Software is Eating the [War]

Why do we need “A DOD Embassy in Silicon Valley?” because our enemies are already utilizing cutting-edge commercial technologies against us, and are likely to continue. While current versions of this meme look manageable (IED’s, reconnaissance drones), even more rapid technological change is coming. We need to engage with the people and industries who are intimately familiar with transformation in order to maintain our strategic edge.

Software (sôft’wâr’): The programs, programming languages, and data that direct the operations of a [computer] system.


Militaries have been running software for millennia, though we have called its installation “training,” its conceptualization “doctrine,” and its execution “combat.” Our hardware (“Recruit, 1") receives a new operating system at bootcamp or officer candidate school, along with a new hair cut, uniform, ways to walk, and even talk (“Sir, [data], Sir!”). We then install advanced software and pair the hardware with specialized peripherals at additional training schools, turning a single unit capable of shining shoes, marching, and communicating, into “fire teams,” “jet pilots” or “gunners’ mates”.

Newly trained, they arrive at, “battalions,” “destroyer squadrons” or “air wings,” and are integrated with other (soldiers/sailors/airmen/marines) into constructs capable of complex executable tasks, from flying and maintaining wing-shaped supercomputers with explosives strapped to their undersides, to taking a long metal tube, placing a nuclear reactor inside of it, sealing it off from the outside environment, submerging it, and sending it across the ocean while keeping its occupants and payloads intact and operable.

Ask yourself: What is a squad flanking maneuver? A Mongol “feigned retreat?”

The answer: Software.

For those of you who know JavaScript, think of it like this:

function phalanx(enemy) {
if (enemy.size == 0) {generals.order("Continue Marching!");}
if (enemy.size > 0) {generals.order("FORM FOR BATTLE!");} }

In the past, the above codebase was installed biologically through training and communication, and then executed at the will of a political entity put in charge of the maneuver unit. It was up to that commander to observe the situation, and, if he saw the enemy, give the order to form up.

Software and the OODA-Loop

Military officer and philosopher John Boyd believed that combat effectiveness derrived from one’s ability to observe, orient, decide, and act (the so-called “OODA-loop”). Boyd, a fighter pilot by training, began his career fighting in, and then thinking about air-to-air engagements — an environment where time is so precious that thinking and acting faster than the opponent is often the difference between life and death.

According to Boyd, the OODA-loop is a verbal representation of the human decision-making cycle. And in challenging circumstances (e.g. air-to-air combat), the ability to make and execute decisions faster than the enemy allows one to act, and then counter enemy reaction before the enemy has even completed their counter-move. Thus, effectiveness increases as your cycle-time (speed of “loop”) decreases.

via Wikipedia

Using the software framework above, the OODA process is simply a way of describing how quickly hardware and software interact to bring about desired actions — and an “OODA-loop”, describes the iterative process of acting in order to achieve an objective.

Decreasing The Loop

With technology, we decrease the OODA-loop by boosting processing power, storage capacity, frame-rates, etc. In war, we decrease the OODA-loop with challenging training, innovative tactics, and cutting-edge technology.

In practice we always base our preparations against an enemy on the assumption that his plans are good; indeed, it is right to rest our hopes not on a belief in his blunders, but on the soundness of our provisions. Nor ought we to believe that there is much difference between man and man, but to think that the superiority lies with him who is reared in the severest school. — Thucydides, “The Peloponnesian War”

But improvement has, over the past century, still been largely been incremental in scope. While certainly satellites and cruise missiles have improved our ability to perceive or act, they have not transformed it. Our ability to observe remains the ability for a human being to observe with aid of a satellites. Predator drones beam information to a pilot in a small box in the Nevada desert, as if they were staring out from the cockpit.

All that is about to change. Machine learning will soon bring us to an inflection point where we will see nearly exponential gains in our ability to observe and orient in near-real-time.

As (Processing Power) → ∞, (OODA)→ 0

Moore’s Law has deep implications for the OODA-loop. Once computers have human-scale computational abilities, they will be able to contextualize information and draw inferences — like determining if a boat floating off the Horn of Africa is running khat to Aden, or simply fishing. Previously, analysts made these judgements with years of learning (read: “software” upgrades): they became intimate familiar with local dhow traffic, came to know local meteorological trends, &c. After all of those years observing, they could make better guesses as to a boat’s true intentions.

As information shifts from analog to digital (and as we gain the ability to deeply contextualize information in real and digital space), data will transition from something that enables military operations, to something closer to its foundation. No longer will one person in a dark room thousands of miles away watch a few ships and pass notes to a “watch captain,” instead, systems with infinite attention span will catalogue every ship, cross-referencing movements with cell phone records, criminal and commercial incidents, shipping manifests, and more. Anti-piracy task force commanders will simply query, “Who is currently trafficking illegal weapons at [the target location],” and be provided with a rough estimate of individuals, their vessels, pattern of life, and optimal intercept locations. Instant observe, instant orient. What remains? Decision and action.

Combat Math Of The Future

The diagram below is an abstract illustration of Moore’s Law, at left. Each sequential blue circle represents a doubling (starting from 1). After approximately twenty doublings, we get to just over 1,000,000. In terms of Moore’s Law, this is transistors per standard chip (roughly analogous to processing power); the space between each circle represents 18 months. The chart illustrates 30 years of progress. From the perspective of 20 doublings, the first 15 doublings look as though nothing is changing (at scale).

DISCLAIMER: This is an imperfect chart, ginned up simply for sake of illustration.

On the right, I have simply reversed the scheme for purpose of illustration, in what I’m calling “Boyd’s Law.” I am using the inverse graph Moore’s Law because the inherent assertion is that being an OODA-loop increment faster than a competitor essentially doubles lethality. This is an imperfect concept, but adequate for illustration: the difference between increments is existential (victorious vs. defeated). Think “Bullet Time” in The Matrix.

Militaries have moved from right to left along the x-axis of Boyd’s Law with better training and better equipment. But to get to that “15th doubling” phase, we will need to upgrade the “observe/orient” processors from biological to digital. We can have the best tools for deciding (i.e. the best training in the world), and the best tools for acting in the world (i.e. the best weapons), but if we maintain our analog tools for observing and orienting we will cede these advantages to individuals or groups that press ahead with pursuing digital technologies for observation and orientation.

Within the next few years, the cost to observe anything, or to orient to anything will approach zero. This means that all actors will be achieve measurable shortening of their OODA-loop without access to cutting-edge military technology and training — they’ll be able to do it from a smartphone. ISIS is already threatening US military spouses over Twitter. What happens when such technologies metasticize into IED-enabled 3D-printed drones? The answer is: military capabilities that seem to rival ours.

Think this stuff is science fiction?

Within five years, Skybox Imaging will be able to provide near-real-time commercial imagery of any location across the world. The implications are manifold. Hedge funds have been using satellite imagery for over half a decade to predict markets. As these capabilities decrease in cost (and as computational abilities move them into the realm of the automatable), vast amounts of meaning will soon become readily knowable.

Machine learning and the increasing datafication of realspace will mean that technologies like Blue Force Tracker will soon give combatants the ability to know the location, capability, and even likely intentions of nearly every object within the battle-space.

Imagine fusing of nearly every electronic aspect in a combat zone, and the placing a reasoning layer on top of it, yielding a dynamic battle-space map. A patrolling platoon sergeant with heads-up-displays receives push notifications when military-age-males with connections to known insurgents approach from tactically sensitive angles; the Sergeant deploys micro-UAV’s that transmit their feeds directly to a cloud architecture that is running complex body-language and neuro-linguistic and translation analysis algorithms to discern the incoming individuals’ intentions. Simultaneously, programs running on the same cloud analyze geospacial aspects of the terrain and buildings, and provide recommendations about optimal defensive positions to the Sergeant, who then issues a simple digital command, and tactical orders are pushed directly out to every soldier on the patrol, so they can silently move into their new positions and await further orders.

In fact, many of these concepts are already happening at shopping malls across the United States and elsewhere, with cutting-edge advertising technology already weaving together complex pattern-of-life data for nearly every connected person on the planet. Walking down the street, posting something on Facebook about how you want a new pair of pants? The Gap will place advertisements in your newsfeed letting you know that the bluejeans of your dreams are 30% off — and the nearest store is just two blocks up on your left. Target already knows before your family when you’re pregnant.

The implications of avoiding advanced technologies and remaining the preeminent industrial military in the world are grave.

Silicon Valley: Victory With Cartography

We now come full circle. Why should Department of Defense open an embassy in Silicon Valley? Because the companies in Northern California (and other places as well, but first and foremost in and around the Bay Area) are at the leading edge of taking information from the analog realm, and not only turning it in to digital information, but turning it into relational information. And that is because these companies have been building digital maps for the past 30 years.

The true art of cartography is taking “realspace” information and making it knowable to a non-present entity (the person reading the map). This is what Facebook, and other technology companies, are actual doing — and their true power. Never mind the seven friends who “liked” your photo of Satoshi the Shiba Inu wearing a pint-sized incredible Hulk costume, that upload means that the network now knows where you uploaded the photo from, what camera you used, who else was in the photo, who liked it, and so on. Every time you “friend” someone, you are creating a small detial on an impossibly large digital map that is approximating reality in 1's and 0's. Knowledge of your friends transitioned from something known only to you and a few others, to something that is now, thanks to adaptable programming interfaces (“Log In With Facebook” et. al.), knowable to others. This data-as-environment coupled with advanced machine learning is how we will need to build and execute defense technology in the future. And the people doing it best right now—the people we need to be learning from—take their coffee at Philz, their hummos from Oren’s, and half-day winter Fridays in order to get a few runs in on the California side of Lake Tahoe.

The technologies that our military-industrial complex built over the past half-century, from long-range precision strike to C4ISR, are philosophical analogues of the hoplite. They enable us to see, farther. Strike, faster. This is all well and good, but it relies on information-as-inject, not information-as-context. Better weapons have meant the difference between life and death. But better “stuff” won’t be enough. Better software will be the key.

Consider the differences that even fifteen years ago would have been unimaginable: an atlas vs. Google Maps, a compact disc vs. Spotify, CBS radio and Netflix. This trend (of analog information becoming exponentially more valuable once turned into 1's and 0's) is what venture capitalist Marc Andreessen referred to when he famously wrote that “Software is eating the world.” It is coming to national security.

Two hundred and forty years ago, Joseph Warren—who acted as both irregular militia leader and intelligence collector—passed information from a highly placed intelligence source to Paul Revere, who carried that message (That the British intended to destroy Colonial arms stockpiles at Concord, via Lexington, and would arrive via water) to be passed to other colonists. The process of transferring several bytes worth of information over less than ten miles required something on the order of one hundred man-hours: first in extracting the information, and likely intentions of the British, then transmitting that information to Joseph Warren, who then passed it to Paul Revere, who then passed it on to the American revolutionaries outside Boston.

As information becomes contextualized, the software we run on top of our defense-centric data infrastructures will be the definitive tool in military operations. Building deep cultural connections to the software industry now will help the defense establishment prepare for future conflicts that look nothing like anything we have ever seen. Today, Joseph Warren would pull up a remote video feed, stare at it for a few hours, consider what British activity when conducting land- or amphibious operations looked like in the past, take a guess as British intentions to attack by by land or by sea, and then pass that information to his courier (Revere), who would then pass it on to the revolutionary commanders. But in the future, it will look something more like this:

Copyright 2015, Google.

In other words, software is eating the war.

One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.