Algorithms – BLDGBLOG

How The City Uses Algorithms

New York City has announced the “Automated Decision Systems Task Force which will explore how New York City uses algorithms.” This makes New York “the first city in the country bringing our best technology and policy minds together to understand how algorithms affect the daily lives of our constituents. Whether the city has made a decision about school placements, criminal justice, or the provision of social services, this unprecedented legislation gets us one step closer to making algorithms accountable, transparent, and free of potential bias.”

(Spotted via Kate Crawford.)

The Search for Bill Ewasko

[Images: Hiking in Joshua Tree National Park; photos by Geoff Manaugh].

“In June 2010, Bill Ewasko traveled alone from his home in suburban Atlanta to Joshua Tree National Park, where he planned to hike for several days.” So begins the story of an avid hiker and Vietnam vet who went missing in Joshua Tree, a mere two-hour drive from Los Angeles, and has never been found to this day.

It has now been nearly eight years since his disappearance, but the search for Bill Ewasko never ended: people with no connection to the Ewasko family have continued to look, trading maps & GIS files online, scouring ever more remote regions of the park on foot, and arguing about the meaning of a mysterious cell-phone “ping” that seemed to place Ewasko so far outside of the original search area that, at first, many hikers simply dismissed the data.

The ongoing search for Ewasko has since become one of the most geographically extensive missing-person searches in U.S. history, with well more than a thousand miles’ worth of routes covered in Joshua Tree National Park alone.

[Image: Joshua Tree National Park; photo by Geoff Manaugh].

I began following the story of the Ewasko search in the late spring of 2016, following a series of posts on a blog called Other Hand, written by retired civil engineer Tom Mahood, and emailing a handful people still involved with the search. In the spring of 2017, I was able to join one of those searchers, Los Angeles musician Adam Marsland, in person on a new hike into a part of the park known as Smith Water Canyon. Then, when I was back in Palm Springs to report on the National Valet Olympics, I stayed in town for a few days to do several more hikes of my own, trying to familiarize myself not only with the landscape of Joshua Tree’s mountainous northwest, where Ewasko disappeared, but with the sensation of being alone there.

In Joshua Tree, even when the roads through the heart of the park are clogged with vehicles, it is often true that the instant you hike just one more ridge away from whatever trail you were meant to follow, you are utterly and completely on your own.

[Image: Joshua Tree National Park; photo by Geoff Manaugh].

A feature I wrote about the Ewasko search is now online over at the New York Times Magazine, part of their “Voyages” issue. The piece not only recounts the known details of Ewasko’s June 2010 hike, it also includes a look at so-called “lost person behavior” algorithms, deployed to anticipate how a stranger will act in an unfamiliar landscape, and it briefly reviews some of the more outlandish theories of what might have happened to Ewasko and how his cell phone appeared to be in such an unexpected region of the park.

[Image: Joshua Tree National Park; photo by Geoff Manaugh].

What drew me to Ewasko’s story in the first place was not just the fundamental mystery of how it could have happened—that is, how a competent outdoorsman could completely disappear from the surface of the Earth only two hours outside Los Angeles—but also why disappearance itself seems to draw so many people in. Trying to understand this led me to a long list of people, including musician Adam Marsland, as well as a cell-phone forensics expert and USC alum named Mike Melson who founded an independent search-and-rescue group inspired by a line from The Book of Matthew: “Your Father in heaven is not willing that any of these little ones should be lost.”

As with all stories of this kind, of course, there is so much more to tell, so many more details that only add to the mystery of Ewasko’s disappearance and to the depth of character of the people involved in searching for him, but there was not enough space to get into it all. This includes questioning the very idea of wilderness, and how we define it, when a step beyond the boundaries of civilized space can occur mere yards from the edge of a popular trail.

Here is a link to the piece, which also features evocative photographs by Philip Montgomery.

(Previously on BLDGBLOG: Algorithms in the Wild).

Nature Machine

[Image: Illustration by Benjamin Marra for the New York Times Magazine].

As part of a package of shorter articles in the New York Times Magazine exploring the future implications of self-driving vehicles—how they will affect urban design, popular culture, and even illegal drug activity—writer Malia Wollan focuses on “the end of roadkill.”

Her premise is fascinating. Wollan suggests that the precision driving enabled by self-driving vehicle technology could put an end to vehicular wildlife fatalities. Bears, deer, raccoons, panthers, squirrels—even stray pets—might all remain safe from our weapons-on-wheels. In the process, self-driving cars would become an unexpected ally for wildlife preservation efforts, with animal life potentially experiencing dramatic rebounds along rural and suburban roads. This will be both good and bad. One possible outcome sounds like a tragicomic Coen Brothers film about apocalyptic animal warfare in the American suburbs:

Every year in the United States, there are an estimated 1.5 million deer-vehicle crashes. If self-driving cars manage to give deer safe passage, the fast-reproducing species would quickly grow beyond the ability of the vegetation to sustain them. “You’d get a lot of starvation and mass die-offs,” says Daniel J. Smith, a conservation biologist at the University of Central Florida who has been studying road ecology for nearly three decades… “There will be deer in people’s yards, and there will be snipers in towns killing them,” [wildlife researcher Patricia Cramer] says.

While these are already interesting points, Wollan explains that, for this to come to pass, we will need to do something very strange. We will need to teach self-driving cars how to recognize nature.

“Just how deferential [autonomous vehicles] are toward wildlife will depend on human choices and ingenuity. For now,” she adds, “the heterogeneity and unpredictability of nature tends to confound the algorithms. In Australia, hopping kangaroos jumbled a self-driving Volvo’s ability to measure distance. In Boston, autonomous-vehicle sensors identified a flock of sea gulls as a single form rather than a collection of individual birds. Still, even the tiniest creatures could benefit. ‘The car could know: “O.K., this is a hot spot for frogs. It’s spring. It’s been raining. All the frogs will be moving across the road to find a mate,”’ Smith says. The vehicles could reroute to avoid flattening amphibians on that critical day.”

One might imagine that, seen through the metaphoric eyes of a car’s LiDAR array, all those hopping kangaroos appeared to be a single super-body, a unified, moving wave of flesh that would have appeared monstrous, lumpy, even grotesque. Machine horror.

What interests me here is that, in Wollan’s formulation, “nature” is that which remains heterogeneous and unpredictable—that which remains resistant to traditional representation and modeling—yet this is exactly what self-driving car algorithms will have to contend with, and what they will need to recognize and correct for, if we want them to avoid colliding with a nonhuman species.

In particular, I love Wollan’s use of the word “deferential.” The idea of cars acting with deference to the natural world, or to nonhuman species in general, opens up a whole other philosophical conversation. For example, what is the difference between deference and reverence, and how we might teach our fellow human beings, let alone our machines, to defer to, even to revere, the natural world? Put another way, what does it mean for a machine to “encounter” the wild?

Briefly, Wollan’s piece reminded me of Robert MacFarlane’s excellent book The Wild Places for a number of reasons. Recall that book’s central premise: the idea that wilderness is always closer than it appears. Roadside weeds, overgrown lots, urban hikes, peripheral species, the ground beneath your feet, even the walls of the house around you: these all constitute “wilderness” at a variety of scales, if only we could learn to recognize them as such. Will self-driving cars spot “nature” or “wilderness” in sites where humans aren’t conceptually prepared to see it?

The challenge of teaching a car how to recognize nature thus takes on massive and thrilling complexity here, all wrapped up in the apparently simple goal of ending roadkill. It’s about where machines end and animals begin—or perhaps how technology might begin before the end of wilderness.

In any case, Wollan’s short piece is worth reading in full—and don’t miss a much earlier feature she wrote on the subject of roadkill for the New York Times back in 2010.

The Ghost of Cognition Past, or Thinking Like An Algorithm

[Image: Wiring the ENIAC; via Wired]

One of many things I love about writing—that is, engaging in writing as an activity—is how it facilitates a discovery of connections between otherwise unrelated things. Writing reveals and even relies upon analogies, metaphors, and unexpected similarities: there is resonance between a story in the news and a medieval European folktale, say, or between a photo taken in a war-wrecked city and an 18th-century landscape painting. These sorts of relations might remain dormant or unnoticed until writing brings them to the foreground: previously unconnected topics and themes begin to interact, developing meanings not present in those original subjects on their own.

Wildfires burning in the Arctic might bring to mind infernal images from Paradise Lost or even intimations of an unwritten J.G. Ballard novel, pushing a simple tale of natural disaster to new symbolic heights, something mythic and larger than the story at hand. Learning that U.S. Naval researchers on the Gulf Coast have used the marine slime of a “300-million-year old creature” to develop 21st-century body armor might conjure images from classical mythology or even from H.P. Lovecraft: Neptunian biotech wed with Cthulhoid military terror.

In other words, writing means that one thing can be crosswired or brought into contrast with another for the specific purpose of fueling further imaginative connections, new themes to be pulled apart and lengthened, teased out to form plots, characters, and scenes.

In addition, a writer of fiction might stage an otherwise straightforward storyline in an unexpected setting, in order to reveal something new about both. It’s a hard-boiled detective thriller—set on an international space station. It’s a heist film—set at the bottom of the sea. It’s a procedural missing-person mystery—set on a remote military base in Afghanistan.

Thinking like a writer would mean asking why things have happened in this way and not another—in this place and not another—and to see what happens when you begin to switch things around. It’s about strategic recombination.

I mention all this after reading a new essay by artist and critic James Bridle about algorithmic content generation as seen in children’s videos on YouTube. The piece is worth reading for yourself, but I wanted to highlight a few things here.

[Image: Wiring the ENIAC; via Wired]

In brief, the essay suggests that an increasingly odd, even nonsensical subcategory of children’s video is emerging on YouTube. The content of these videos, Bridle writes, comes from what he calls “keyword/hashtag association.” That is, popular keyword searches have become a stimulus for producing new videos whose content is reverse-engineered from those searches.

To use an entirely fictional example of what this means, let’s imagine that, following a popular Saturday Night Live sketch, millions of people begin Googling “Pokémon Go Ewan McGregor.” In the emerging YouTube media ecology that Bridle documents, someone with an entrepreneurial spirit would immediately make a Pokémon Go video featuring Ewan McGregor both to satisfy this peculiar cultural urge and to profit from the anticipated traffic.

Content-generation through keyword mixing is “a whole dark art unto itself,” Bridle suggests. As a particular keyword or hashtag begins to trend, “content producers pile onto it, creating thousands and thousands more of these videos in every possible iteration.” Imagine Ewan McGregor playing Pokémon Go, forever.

What’s unusual here, however, and what Bridle specifically highlights in his essay, is that this creative process is becoming automated: machine-learning algorithms are taking note of trending keyword searches or popular hashtag combinations, then recommending the production of content to match those otherwise arbitrary sets. For Bridle, the results verge on the incomprehensible—less Big Data, say, than Big Dada.

This is by no means new. Recall the origin of House of Cards on Netflix. Netflix learned from its massive trove of consumer data that its customers liked, among other things, David Fincher films, political thrillers, and the actor Kevin Spacey. As David Carr explained for the New York Times back in 2013, this suggested the outline of a possible series: “With those three circles of interest, Netflix was able to find a Venn diagram intersection that suggested that buying the series would be a very good bet on original programming.”

In other words, House of Cards was produced because it matched a data set, an example of “keyword/hashtag association” becoming video.

The question here would be: what if, instead of a human producer, a machine-learning algorithm had been tasked with analyzing Netflix consumer data and generating an idea for a new TV show? What if that recommendation algorithm didn’t quite understand which combinations would be good or worth watching? It’s not hard to imagine an unwatchably surreal, even uncanny television show resulting from this, something that seems to make more sense as a data-collection exercise than as a coherent plot—yet Bridle suggests that this is exactly what’s happening in the world of children’s videos online.

[Image: From Metropolis].

In some of these videos, Bridle explains, keyword-based programming might mean something as basic as altering a few words in a script, then having actors playfully act out those new scenarios. Actors might incorporate new toys, new types of candy, or even a particular child’s name: “Matt” on a “donkey” at “the zoo” becomes “Matt” on a “horse” at “the zoo” becomes “Carla” on a “horse” at “home.” Each variant keyword combination then results in its own short video, and each of these videos can be monetized. Future such recombinations are infinite.

In an age of easily produced digital animations, Bridle adds, these sorts of keyword micro-variants can be produced both extremely quickly and very nearly automatically. Some YouTube producers have even eliminated “human actors” altogether, he writes, “to create infinite reconfigurable versions of the same videos over and over again. What is occurring here is clearly automated. Stock animations, audio tracks, and lists of keywords being assembled in their thousands to produce an endless stream of videos.”

Bridle notes with worry that it is nearly impossible here “to parse out the gap between human and machine.”

Going further, he suggests that the automated production of new videos based on popular search terms has resulted in scenes so troubling that children should not be exposed to them—but, interestingly, Bridle’s reaction here seems to be based on those videos’ content. That is, the videos feature animated characters appearing without heads, or kids being buried alive in sandboxes, or even the painful sounds of babies crying.

What I think is unsettling here is slightly different, on the other hand. The content, in my opinion, is simply strange: a kind of low-rent surrealism for kids, David Lynch-lite for toddlers. For thousands of years, western folktales have featured cannibals, incest, haunted houses, even John Carpenter-like biological transformations, from woman to tree, or from man to pig and back again. Children burn to death on chariots in the sky or sons fall from atmospheric heights into the sea. These myths seem more nightmarish—on the level of content—than some of Bridle’s chosen YouTube videos.

Instead, I would argue, what’s disturbing here is what the content suggests about how things should be connected. The real risk would seem to be that children exposed to recommendation algorithms at an early age might begin to emulate them cognitively, learning how to think, reason, and associate based on inhuman leaps of machine logic.

Bridle’s inability “to parse out the gap between human and machine” might soon apply not just to these sorts of YouTube videos but to the children who grew up watching them.

[Image: Replicants in Blade Runner].

One of my favorite scenes in Umberto Eco’s novel Foucault’s Pendulum is when a character named Jacopo Belbo describes different types of people. Everyone in the world, Belbo suggests, is one of only four types: there are “cretins, fools, morons, and lunatics.”

In the context of the present discussion, it is interesting to note that these categories are defined by modes of reasoning. For example, “Fools don’t claim that cats bark,” Belbo explains, “but they talk about cats when everyone else is talking about dogs.” They get their references wrong.

It is Eco’s “lunatic,” however, who offers a particularly interesting character type for us to consider: the lunatic, we read, is “a moron who doesn’t know the ropes. The moron proves his [own] thesis; he has a logic, however twisted it may be. The lunatic, on the other hand, doesn’t concern himself at all with logic; he works by short circuits. For him, everything proves everything else. The lunatic is all idée fixe, and whatever he comes across confirms his lunacy. You can tell him by the liberties he takes with common sense, by his flashes of inspiration…”

It might soon be time to suggest a fifth category, something beyond the lunatic, where thinking like an algorithm becomes its own strange form of reasoning, an alien logic gradually accepted as human over two or three generations to come.

Assuming I have read Bridle’s essay correctly—and it is entirely possible I have not—he seems disturbed by the content of these videos. I think the more troubling aspect, however, is in how they suggest kids should think. They replace narrative reason with algorithmic recommendation, connecting events and objects in weird, illogical bursts lacking any semblance of internal coherence, where the sudden appearance of something completely irrelevant can nonetheless be explained because of its keyword-search frequency. Having a conversation with someone who thinks like this—who “thinks” like this—would be utterly alien, if not logically impossible.

So, to return to this post’s beginning, one of the thrills of thinking like a writer, so to speak, is precisely in how it encourages one to bring together things that might not otherwise belong on the same page, and to work toward understanding why these apparently unrelated subjects might secretly be connected.

But what is thinking like an algorithm?

It will be interesting to see if algorithmically assembled material can still offer the sort of interpretive challenge posed by narrative writing, or if the only appropriate response to the kinds of content Bridle describes will be passive resignation, indifference, knowing that a data set somewhere produced a series of keywords and that the story before you goes no deeper than that. So you simply watch the next video. And the next. And the next.

Spells Against Autonomy

[Image: From “Autonomous Trap 001” by James Bridle].

By now, you’ve probably seen James Bridle’s “Autonomous Trap 001,” a magic salt circle for ensnaring the sensory systems of autonomous vehicles.

By surrounding a self-driving vehicle with a mandala of inescapable roadway markings—after all, even a person wearing a t-shirt with a STOP sign on it can affect the navigational capabilities of autonomous cars—the project explores the possibility that these machines could be trapped, frozen in a space of infinite indecision, as if locked in place by magic.

[Image: From “Autonomous Trap 001” by James Bridle].

Five years from now, rogue highway painting crews well-versed in ritual magic and LiDAR sigils shut down all machine-vision systems on the west coast…

As Bridle explained to Creators, there should be at least a handful of more examples of this automotive counter-wizardry to come.

(Earlier on BLDGBLOG: Robot War and the Future of Perceptual Self-Deception. See also The Dream Life of Driverless Cars.)

Hard Drives, Not Telescopes

[Image: Via @CrookedCosmos].

More or less following on from the previous post, @CrookedCosmos is a Twitter bot programed by Zach Whalen, based on an idea by Adam Ferriss, that digitally manipulates astronomical photography.

It describes itself as “pixel sorting the cosmos”: skipping image by image through the heavens and leaving behind its own idiosyncratic scratches, context-aware blurs, stutters, and displacements.

[Image: Via @CrookedCosmos].

While the results are frequently quite gorgeous, suggesting some sort of strange, machine-filtered view of the cosmos, the irony is that, in many ways, @CrookedCosmos is simply returning to an earlier state in the data.

After all, so-called “images” of exotic celestial phenomena often come to Earth not in the form of polished, full-color imagery, ready for framing, but as low-res numerical sets that require often quite drastic cosmetic manipulation. Only then, after extensive processing, do they become legible—or, we might say, art-historically recognizable as “photography.”

Consider, for example, what the data really look like when astronomers discover an exoplanet: an almost Cubist-level of abstraction, constructed from rough areas of light and shadow, has to be dramatically cleaned up to yield any evidence that a “planet” might really be depicted. Prior to that act of visual interpretation, these alien worlds “only show up in data as tiny blips.”

In fact, it seems somewhat justifiable to say that exoplanets are not discovered by astronomers at all; they are discovered by computer scientists peering deep into data, not into space.

[Image: Via @CrookedCosmos].

Deliberately or not, then, @CrookedCosmos seems to take us back one step, to when the data are still incompletely sorted. In producing artistically manipulated images, it implies a more accurate glimpse of how machines truly see.

(Spotted via Martin Isaac. Earlier on BLDGBLOG: “We don’t have an algorithm for this.”)

Sovereign Flocking Algorithms

[Image: Flocking diagram by “Canadian Arctic sovereignty: Local intervention by flocking UAVs” by Gilles Labonté].

One of many ways to bolster a nation-state’s claim to sovereignty over a remote or otherwise disputed piece of land is to perform what’s known as a “sovereignty cruise.” This means sending a ship—or fleet of ships—out to visit the site in question, thus helping to normalize the idea that it is, in fact, a governable part of that nation’s territory.

It is, in essence, a fancy—often explicitly militarized—version of use it or lose it.

Last summer, for example, Vietnam organized a private tour of the Spratly Islands, an archipelago simultaneously claimed by more than one nation and, as such, part of the much larger ongoing dispute today over who really owns and controls the South China Sea [sic].

Vietnam’s effort, Reuters reported at the time, was a strategic visit “to some of Asia’s most hotly contested islands, in a move likely to stoke its simmering dispute with Beijing over South China Sea sovereignty.”

It made “little attempt to disguise its political flavor, and comes as Vietnam pursues a bolder agenda in pushing its claims in the face of China’s own growing assertiveness.” Indeed, the cruise was apparently just the beginning, a mere “trial run ahead of Vietnam’s tentative plans to put the Spratlys on its tourism map, including scheduled passenger flights, possibly this year.”

Bring the people, in other words, and you bring evidence of governmental control.

Against this, of course, we must place the construction of entire islands by China, including the recent installation of a new primary school there, on an artificial island, a school whose opening lecture “was a geopolitical class that focused on China’s ownership of the sea.”

These sovereign games of Go taking place in disputed waters could sustain an entire blog on their own, of course, and are a topic we’ll undoubtedly return to. (Briefly, it’s worth noting that the sovereign implications of artificial islands were also part of a course I taught at Columbia a few years ago.)

Surprisingly, however, another region seen as potentially subject to future disputes over sovereignty is the Canadian Arctic. As such, arguments over such things as whether or not the Northwest Passage is an “international strait” (open to use by all, including Russian and Chinese military ships) or if it is actually a case of “internal waters” controlled exclusively by Canada (thus subject to restricted access), are still quite active.

Add to this a series of arguments over indigenous political rights as well as the specter of large-scale terrestrial transformation due to climate change, and a series of intriguing and quite complicated political scenarios are beginning to emerge there. (Who Owns The Arctic? by Michael Byers is an excellent introduction to this subject, as is Mia Bennett’s blog Cryopolitics.)

[Image: Flocking diagram by “Canadian Arctic sovereignty: Local intervention by flocking UAVs” by Gilles Labonté].

With all this in mind, consider a fascinating report issued by Defence R&D Canada back in 2010. Called “Canadian Arctic sovereignty: Local intervention by flocking UAVs” (PDF), and written by Gilles Labonté, it opens stating that “the importance of local intervention capability for the assertion of Canadian Sovereignty in the Northwest passage is recognized.”

However, Canada presently lacks the ability to deploy at any northern position, on demand, assets that could search a wide area for rescue or surveillance purposes. This fact motivated the exploration we report here on the feasibility of a rapid intervention system based on a carrier-scouts design according to which a number of unmanned aerial vehicles (UAVs) would be transported, air launched and recovered by a larger carrier aircraft.

In other words, if Canada can’t send actual Canadians—that is, living human beings—on aerial “sovereignty cruises” by which they could effectively demonstrate real-time political control over the territories of the north, then they could at least do the next best thing: send in a flock of drones.

Doing so, Labonté suggests, would require a particular kind of flocking algorithm, one with an explicitly political goal. “In the present report,” he adds, “we propose a solution to the remaining problem of managing simultaneously the many UAVs that are required by the vastness of the areas to be surveyed, with a minimum number of human controllers and communications.”

Namely, we present algorithms for the self-organization of the deployed UAVs in the formation patterns that they would use for the tasks at hand. These would include surveillance operations during which detailed photographic or video images would be acquired of activities in a region of interest, and searching an area for persons, vehicles or ships in distress and providing a visual presence for such. Our conclusion is that the local intervention system with flocking UAVs that we propose is feasible and would provide a very valuable asset for asserting and maintaining Canadian Sovereignty in the North.

There are “formation patterns” and flocking algorithms, this suggests, that would specifically be of use in “asserting and maintaining Canadian Sovereignty in the North.”

Hidden within all this is the idea that particular flocking algorithms would be more appropriate for the task than others, lending an explicit air of political significance to specific acts of programming and computation. It also implies an interesting connection between the nation-state and behavioral algorithms, in which a series of behavioral tics might be ritually performed for their political side-effects.

For some context, the report adds, “the Canadian Government has had serious considerations of establishing a presence in the north through purchasing nuclear submarines and ice-breakers.” But why not side-step much of this expense by sending UAVs into the Arctic void instead, reinforcing nation-state sovereignty through the coordinated presence of semi-autonomous machines?

Simply re-launch your drones every two or three months, just often enough to nudge the world into recognizing your claim, not only of this remote airspace but of the vast territory it covers.

A halo of well-choreographed aerial robots flocks in the Arctic skies before disappearing again into a bunker somewhere, waiting to reemerge when the validity of the government appears under threat—a kind of machine-ritual in the open three-dimensional space of the polar north, a robotic sovereignty flight recognized around the world for its performative symbolism.

Read the rest of Labonté’s paper—which is admittedly about much more than I have discussed here—in this PDF.

“We don’t have an algorithm for this”

[Image: Comet 67P, via ESA].

In the story of how European Space Agency researchers are scrambling to locate—and possibly move—the Philae probe, which they successfully landed on Comet 67P two days ago, there’s an interesting comment about computer vision and the perception of exotic landscapes.

[Image: Comet 67P, via New Scientist].

“We’re working our eyes off,” one of the scientists says to New Scientist, describing how they are personally and individually poring over photographs of the comet.

“It’s an entirely manual process,” New Scientist continues, “because the complex and bizarre landscape of comet 67P defies any kind of automated search. ‘We don’t have an algorithm for this,’ he says.”

We don’t have an algorithm for this.

[Image: The irregular terrain of Comet 67P, via ESA].

It would be interesting to develop a taxonomy of landscapes based on their recognizability to algorithms. This would tell you as much about how computers see the world as it would about the aesthetic assumptions—even the geological biases—of the people who programmed those computers.

Think, for example, of Adam Harvey’s work, asking When Is An Apple No Longer An Apple? That project explored the point at which machine-learning algorithms could no longer distinguish the iconic fruit from a jumble of colorful objects.

Or take Harvey’s more recent CV Dazzle experiment, which looked at how to prevent facial recognition software from identifying a face at all through the clever use of cosmetic camouflage.

However, in the case of Comet 67P and other extreme topographic environments, we would be looking at when a landscape is no longer a landscape, so to speak, at least in terms of the computer-vision algorithms programmed to analyze it.

[Image: Comet 67P, via ESA].

What other landscapes fall within this category—of spatial environments unrecognizable to machines—and what do those spaces reveal about the dimensional prejudices of the algorithm? Light and shadow; depth and range; foreground and background; geometry and complexity.

Bump Adam Harvey’s investigations up to the scale of a landscape, and a million potential design projects beckon. Learning from Comet 67P.

(Earlier on BLDGBLOG: The Comet as Landscape Art).

The Comet as Landscape Art

[Image: Photo courtesy ESA].

Intrigued by these images as an example of how the tradition of landscape representation has rapidly progressed—from the Romantics and the Hudson River School to Rosetta—I felt compelled to post a few photos of the craggy and glacial surface of Comet 67P/Churyumov–Gerasimenko, sent back to Earth yesterday from the European Space Agency’s Rosetta spacecraft.

The surface of the comet “is porous, with steep cliffs and house-sized boulders,” making it earth-like yet deeply treacherous, an irregular terrain to photograph and a dangerous place to land.

[Image: Photo courtesy ESA].

It is the notion of “land” here that is most interesting, however, as this is really just the imposition of a terrestrial metaphor onto a deeply alien body. Yet the comet is, in effect, literally a glacier: a malleable yet permanently frozen body of ice hurtling through space, occasionally exploding in comas and tails of vapor.

It is “an ancient landscape,” we read, “and yet one that looks strangely contemporary as the sun vaporizes ice, reworking the terrain like a child molding clay.”

Think Antarctica in a winter storm, not southern Utah—or Glacier National Park, not the Grand Canyon.

[Image: Photo courtesy ESA].

Along those lines, some of the most provocative writing on what it means to visually represent the frozen and hostile landscapes of the Antarctic is by writer William L. Fox, whose work offers some useful resonance here.

Fox has written, for example, about the technical and even neurological difficulties in accurately representing—let alone comprehending or simply navigating—Antarctic space and the vast forms that frame it.

Distant landscapes distorted by thermal discontinuities; white levels pushed to the absolute limit of film chemistry; impossible contours throwing off any attempt at depth perception; even the difficulty of distinguishing complicated mirages from actual landforms: these are all part of the challenge of creating images of an exotic landscape such as the Antarctic.

As Fox writes, it was even specifically the tradition of Dutch landscape painting, combined with the maritime practice of sketching coastal profiles, that first introduced the visual world of the Antarctic to western viewers: it was thus seen as an ominous, ice-clogged horizon of fog and low clouds looming always just slightly out of ship’s reach at the bottom of the world.

He calls this the genre of “representational exploration art.”

[Image: Photo by Stuart Klipper from his fantastic book, The Antarctic: From the Circle to the Pole, with a foreword by William L. Fox].

In one interesting passage in his book Terra Antarctica, he suggests that the south polar landscape is so extreme, it often resists natural analogy. As Fox describes it, the wind-carved boulders and isolated pillars and cliffs of ice are more like “artworks by Salvador Dalí and Henry Moore, evoking the spirit of surrealism with the former and modernist forms with the latter. The Antarctic is so extreme to our visual expectations that, once we attempt to move beyond measurement to describe it, analogies with other parts of nature fall short, and we resort to comparisons with cultural artifacts that push at the boundaries of our perceptions.”

These include “cultural artifacts such as sculpture and architecture, products more of the imagination than of nature.”

Consider, for instance, that comet 67P is widely known today as the “rubber-duck comet” due to its bifurcated structure, implying, as Fox suggests with the Antarctic, that no natural analogy seemed adequate for describing the comet’s geometry.

[Image: The gateway arches of the Antarctic; photo by Stuart Klipper from, The Antarctic: From the Circle to the Pole, foreword by William L. Fox].

But what are we to make of comet 67P now that we can see it as a physical landscape, not just an ephemeral optical phenomenon passing, at great distance, through the sky? When a blur becomes focused as terrain, what is the best way to describe it? What visual or textual traditions are the most useful or evocative—vedas and sutras or laboratory reports?

Put another way, is poetry as appropriate as a scientific survey in such a circumstance—should “we attempt to move beyond measurement to describe it,” in Fox’s words—and, if not, what new genres of exploration art might result from this spatial encounter?

I’m reminded here of poet Christian Bök’s wry remark on Twitter: “I am still amazed that poets insist on writing about their divorces, when robots are taking pictures of orange, ethane lakes on Titan…”

Now that humans are beginning to land semi-autonomous camera-ships on the frozen ice fields of passing comets, sending back the (off)world’s strangest landscape art—as if a direct line runs from, say, the pastoral landscapes of Claude Lorrain or the elemental weirdness of J.M.W. Turner to the literally extraterrestrial boulders and gullies depicted by Rosetta—how should our own descriptive traditions adapt? What, we might ask, is comet 67P’s role in art history?

[Image: Approaching 67P, via the ESA].

Computational Mythologies: An Interview with Zachary Mason

[Image: “Homer, the Classic Poets,” by Gustave Doré, from Canto IV of The Inferno].

Novelist Zachary Mason’s Lost Books of the Odyssey has been described by The New York Times as “dazzling… an ingeniously Borgesian novel that’s witty, playful, moving and tirelessly inventive.”

As Slate’s John Swansburg describes it, the book is a fictional anthology of “Homeric apochrypha—versions of the Odysseus story that circulated in the time before Homer but were left out of the epic as we came to know it.” Yet “Mason’s enterprise never devolves into a mere high-concept exercise,” Swansburg adds. And I agree: the book’s constantly shifting short narratives offer a kind of stratigraphic road-cut straight to the contested origins of Western mythology, where a storm-wracked, war-torn archipelago is ceaselessly crossed by a homesick husband fighting to return to his family—only Mason has taken these elements and cross-wired them, creating a dreamlike, parallel landscape of new heroic sequences, echoes, and myths.

In the following interview, Zachary Mason speaks to BLDGBLOG about his book; its use of the archipelagic landscapes of ancient Greece for new, combinatorial ends; the algorithmic templates underlying much of his fiction; his current work on Artificial Intelligence; the future of automated construction technologies, including 3D-printing, a theme explored in Mason’s most recent work; other possible narrative directions for further rewritings of The Odyssey (including a version set in the Caucausus Mountains, with, as Mason describes it below, “a huge system of unreliable, unmapped and essentially creaky rope-bridges strung up between the peaks”); and much more. We spoke by phone.

[Images: Illustrations by Willy Pogány for The Adventure of Odysseus and the Tale of Troy, Padraic Colum’s own retelling of The Odyssey].

• • •
BLDGBLOG: I’d like to start with one of the most memorable images in the book, that of “Agamemnon’s Fortress.” Could you describe that briefly?

Zachary Mason: In that chapter, the Greeks, having no building materials except for the timbers of their ships, and expecting the siege to last a long time, have excavated their forward base in the sand in front of Troy.

This chapter has the structure of a fairy tale, or something out of the Arabian Nights. It starts with Agamemnon’s sense of helplessness; for all his armies and his heroes he can’t take a single city, which leads him also to reflect on the extent of his ignorance, so he calls together his three wisest counselors and, not being one for half-measures, asks them to explain, essentially, everything in the world. Three times he asks them, and each time they come back with a denser and perhaps pithier solution, and with each iteration more time passes.

The underground base becomes first a city, then a network of cities, that keep getting deeper as the old cities crumble and are used only for storage chambers and secret passages, and, all this time, Troy is only about half a mile away. By the end of the story Troy has been abandoned, so there’s no further reason for the Greeks to be there, but they still are, and they’re still digging deeper.

Part of what I was doing was taking the structure of a fairy tale—often there are three questions, animals, obstacles or what have you—and making the progression between the iterations exponential, rather than constant, so there’s a drastic acceleration.

Also, there’s something fascinating about this improvised, temporary, and quite uncomfortable underground base becoming permanent and entrenched, and going ever deeper, starting to dominate the lives of the residents with its deranged logic. It’s reminiscent of an ants’ nest, or the World War II eras quonset huts still in use at SRI.

BLDGBLOG: Or Kafka’s “Burrow“, another story of tunneling. What I like about the image is its dichotomy between the aboveground walled fortress of Troy, with its stone walls and permanent streets and houses, and its long-term sense of history, compared to the underground maze of the invading Greeks, constantly turning this way and that and digging deeper into the earth. It’s a nice juxtaposition.

Mason: Troy is the absence of possibilities, in a sense; it’s just there and the Greeks can’t do anything about it, no matter how much they try. In the sand, though, there are infinite possibilities, all of them fairly useless.

[Images: Illustrations by Willy Pogány for The Adventure of Odysseus and the Tale of Troy, Padraic Colum’s retelling of The Odyssey].

BLDGBLOG: You also describe how, when the sand walls collapse, the Greeks implement laws that say the soldiers can’t excavate or uncover things that have been buried. They’re forced to avoid their own past, in a sense, and keep digger new tunnels. It’s like a legislatively enforced amnesia, or a living archive that refuses to excavate itself.

Mason: I liked the idea that they would become trapped by their own superstitions: prevented from doing the rational thing—as far as planning went—and obliged by this unfortunate belief to keep digging themselves in deeper.

In a way, it’s the opposite of amnesia, if you think of the collapsed chambers as preserved. As though we were forbidden to repair collapsed or damaged buildings on the surface, and cities become theme-parks of stratified decay.

BLDGBLOG: There’s another image in the book that really caught me: Ilium, “death’s city,” full of “uncountable mausoleums” and constructed from bones. “The high walls of Death’s city became the ubiquitous background of the Greek’s dreams,” you write.

Mason: In that chapter Troy has become Death’s city, and it is implied that all of Hades is contained within its walls. I imagined Death’s city as a place of levels, reaching down forever; it goes so deep, that not even its inhabitants have seen all of it, which somehow seems to gel with the way the representation of Death as an object of obsessive focus.

In this story, Menelaus eventually defeats and overthrows Death, and though he intends to destroy his city, but he end up doing no more than taking Death’s place, and adding new levels to the already infinite levels of the city.

[Images: Illustrations by Willy Pogány for The Adventure of Odysseus and the Tale of Troy, Padraic Colum’s retelling of The Odyssey].

BLDGBLOG: I’m curious, pulling back a level, how a particularly evocative city description or landscape description can, in and of itself, achieve something on a narrative level that other rhetorical devices often aren’t able to or have a harder time accomplishing. It interests me, for instance, that if your book was set in a very different place or geography—in central Illinois, say, wandering from town to town—those facts alone, before characterization even kicks in, would hugely affect the mood or tone of the story. Part of the imaginative appeal of The Odyssey itself, I would say, has a lot to do with the archipelagic landscape it takes place within; if Odysseus had just wandered around the Caucasus Mountains, from peak to peak, instead of island to island, then the story’s cosmic overtones—wherein each island is its own micro-cosmic world, with its own sequences of experience—would have been achieved only in a quite different rhetorical way.

Mason: [laughs] I’m imagining The Odyssey set in Illinois—how the Trojan War would be a fight for one particular bit of plain amidst otherwise completely identical expanses of plain, and how that would add a sense of futility to Odysseus’s homeward journey—he weeps when he finally sets foot on Ithaca, which is fifty hectares of absolutely undistinguished farmland.

Here’s an idea: set The Odyssey in the Caucasus, but with a huge system of unreliable, unmapped and essentially creaky rope-bridges strung up between the peaks. The valleys are full of bandits, and hardship, and take a long time to navigate—but, though the ropes are faster, no one knows quite why they’re there, or their connectivity. Then you’d have something that feels at least a little like The Odyssey.

A nice thing about islands, as opposed to regular old landscape, is that they seem completely knowable. With an island, one could have a clear view of all of the elements in play in whatever narrative, and of the island’s history, and of the full significance of everything. One’s understanding of a continent is necessarily hand-wavey, and things are probably changing faster than one can keep track of them.

There was an older version of the Lost Books—or, at any rate, another book that ended up getting folded into what eventually became the Lost Books—which was going to be much more explicitly geographical. Every story was going correspond to an island, and the elements of those islands would be specified by a combinatoric system. I made up a table of elements, and I was duly working my way through the possible combinations, but it turned out to be very, very difficult to make this work; I couldn’t finish it, though you can still see echoes from time to time.

I think art tends to turn out best under moderate constraint; the combinatoric system was probably a little too strong. But I kind of like the way the character of the old, never-quite finished book shows up in the Lost Books (there’s actually more than one unfinished ghost-book lurking in the Lost Books), because its interesting when there are multiple patterns that partially describe, in this case, a book, but where none of them completely describe it. Its a little like complexity theory—too much order and you get banal rigidity, but too little and you get chaos, and the interesting things are on the boundary between the two.

[Images: Illustrations by Willy Pogány for The Adventure of Odysseus and the Tale of Troy, Padraic Colum’s retelling of The Odyssey].

BLDGBLOG: Are you drawn to things like the Oulipo, or other sorts of literary games?

Mason: I’ve always been drawn to Oulipo. I have one life in math and science, and another in literature, so Oulipo is compelling as the intersection between the two. On the other hand, Oulipan games don’t always work. There are a few products of Oulipo that are brilliant, and some that are interesting, and more that are the literary equivalent of musical scales.

The book was, in its original conception, intensely Oulipan, but I couldn’t get it work that way, so I ended up relaxing the constraints I had imposed on myself, lest I end up with something that felt like a sterile exercise rather than an organic whole.

So you might say that, for me, Oulipo is a good starting point but not a good finishing point.

[Images: Illustrations by Willy Pogány for The Adventure of Odysseus and the Tale of Troy, Padraic Colum’s retelling of The Odyssey].

BLDGBLOG: Is the final sequence of chapters arranged for narrative effect, then, or is it based on some other sort of underlying structure or combinatorial path?

Mason: In an early version of the book I did use an algorithm to order the chapters. In those days, each chapter was associated with a handful of keys—broad themes like “time” and “the gods” and “revenge” and so forth. I wrote a program that used simulated annealing to order the book in a more-or-less optimal way, where optimality was defined as maximizing the number of overlapping keys between adjacent chapters. The intent was to produce an ordering where there was always a strong sense of continuity between chapters, but where the nature of that continuity varied with every boundary.

In the end, I didn’t like the ordering the algorithm produced, and realized that there were actually other rules I wanted to follow, some of which didn’t lend themselves to formalization, so I ended up arranging the chapters by hand. I try to alternate long and short chapters, and its good when adjacent chapters rhyme, thematically; also, the book now starts off by establishing the kind of recombinatoric game I’m playing with The Odyssey, and then, as you get toward the end, that pattern breaks down, and you get all sorts of strange things—The Odyssey interpreted as a chess manual, for instance.

[Images: Illustrations by Willy Pogány for The Adventure of Odysseus and the Tale of Troy, Padraic Colum’s retelling of The Odyssey].

BLDGBLOG: The chess manual chapter—“Record of a Game”—is fantastic. Could you describe that chapter briefly and explain its conceit?

Mason: “Record of a Game” is a chapter that explains how The Iliad is not, in fact, an epic, but an ancient chess manual. This chapter explains that chess radiated out from India and took on locally idiosyncratic forms in most Indo-European cultures; in ancient Greece, it assumed a form in which the pieces, rather than being faceless icons, are strongly individuated. There were a few particular games that were considered to embody everything that was worth knowing about the game, and chess masters had to memorize those games precisely. Various mnemonics were added to make this task easier and eventually, to the uninitiated, the records of these games came to seem like heroic narratives, which was aided when the mnemonics were misinterpreted as epic clichés—Thetis being “trim-ankled,” Achilles “fleet-footed” and so forth.

A lot of the book is about interpreting The Odyssey as a code, so that Homer’s text is understood as a distortion of some underlying signal, and it is that signal, under various assumptions, that one is trying to infer. “Record of a Game” is perhaps the most extreme example of this, in that it explains away almost everything about The Iliad.

The coda to this chapter explains that The Odyssey is a sort of fictive chess manual, describing the motion of the pieces after the game has finished and the players have departed, in which the Odysseus piece is trying to get back to its home square. So it a sort of second-order game.

BLDGBLOG: Interpretation, here, becomes a form of paranoia—more an act of invention than one of reading.

Mason: There are some aspects of The Iliad that lend themselves almost eerily to this kind of interpretation—like the famous catalog of ships, which is also famously boring. In “Record of a Game,” it’s explained that the catalog of ships is properly understood as a description of the opening in a chess game.

Then there are all the lists of killing—this warrior slew that warrior, and that warrior slew this other warrior—which is not, I think, hugely interesting in itself, but, if you look at it as a series of exchanges in the middle game, begins to make sense sense.

On the other hand, Homer has been what one might call exhaustively interpreted. You can sit alone in your living room and make up the craziest, most implausible theory about Homer that you can, and then go to Google, you’ll find that some serious person with solid academic credentials has dedicated his career to espousing your preposterous theory.

BLDGBLOG: [laughs] Like Shakespeare wrote Homer.

Mason: Exactly.

[Images: Illustrations by Willy Pogány for The Adventure of Odysseus and the Tale of Troy, Padraic Colum’s retelling of The Odyssey].

BLDGBLOG: If that’s the case, do you see your own book as participating in, and thus continuing, this sort of interpretive culture? Or is it more of a parody?

Mason: It’s a bit of both. I was certainly aware of the exhaustive interpretation of Homer, and I guess I thought of the Lost Books as enabled by that, and somehow setting a cap on it, by being the logical culmination and maximum expression of this tendency. It’s as though I was saying, “You call that interpretive chaos? I’ll show you interpretive chaos!”

That said, I wasn’t trying to put Homeric interpretation out of business or make any big, stomping academic points. It just seemed like this tradition both suggested and licensed a really fun thing to do with the book.

And then, The Odyssey seems to lend itself uniquely to this kind of remixing, in that it’s compelling at almost any granularity. The way its written is compelling in the details, but, at a coarser level, what one might call its language of imagery is powerful. The Odyssey retains considerable power even when reduced to a plot synopsis, which isn’t true of many books—a plot synopsis of The Inferno or Lolita is unlikely to be hugely interesting. Cormac McCarthy’s The Road might come close, but, really, it just has a single image. As I say this, it occurs to me that many Borges stories would still be compelling as a single paragraph precis.

[Images: Illustrations by Willy Pogány for The Adventure of Odysseus and the Tale of Troy, Padraic Colum’s retelling of The Odyssey].

BLDGBLOG: Stepping back a bit, your author bio refers to you as an Artificial Intelligence researcher, but I’m curious what that actually means.

Mason: The first thing to notice about AI is that there isn’t any; in some sense, there’s been no real progress in the field. We don’t know much more about the computational character of cognition than we did in 1950. So I, like a number of people, am interested in trying new ways of approaching the problem. New kinds of computational models of perception and language are, I think, one promising path.

A particular interest of mine is computational models of design. Design problems are kind of a sweet spot, insofar as they offer deep domain richness, but they don’t too much background knowledge, which is very difficult to handle, computationally. You just need models of artifacts, and the way those artifacts are interpreted.

One of the problems with A.I. is that interacting with the world is really tough. Both sensing the world and manipulating it via robotics are very hard problems, and solved only for highly stripped-down special cases. Unmanned aerial vehicles, for instance, work well because maneuvering in a big, empty, three-dimensional void is easy—your GPS tells you exactly where you are, and there’s nothing to bump into except the odd migratory bird. Walking across a desert, though—or, heaven help us, negotiating one’s way through a room full of furniture in changing lighting conditions—is vastly more difficult.

BLDGBLOG: Saying this purely as a dilettante, it seems like there are at least two models of Artificial Intelligence. One of them is about spatial navigation, as you say, but another is more textual, or language-based. This latter version touches on things like the Turing Test, of course, but also on things like the text-mining industry, where they’ve developed intelligent software programs that can read through hundreds of thousands of pages in a flash and find the keywords or phrases that you’re looking for—which is different from a Google search.

Mason: Text-mining is well and good, but there’s a sense in which it’s not A.I. The programs don’t understand the text in any meaningful way. They manipulate it statistically, and, in that way, they’re able to accomplish things that appear intelligent—but there’s no actual comprehension.

You can take text analytics and that sort of thing up to a certain point, and you can get some pretty impressive results—Google works well—but there’s a hard boundary that you’re not going to be able to cross if you don’t have a full-fledged model of cognition.

Nobody’s figured out how to make that model, so there are hard limits on how far things like Google and text analytics can go. This is tacitly understood, for the most part (though I’ve spoken with some Googlers who have seemed guilty both of hubris and of not understanding A.I.’s history), but it’s bad business to admit it. So, when they say their algorithms are intelligent, or that their algorithms understand the text and so forth, its just fatuous marketing-speak.

[Images: Illustrations by Willy Pogány for The Adventure of Odysseus and the Tale of Troy, Padraic Colum’s retelling of The Odyssey].

BLDGBLOG: I’m curious how all this comes together in The Lost Books of the Odyssey. So far, we’ve talked about combinatorics, allegories, Artificial Intelligence, and the Oulipo, and I know that, for me, there were moments while reading the book when it felt almost as if a program had been fed certain narrative parameters—cave, cyclops, Odysseus, boat—and the remixed results became the Lost Books, as if it were the output of a demented A.I. program. Were you hoping to use the book itself as a model for computational literature?

Mason: There was a time in my life when I would have been very happy to have it suggested that my work was the output of a demented A.I. program—

BLDGBLOG: I meant that in a positive way!

Mason: A counter-question for you is: do you think you would have thought that if my bio hadn’t said that I worked with A.I.?

BLDGBLOG: Perhaps not. But, on another level, especially with a book like yours, doesn’t the author bio become a deliberate way to frame the book’s contents? It helps to flavor how a book is received and interpreted.

Mason: That’s a fair point. In fact, in the first edition of the book, I used a fake author bio. I claimed to be an archaeo-cartographer and paleo-mathematician at Magdalen College, Oxford, the holder of the John Shade Chair. Note that archaeo-cryptography and paleo-mathematics don’t exist as disciplines, and John Shade is a character in a Nabokov novel. I was perhaps unreasonably pleased with this trick, not least because much of the book is about endless recursions of false and manipulative identity—so it seemed to fit, rather than being arbitrary hijinks.

I was persuaded to use a real biography for the FSG edition, and have since regretted it. My author-bio says I do A.I., because I thought it was an interesting hook, but it seems to color the way people approach the book now, and not necessarily in desirable ways. On the whole, I’d like the book to be read without reference to my biography. Perhaps I should really have gone beyond the bounds of the plausible and claimed to be a graduate of the Iowa Writer’s Workshop, and that I now teach creative writing at a small Midwestern liberal arts school.

I’ve been sufficiently cranky about this that I’ve considered saying that, in fact, I’m just a writer, but there’s another guy with the same name as me who is a computer scientist specializing in A.I. We met because we have similar gmail addresses, and sometimes get each other’s mail. He was kind enough to read an early draft of my book, and, being a Borges fan, like disproportionately many well-read scientists, wondered how differently the book would read if it had been written by an A.I. guy. I thought that sounded like a good hook, and ran with it.

But, to answer your question more directly, I certainly wasn’t going for anything overtly combinatoric, or at least not after the book’s very earliest days.

It sounds like you’re reacting to my preoccupation with what I might call the primes of the story. There are aspects of the Odyssey that seem essential, and these are few in number, just a handful of images. There’s a man lost at sea, an interminable war a long way behind him, and a home that’s infinitely desirable and infinitely far away. There’s the man-eating ogre in his cave; there are the Sirens with their irresistible song; there’s the certain misery of Scylla and Charybdis.

I feel like these images are responsible for the enduring power of the story, and its survival, more than the particular details of, say, dialogue among the suitors, or what have you. I wanted to work directly with these primes, to present them in as powerful and stripped-down a way as possible, and to explore how they could interact, and how they could combine to make new forms. I suppose this kind of minimalist, reductive aesthetic does has a mathematical flavor.

[Images: Illustrations by Willy Pogány for The Adventure of Odysseus and the Tale of Troy, Padraic Colum’s retelling of The Odyssey].

BLDGBLOG: In the more recent fiction that you sent me, you’ve been exploring quite strong architectural and urban imagery. I’m curious to hear more about how urban imagery, in particular, features in your more recent fiction, and other ways that urban and architectural descriptions are being foregrounded in your work.

Mason: I sent you some fragments of a book, tentatively titled Void Star, in which architecture does feature rather prominently.

The book is set in the murkily indefinite future. Technology has improved, and robotics works much better, so much so that it has become a cheap, boring technology, and construction robots, in particular, are ubiquitous—they’re essentially 3D printers scuttling around on insect-like legs. Right now, researchers are taking the first steps toward building robots where you can set them loose and they’ll assemble complicated structures—often, interestingly, mimicking the control principles used by social insects—and I thought how interesting it would be, and how different the world would look, if these things ever actually work.

Today, anybody can go to Home Depot or its equivalent and buy the materials to build a shed; but construction, on a large scale, is very expensive and reserved for wealthy organizations. It’s a rare privilege to actually get to build something. If these robots exist, then architecture is democratized: anyone with a few bucks can build a structure to whatever specifications they like.

Once you have that, cities start to metastasize and grow. Favelas and other improvised and illegal shadow cities become marvelous, growing layer upon layer, like coral reefs.

Another architecturally salient aspect of Void Star is that, in the book, A.I.s exist, but they’re not like anyone expected. They’re intelligent but not human; in fact, their minds and perspective and languages are so different that people can’t really talk to them—they’re much more like Stanislaw Lem’s Solaris than Commander Data, the Terminator, Agent Smith, or HAL. Despite this, they can still be useful—in design tasks, for instance. They write most of the world’s software, and do it very quickly—the amount of code in the world increases by many orders of magnitude, but nobody knows how it works. Software development becomes a process less of hacking code than establishing some sort of shared understanding with these strange, essentially foreign intelligences.

The A.I.s also design buildings, and they think so fast, and with such breadth, that their designs are more complete than is otherwise possible. Buildings become much more complicated, and better thought-out—in a sense, absolutely thought-out. The A.I. might consider, say, the light and the acoustics at every spot in the building at every time of day and every day of the year, and the kinds of relationships that you could then create between the experiences at these different locations. Also, because the machines have such fine-tuned control of the way buildings are constructed, they can implement design motifs that go down almost to the molecular level. In buildings as they are, there is, inevitably, unarticulated matter—a girder is just a girder, concrete is just concrete—but the machines could make their artifacts fractally ornate at every level. They would, in some sense, be complete artifacts.

It will be an interesting world.

• • •
Thanks again to Zachary Mason for taking the time to have this conversation. Pick up a copy of The Lost Books of the Odyssey, meanwhile, which came out in paperback last month, and see what you think.