or that they are really someone else’s fault? If so, we are all in the kind of trouble that Agency depicts.

If not, we are surely still in trouble. We are going to have to confront a series of mutually reinforcing crises, in a global system where states may be moving to a politics of the jungle, each fending for itself. But we may have some agency to confront these problems and to mitigate them. It might or might not be enough.



David 04.08.20 at 3:14 pm

I can’t believe no-one has responded yet. So let me kick things off by saying that the feature of The Peripheral that struck me most was the sense that the Jackpot had come about basically through carelessness – that the people who might have been in a position to prevent it fundamentally couldn’t be bothered to do that.

My second thought was that lack of agency makes for a challenging thesis but a dull novel. Netherton at one point in the new book finds himself missing Flynne – not as much as the reader does, Wilf, believe me …


Barry 04.08.20 at 3:18 pm

Henry, I just have this overwhelming feeling that Gibson’s work has been so overtaken by events that it’s archaic.

I’ve tried to watch a few 90’s shows that I never watched, and it feels like an old movie from another world (or like an Edwardian play watched after WWI).


oldster 04.09.20 at 8:03 am

“Agency suggests that the reason why Brexit happened and Trump won is because Russia manipulated social media.”

Weakened from “the reason” down to “one significant reason”, this is surely beyond dispute.

Anyone who had watched Manafort do Putin’s bidding in the Ukraine could see this coming as soon as he joined Trump’s team.


Phil Koop 04.09.20 at 3:19 pm

lack of agency makes for a challenging thesis but a dull novel

“Lack of agency” is basically every StanisÅ‚aw Lem novel ever; possibly, that proves your point.

Hugo Mercier’s new book presents a lot of very good evidence that human beings are more stubborn and better able to detect falsehoods than the commonly accepted myths would suggest.

We’ve already had an exchange on Twitter in which I suggested that current events are not very supportive of Mercier’s thesis. Cell phone tracking data, which is fairly objective, says that belief in deadly viruses is divided along partisan lines. My interpretation of pizzagate is different from Mercier’s. After all, I very strongly believe as a factual matter that millions of people are being immiserated and worse in Syria and Yemen, yet I don’t rush down there with a gun to intervene; and I am in the company of hundreds of millions of others like me. I think it is entirely possible that many Trump supporters believe in pizzagate in whatever sense it is possible to believe in anything, and yet do nothing about it.


Henry 04.09.20 at 8:47 pm

Yes – I think Agency isn’t as good a novel as The Peripheral – it has some very interesting ideas, but it somehow isn’t as generative as the first book was. Still well worth reading.


GMcK 04.09.20 at 11:44 pm

Gibson has stated in interviews that he had originally intended both The Peripheral and Agency to be stand-alone novels (Archangel notwithstanding). Then Trump’s election came along while he was writing Agency and he had to throw out major portions of its plot. It apparently took him awhile to realize that the world of Agency could be viewed as a stub. Then he had to rewrite large portions of the draft to incorporate the post-jackpot world and its characters. I think this slashing and burning shows. The Peripheral contains a couple of chapters that are amazing literary tours de force; Agency doesn’t have any of those.

The idea of agency has been bubbling around underneath Gibson’s stories since the beginning. It’s not hard to see Eunice as a counterpart to Wintermute in Neuromancer, as an AI struggling to break free of the restrictions imposed by its “masters”. There’s a not insignificant crop of people these days who are worrying about how AI’s may begin acting in ways that aren’t benign and properly subservient to their human creators. If those thinkers get their way, we may yet see the emergence of the Turing Police that Wintermute was afraid of. Trying to avoid spoilers, Eunice in Agency fights the same battle, with a 21st century solution.

Although I would love to see an entire series of stories built on the life and adventures of Ainsley Lowbeer, I fear that she’s relegated to the role of Mysterious Benefactor that is a staple of so many of Gibson’s novels.

As a socioeconomic system, I have my doubts about the viability of the klept that Lowbeer keeps from destroying itself. I’ve looked for a long time for a credible estimate of how many people it would take to maintain a civilization with our level of technology. Ideas like Wikipedia as a “handbook for civilization” and a project organized by the Long Now Foundation a few years ago are laughably oversimplified and unserious. Okay, if the jackpot kills off 85% of 7 billion people, that still leaves a billion people to carry on — maybe they could hold things together. The global population was 1 billion in 1800, so that’s still over the takeoff size for the industrial revolution. If we are able to approach the AI capability of Philip K. Dick’s autonomic factories (who were themselves pathetic losers in more than one story) we mght have a chance at maintaining the comfort level we’ve become accustomed to. I’m doubtful about the reachability of the technology in Lowbeer and Netherton’s world under those circumstances, though.


bianca steele 04.10.20 at 4:49 pm

I finished Agency a couple of days ago. Yeah, it’s not as entertaining as The Peripheral, though it is clever. As a concept, comparing as narrators a robot with a bunch of remote observers in its non-existent head, and a real human subjectivity, is an interesting experiment, but in practice doesn’t make for gripping reading. The action sequences are probably the best part. I find I really want to see how Gibson’s novels speak to me, unfortunately, but there’s something so uncongenial about parts of its worldview (probably its masculinist nature) that I’m pretty uncertain what I think about it.

The obviousness of Gibson’s novels being some kind of engagement with masculinity in the modern world, with unstably consistent insertions of attempts to envision how women might live there too, makes it


ozajh 04.11.20 at 10:04 am

David @ 1,

the Jackpot had come about basically through carelessness – that the people who might have been in a position to prevent it fundamentally couldn’t be bothered to do that.

The initial responses to COVID-19 of several, even many, national administrations spring to mind here.


Barry 04.11.20 at 3:56 pm


” Okay, if the jackpot kills off 85% of 7 billion people, that still leaves a billion people to carry on — maybe they could hold things together.

Remember that something which kills off 85% of the world’s population is (a) on the order of a global thermonuclear war, and (b) would likely not be distributed in such a way as to maximize the situation of survivors.

For example, wiping out urban areas (which would include the suburbs) would destroy almost all utilities serving even rural areas and eliminate the tech knowledge base.

“The global population was 1 billion in 1800, so that’s still over the takeoff size for the industrial revolution. ”

That was also a functioning society, with hundreds of years behind it in agriculture, cities, ports, shipping, organization, etc.

“If we are able to approach the AI capability of Philip K. Dick’s autonomic factories (who were themselves pathetic losers in more than one story) we mght have a chance at maintaining the comfort level we’ve become accustomed to. I’m doubtful about the reachability of the technology in Lowbeer and Netherton’s world under those circumstances, though.”

I doubt it. First, nobody could deal with the software or keeping it and the hardware running. Even if the AI’s were trying to keep things running, they’d be trying to run vast 21st century systems using the tiny non-destroyed remants.


Barry 04.11.20 at 3:58 pm

David @ 1:

” the Jackpot had come about basically through carelessness – that the people who might have been in a position to prevent it fundamentally couldn’t be bothered to do that.”

ozajh: ” The initial responses to COVID-19 of several, even many, national administrations spring to mind here.”

One way to think of this type of situation is that the elites went from parasitizing civilization to flat-out consuming it.

Comments on this entry are closed.