Skip to main content

The Fortune at the Edge of the Network [Venkatesh Rao, annotated]

Fresh Venkatesh Rao newsletter instalment that does a pretty good job of teasing out the implications of taking a tektological look at infrastructure through the lens of network theory... so good a job, in fact, that I'm going to grab and notate the whole thing, because he's managed to capsule a bunch of points I've been struggling to phrase clearly.


1/ “The last mile” is a phrase used by engineers to talk about the last (“leaf”) like segments of large networks with approximate center-to-edge topologies.

2/ In all sorts of network logistics (transport, telegraph, telephone etc), historically the "last mile" has been the bane of infrastructure. It’s where the messiest practical issues live.

3/  Right-of-way/eminent domain issues are politically/legally more complex (10 miles of cable laying in the countryside is easier than 1 block in a major city)

4/ Physical issues are more complex as well (water pipes, package deliveries, and fiber optics have different needs but often share pathways for geometry reasons).

[The above covers the basics, though it's far from basic -- see Keller Easterling's Organisation Space.]

5/ Last-mile regimes need not look like “paths” at all: waterways, spectrum rights, line-of-sight (view obstruction in real estate, glide paths for airplane landing approaches, building shadows) 

6/ In the future, drone landing/takeoff logistics, Pokemon Go type AR-conflict rights, bikes vs self-driving cars, will present novel, subtle last-mile issues.

7/ Generally though, the bottleneck is increasingly moving from literal last mile to literal last inch. Phone-to-ear, UPS-truck parking spot to porch, NFC/bluetooth, cafe power outlets.

[In my own taxonomy, this means the bottleneck has moved to the interface layer.]

8/ In raw flow volume terms, the last mile probably accounts for the bulk of actual miles traveled by anything on a network due to sheer number of endpoints.

[Note this is the exact opposite of the way in which money tends to be allocated to network development and maintenance.]

9/ The last mile is the typically the last to go hi-tech. Containerization still stops and turns into break-bulk at city limits. Fiber optics still turns into local-loop copper (DSL) in many places.

10/ As the red !!! show in the cartoon, issues get more tricky in last-block to last-inch land. It's still physically and legally complex, but that isn't the hardest part anymore.

11/ Two forces make the last block especially hard: increased demand and inequality. The case of physical packages illustrates this well.

12/ Increased demand is obvious: postal systems/FedEx etc weren't built with this much small-package flow in mind. Neither were front porches or mailboxes.

13/ Inequality is less obvious: in an unequal society there is more incentive for low-level theft and pilfering, easiest at the last block.

[Less obvious to those of us used to taking a systems perspective, perhaps; the incentive factor demonstrates just how obvious it is to those who live at the ragged edges of networks.]

14/ Anecdotally, theft from porches etc. has risen: more temptation, more people in an economic condition where they can be tempted. But careful how you interpret this. 

15/ As Anatole France sardonically observed, “The law, in its majestic equality, forbids the rich as well as the poor to sleep under bridges, to beg in the streets, and to steal bread.”

16/ Concierge services for accepting packages are now increasingly a necessity in bigger cities in middle class apartment buildings. More people are getting personal packages delivered at workplaces.

[Note that this may be a convenience issue as much as a security issue, at least in the UK context... I'd happily take the risk on the occasional pilfered package if it meant I never had to arrange another red-card redelivery, but YMMV, obvs.]

17/ You also increasingly have both large, low-value packages (e.g. cat litter) that are awkward for small locker-based systems or stairwells, and small jewelry-level value packages (iPhones)

18/ Buildings change slowly, especially in old cities with civic gridlock. It will take a decades for new buildings to reflect last-block needs. Follow the writing of Kim-Mai Cutler for this action in San Francisco.

[So now we shift from (relatively) simple material logistics and on to service and data logistics...]

19/ Similar issues occur in other networks. Consider net metering models for solar power, charging needs of electric vehicles, shopping cart services, 1-hour delivery, meal-kit businesses, etc.

20/ There are now fights over charging in charging stations, homeowners are setting up informal charging services on lawns. Blue Apron customers pile up ice packs.

21/ Even more subtleties at the informational level: Airbnb etc. require more sophisticated security for the last block: key transfers, digital locks etc. Your wallet needs RFID scanner protection.

22/ And as more and more value in flow (VIF) is in the last block at any given time, incentives for conflict and crime increase.

23/ "Stealing" cable or electricity required some sophistication, "stealing" wifi was much easier…for a while. The opportunity space will increase at all levels of difficulty.

[Ubiquity of infrastructures plus proliferation of multi-system interfaces divided by privatisation/unbundling/splintering of 'utilities'... when markets encounter habituation, ugliness happens.]

24/ The Dyn DDoS attack relied heavily on IoT devices, particularly insecure surveillance cameras. The “attack surface” as security people call it, will only increase.

[Every new interface device is potentially an interface to any other networked interface. Chips with everything, as the headlines used to go.]

25/ ATM card fraud now uses very sophisticated last-inch tech: molded plastic fake keypads, fake stripe readers on top of real ones, tiny cameras. I recently had an ATM card compromised that way.

26/ The last block/inch is also has a non-criminal economy developing: from unlocking smart-contract rental cars to power outlets in cafes that charge for a charge.

[Criminal economies are a signal of opportunity; this is just as true at the edge of the network as it is at the centre.]

27/ A lot is low-value/high volume so online micropayments arguments ("just make it free"/"not worth financializing") apply. But not all.

[Note that in this case it can be obfuscatory to focus overmuch on the material technology involved; what's interesting about these cases is how the technology gets folded into a service offer. Ownership and control over the interface layer is the opportunity recognised by criminal an non-criminal economic actors alike.]

28/ Frederik Pohl once said “the job of the sci-fi writer is to predict not the automobile but the traffic jam." Traffic jams are usually at the leaves of infrastructure trees.

[Smart guy, Pohl. Good writer, too.]

29/ Literal traffic jams happen most near/in city downtowns.  As s/w eats any network-provisioned service, traffic jams moves further down into capillaries.

[s/w = software, I think?]

30/ I like the holographic principle as a metaphor for for thinking about the effects of s/w-eats-a-network: more of the valuable information within a  volume of space can live on its surface. 

[OK, so this is where Rao's metaphor and one of my own come so close together that they almost bump noses: the infrastructural metasystem is also the metamedium, the medium of all media; hence all media is infrastructurally mediated; hence the metasystem is the veil upon which the Spectacle is projected. Logic of the Spectacle, cf. Debord: "that which is good appears, and that which appears is good"; extended by McKenzie Wark via William Gibson, "that which is secret is better [...] the secret is to the spectacle as art once was to culture. The secret is not the truth of the spectacle, it is the aesthetic form of the spectacle." So when "s/w-eats-a-network", what's really happening is that software is wrapping the deep function of the network up in a glossy package which takes Clarke's Third Law as its primary design principle.]

31/ For a network, the “volume” is the part behind the endpoints, which usually converges on one or more back-end centers. The “surface” is the set of all endpoints.

[This metaphor is really, really useful to me.]

32/ As a result, there is a LOT of economic value in the last block to last inch zone. C. K. Prahlad’s famous fortune at the bottom of the pyramid idea generalizes to “edge of any network.”

33/ In future, if current progress in brain implants continues, there may be an even bigger fortune in the “negative 1 inch” that goes into your head (disclosure: company mentioned in that article, Kernel, is a client).

[That's a pretty big 'if', IMO. But Rao knows his wider audience well, I suspect.]

34/ A general topological theory why this happens is that a more informationally powerful technology induces a higher-resolution network structure.

35/ World-eating new technologies extend the resolution of basic infrastructure networks: tens of miles for trains/planes, miles for cars, blocks for electricity, inches for wireless


36/ A network core can be defined as the low-resolution backbone where economics allows aggregation leverage, and low transaction costs for huge financial flows.

37/ This is anything you can call a “cloud” in some sense: a datacenter, a large dam, a power plant, a major interstate highway, a rail depot. I wrote about this idea in my Aeon essay American Cloud

[Personal aside: Rao's American Cloud essay was part of the inspiration for m'colleague Adam Rakunas's second novel, Like A Boss.]

38/ At the edge otoh technology stops being organized by economics, and starts being organized by social norms at its resolution limit set by transaction costs: the price of an in-app purchase for example.

39/ So sociologically, the last mile/block/inch is where the market stops and what I call an economics of pricelessness, based on values and norms, starts to kick in.


40/ When large-scale disruption happens due to a major technology like s/w, social-norms space gets systematically pushed back by market space.

[Cf. Uber, Airbnb etc etc.]

41/ The ultimate reason is physics: this is tendency towards "plenty of room at the bottom" (Feynman). As the market occupies that room, sociology (and in the future, psychology) yields to economics

42/ The transient is ugly because while you're shifting regimes, you’re converting social capital into financial capital, hurting social-capital-rich types (think priests) and enriching platform builders (think unicorn CEOs).

43/ The urban manifestation of these dynamics is gentrification: technology extending the power of markets into our community lives at increasing resolution.

44/ But if you think this process is almost over, think again. It's just beginning. You could say iOS and Android represent gentrified and slum-like digital neighborhoods in the last inch.

[There's a side-spur argument to be made about FOSS and open systems in general, here; as Rao is suggesting, FOSS can't remove these tendencies from networks, but can make it easier for people to have some control over their interfaces.]

45/ You know the old saying, "your freedom of action ends where my nose begins”? This is about to get pretty literal. There is a power struggle right by your nose/ear.

46/ But it isn’t between free individuals and an enslaving techno-capitalist cloud. You never were that free an inch from your face. You were merely the captive of non-economic forces.

47/ At worst the struggle is between the tyranny of markets and the tyranny of unchosen neighbors. The tyranny of money and the tyranny of taboos.

[Scylla and Charybdis, eat your heart out.]

48/ At best though, what we have here is technology liberating you from the tyranny of neighbors. And which view is true for you is more within your control than you think.

49/ If you see technology as potential for increased agency, you can learn to rule the last mile like a gritty cyberpunk novel protagonist, even if you don’t own a billionaire platform.

50/ If you see technology as increasing agency only for privileged others, it will become a self-fulfilling prophecy and you will end up on the losing side of this process.

51/ You will also be on the losing side if you don’t recognize that tyranny of neighbors (“hell is other people”) is a factor, a dynamic the dystopian show Black Mirror explores well.

52/ In the Black Mirror future, technology does not contend with the power of communities. It becomes allied with it to suppress individual freedom even more.

[As the title of the series makes clear: it is merely reflecting society back at itself. Brooker repeatedly makes the point that he's not writing about technology, but that technology has become a handy way to enable plot points that would have been impossible just a decade ago (though the same phenomenon has killed off older plots, e.g. the missed phonecall). The (largely good-natured) joshing that BM has become "what if phones, but too much?" misses the point; BM's not about the phones, it's about the too much, and that's not a function of the phones.]

53/ If you think this is unlikely in the real world, think again, entire countries like France seem to be exploring that direction of evolution.  

[UK, also.]

54/ This is not to absolve infrastructure titans and CEOs of big platform companies from all responsibility, or to abandon everybody to their own devices (heh!)

[No, but their position effectively denies us the possibility of taking that responsibility for ourselves; networks perform optimally as organisational monopolies, and as such are fundamentally incompatible with private ownership.]

55/ My buddy Tristan Harris has good thoughts on ethics in design for technology builders. I don’t always agree with the details of his thinking, but he’s right that with last-inch power comes great responsibility.

56/ If you’ve already decided “infrastructure creep” is bad, you’ll use dystopian metaphors like “tentacles of capitalism” or “eye of Sauron” or “the participatory panopticon” (for Black Mirror version).

57/ I personally tend to think of technology as ideology agnostic: this would happen even if we had a different ideology than neoliberal clickbaitism driving it. 

[We part ways a bit here: I'm with Kranzberg regarding the agnosticism or neutrality of technology, not least because technology is people and practices as well as material things, and people and practices are never ideologically neutral. However, I agree that a lot of the functions Rao is talking about here are endemic characteristics of networks in general, and would as such tend to occur even under different regulatory or socioeconomic regimes... but would they occur to the same extent, or at the same rate? I'm not sure, but I think it's a good question.]

58/ My preferred metaphor is the fingers/eyes of technology itself, considered as a whole (what Kevin Kelly calls the ‘technium’). 

[Ugh, Kevin Kelly. Swap all of this guff out for Haraway's cyborg metaphor, which does all the same work without trying to pretend that people and the technologies they use in their daily lives are analytically separable in any useful or believable way.]

59/ The “eyes” (or senses more generally) are getting incredibly precision in what they can see. I think of last-inch/click-tracking level “seeing” as “retina logistics” by analogy with Mac displays.

60/ The “fingers” of technology are getting increasingly delicate and precise as well. If the last-mile actuation capacity of the cloud was a sledgehammer, we’re at needlepoint now. Did your phone ding when this email arrived?

61/ This is scary to a majority, exhilarating to a minority, and as is the case for all big technology shifts, an existential crisis to those who don’t break smart.

62/ And consistent with the general political/ideological position I generally adopt in breaking smart writings, overall, increasing sensing/actuation resolution of infrastructure is a good thing.

63/ The more fine-grained the presence of technology in our lives, the more generative potential there is for humans to level-up to new, more powerful modes of being.

[Generative potential is a double-edged sword.]

64/ Whether powerful technology existing an inch from your face is good or bad depends on how good you are at using it from that locus.

[True enough. Cropping off the last few points, which are mostly marketing, but the last one's worth saving for the first sentance in particualr:]

70/ There is a nonzero-sum fortune to be created at the edge of the network...

[Yes... yes, there is. But it's slipping away, moment by moment.]