Today there are over 15,000 nuclear weapons in the world, spread over nine countries. Many of these warheads are kept on hair-trigger alert, fully primed and ready to fire on command. Or by mistake.

Why the constant alertness? To guard against surprise attack. The flight time of a submarine-launched missile targeting Washington DC is less than 15 minutes. During that time warning systems must detect the launch and alert the leadership. During those few minutes the president must decide if the threat is valid and respond. That's why nuclear protocols require quick reaction times with no room for doubt or error. Russia's president is under the same time constraints as is the premier of China. Other world leaders with nuclear arsenals work under similar constraints.

Once orders are given, command signals go out to airfields, missile bases and submarines. These military assets might themselves be destroyed in a few minutes, so they must react the moment they get the command. This system is made possible by a complex distributed early-warning and command-and-control network.

Due to the extreme alert level and overall complexity of the system, there have been many false alarms and close calls over the years. These have taken many forms.

Given our simulation involves a Russian attack, here are two example close-calls involving their forces:

On September 26, 1983 Russian defenses detected five incoming ICBMs. In fact these ICBMs were computer-generated phantoms, but the Russians didn't know that. Alarms went off in a command bunker near Moscow. This bunker was the last stop. If it validated the signal then an emergency We-Are-Being attacked message would have gone straight to the Kremlin. The pressure was intense and the system was positive in its assessment of the ICBMs. However the commander of the bunker decided - by instinct - that it was a false alarm. He over-ruled the computers. Had that not happened there was only last step before Russia launched a counter-attack.

Another example. In 1995 Norway launched a meteorological rocket. That was a routine event and the Norwegians informed everyone. Unfortunately not everyone got the memo. The Russians were surprised and their air-defense went on full alert. This meteorological rocket looked like an attack by a US submarine. Launch orders actually reached the desk of the Russian president. Luckily the Norwegian rocket flamed out over the sea and the Russians realized their error.

Lest we think only Russians have problems with nukes, we should briefly note that the United States has had its share of incidents as well. Consider the notorious Damascus Incident. Never was there a better example of Murphy's Law.

A worker was doing routine maintenance at an ICBM silo in Arkansas. He dropped his wrench. The wrench punctured the missile's fuel tank. Fires started and one thing led to another until eventually the missile tried to launch itself. Fortunately only the second stage ignited, which detonated the rocket shortly after liftoff. In this fashion Arkansas failed to incinerate Moscow. However the force of the blast also blew the missile's 9 megaton warhead straight into the air. It landed just outside the base's front gate. Had it exploded it would have irradiated a large part of Dixie. Worse, its impact could easily have led US defenses to believe that an attack was underway.

A dropped wrench, a computer error, some mislaid paperwork - on a planet packed with 15,000 warheads wired to hair-trigger computer systems, it doesn't take much to bring on doomsday.

But as history amply demonstrates, it doesn't necessarily take an accident to start a war. Nine countries have the means to destroy the world. That's an unstable arrangement. Every state must watch every other state, all the time and with never a miscommunication. In such an environment it's easy for a miscalculation or a mistake to get out of hand. Or for an leader to simply make a bad call. The U.S. president, for instance, can order a nuclear attack at any time and for any reason. There are no safeguards. Therefore it's imperative the president be psychological and emotionally stable, as the fate of the planet rests in that person's hands. The same goes for the leaders of the other nuclear powers.

So we live in a world filled with nuclear weapons primed to fire at a moment's notice. Our physical survival demands that no person and no computer ever makes an error. That means our lives are constantly on a knife-edge, one turn of the key away from oblivion. All it takes is bad leadership or a good leader having a bad day. Or a series of unlucky or accidental events. Or simply a random psychopath in the wrong place at the wrong time.

But what would an attack actually look like?

It's hard to comprehend the brutal scale and destructiveness of a modern weapons. To help, let's look at some visualizations.

The "Little Boy" bomb that destroyed Hiroshima had a yield of 15 kilotons (15kt). In other words, the bomb's explosive power was equal to 15 thousand tons of TNT. This single weapon killed over 140,000 people. Partly from the force of the explosion itself, but mainly from the severe burns that tens of thousands of people suffered.

There have been hundreds of nuclear tests over the years. Therefore we have a good mathematical handle on how to calculate their direct effects. For this simulation I use the well-known scaling equations found here.

So how destructive are these weapons? Let's answer that by attacking Los Angeles with Little Boy. For such urban areas it's optimal to air-burst the weapon high above the target. Exploding the weapon at a suitable altitude maximizes the weapon's destructive reach. In contrast, surface-bursts destroy more difficult targets. These include military structures and command bunkers, but also naturally hardened structures such as airport runways and dams. A surface-burst digs a crater, while an air-burst channels that energy towards a greater area of destruction.

So let's drop Little Boy on Los Angeles. We detonate it 2,000 feet over the target. The resulting blast pattern looks like this:

Four nested circles centered on ground zero illustrate the bomb's key effects. The inner red circle is the 20 psi zone. Here the blast wave exceeds 20 pounds per square inch - a truly enormous pressure. 20 psi is enough to take down a skyscraper. Lethal gamma radiation also flood the area while temperatures spike to many thousands of degrees. You might as well be on the surface of the sun. This is the overkill zone.

For this attack the 20 psi circle extends out 0.42 miles from ground zero. No one survives.

The 2nd ring marks the 5 psi area. That pressure blows out buildings instead of vaporizing them. Temperatures and radiation decline, although gamma radiation is still lethal in the inner parts of the ring. Some people survive. However almost all these survivors are severely injured by burns, shrapnel and radiation. In Los Angeles this area extends 1.12 miles from ground zero.

The 3rd ring marks the lethal thermal area. Anyone exposed to the blast suffers immediate 3rd-degree burns. Without prompt medical attention such burns are fatal. The blast wave also remains powerful enough to damage wooden structures and blow out windows. That kills half the population. This ring extends 1.29 miles from ground zero.

The last ring is the non-lethal burn area. Here the blast lessens to a mere hurricane and the main threat is the pulse of heat. Anyone caught in the open suffers 1st or 2nd-degree burns. Some people are also blinded, their retinas scarred by the incredible burst of light. There are many such injuries and deaths, from burns and shrapnel. This ring extends 2.5 miles from ground zero.

Overall this bomb directly affects 20 square miles of LA.

Blast, radiation and heat are just the first effects. Afterwards fires ignite throughout the target area. These fires spread and merge. Under proper conditions such fires self-feed, becoming firestorms that suck in oxygen from surrounding areas and burn everything that can possibly burn. People who luckily survived the bomb's detonation are then incinerated in the resulting inferno.

All of these effects add up to a catastrophe. Hundreds of thousands of people die.

Now let's consider modern weapons. These bigger bombs make Little Boy look like a firecracker. For example, the American B83 has a yield of 1.2 megaons - 80 times more powerful than what we just dropped on LA. But even that weapon isn't anywhere near the largest. The Chinese Deng-Feng 5 delivers 5Mt. And the Russians have a 20Mt weapon (currently inactive, thankfully).

Here's Little Boy versus LA again, but with the map zoomed out:

Holding that zoom constant, let's now drop a Deng-Feng on the same unlucky spot:

Because of the increased yield we detonated this weapon at 16,000 feet.

We see the same effects as before, except the affected areas are now enormous. The central 20 psi no-survivor zone extends 2.78 miles from ground zero and the 5 psi ring extends for 7.34 miles. People burn and are blinded 20 miles from ground zero - all the way out into the Pacific Ocean. Gigantic firestorms engulf much of the LA basin. The total blast area exceeds 1,200 square miles. Millions die.

All this from a single bomb. Humanity has made great progress since Little Boy.

If you survived the attack and firestorm, you would then have to worry about fallout.

Fallout is radioactive ground material blasted into the air by the mushroom cloud. These irradiated particles are then carried onwards by the high-altitude prevailing winds, generating a lethal plume. Eventually this plume drifts down to earth. The effects of that can range from the inconsequential to the lethal, based on intensity and length of exposure.

You can measure fallout in several ways, depending on what interests you. Given we're interested in the biological damage that radiation causes, we'll use the sievert (Sv). This unit is generally expressed in milli-sieverts (mSv) or micro-sieverts (uSv). Geiger-counters typically measure in uSv per hour.

People respond differently to radiation. What makes one person slightly sick to their stomach might outright kill another. But as a basic guide, a dose of 500mSv is the first danger point. At that dosage signs of radiation sickness appear: headaches, nausea and vomiting. You won't die, although you might prefer it.

Bigger doses induce worse symptoms. A dosage above 1,000mSv brings the first fatalities and a dosage above 3,000mSv leads to a 50% death rate. Anything above 10,000mSv is invariably lethal and gruesome.

A single-dose of radiation is more deadly than the same dose spread out over time. For instance, imagine you're considering a nice morning walk through a fallout zone. Your Geiger counter registers 3,000mSv in this area. That means that if your walk lasts one hour you would get a total dosage of 3,000mSv. That's enough to kill you. In contrast, if your Geiger counter registered 3mSv, you could spend 1,000 hours outside before getting the full 3,000mSv dose. In that case your chances of survival would be higher given the dosage accumulated more slowly.

To get a sense of scale, yearly exposure in the United States averages around 6mSv. A dental X-Ray is .01mSv and eating a banana costs you .0001mSv. Years after its meltdown, the Fukushima reactor core still burned at 9,700mSv (9.7sV). Today's health tip: don't step inside a damaged nuclear reactor core.

So how much radiation would a bomb produce? And where would it fall?

Compared to measuring the effects of bomb blast and heat, this calculation isn't so easy. There are a variety of complicating factors.

First, the physics of the weapon regulates the amount of radiation produced. Fission produces more radiation than fusion. Modern weapons are fusion-based, however their ignition requires fission. They are in effect two bombs in one, where the energy released by the fission bomb drives the resulting fusion reaction. Therefore the amount of radiation produced depends on the level of fission in the weapon. This fission fraction varies across weapon designs.

The amount and intensity also varies due to the mode of targeting. Because the fireball from an air-burst doesn't reach the ground, relatively little surface material gets thrown into the stratosphere. For that reason, an air-burst on a non-military target won't generate much fallout. A surface-burst is a different matter. It interacts with the ground by design, throwing up irradiated soil. Attacks on missiles sites and command posts can generate plumes encompassing enormous areas. Airports and other infrastructure (ports, dams, railway junctions) can get hit with surface-bursts as well, generating similar amounts of fallout.

Wind is the biggest external factor. Ground-level winds will carry some fallout locally. However this is a relatively minor effect, particularly given that area will have already suffered from the blast, heat and intense localized gamma radiation. The fallout plume mainly depends on high-altitude winds. These pick up the cloud and carry it long distances. In fact, they will carry some level of fallout around the planet. These high-altitude winds blow in certain prevailing directions, but they also seasonally shift and interact with other weather patterns. These shifting winds make fallout calculations imprecise.

There are further complexities. For instance, surface composition affects the type of material and level of radiation. A surface-burst in a sandy desert will behave differently than one in an urban center. Local weather conditions may bring some fallout down as rain. Other meteorological effects generate hot spots. These can crop up anywhere, bringing death down to an area that wouldn't otherwise be fatally affected by the fallout.

Then there are the target outliers. What happens if a nuclear reactor gets hit by a surface-burst? Reactors store an immense amount of radioactivity. That's why nuclear-disaster sites such as Chernobyl and Fukushima will remain uninhabitable for a long time. A nuclear strike on a reactor would unleash this radioactivity and spray it over an immense area of the country.

For all these reasons, it's not possible to plot fallout with great confidence. The best we can do is an approximation.

Therefore I chose a simple and conservative simulation model. I ignored the possibility of target outliers (such as reactors) and I chose the most-prevalent wind-patterns for each target. To map the fallout plumes themselves I used Carl Miller's Simplified Fallout Scaling System. Based on atmospheric test data from the 1950s, this provides a simple average way to map the fallout contours. I chose to map the area exceeding 10mSv (10,000uSv). At that base rate, 2 days of exposure brings on the start of radiation sickness. After 2 weeks of exposure people begin to die.

As you move deeper into a plume the radiation and the threat level rises. One can use Miller's model to plot these higher radiation levels at more detail, but this implies a greater precision than the data warrants. Due to that fact, I chose to keep the calculations modest and limit them to 10mSv/hour.

Let's return to LA. Here's what a 5Mt Deng-Feng surface-burst does to it:

The fallout plume extends 426 miles and irradiates 30,000 square miles. Lethal radiation reaches all the way to Phoenix.

Again, one shouldn't put total confidence in these calculations. Per the above discussion, many variables will influence the result. That said, this is probably a good approximation. Certainly - should you find yourself anywhere to the east of LA following a nuclear strike - it's a good idea to take shelter.

How long do you need to hide in your custom radiation-shielded basement? Well, here at least we have some relative good news: fallout rapidly decays. This decay rate is calculated by the rule of seven: after the initial explosion radiation falls by one magnitude for every power of seven of hours elapsed. This means that 7 hours after the explosion fallout radiation will be 1/10 as high and after 49 hours (7 x 7) it will be only 1/100 as intense. Said another way, in about 2 days radiation readings will be 1% of their original levels.

Given that, barring extreme levels of radioactivity, most areas will be habitable again after a couple of weeks. Granted, every other living thing in the area might be dead, but people themselves can survive should they shield themselves for the required length of time.

But how would nuclear weapons be used in practice? An attack could take many forms. Terrorists might explode a primitive radiological bomb, or thousands of thermonuclear warheads might vaporize large chunks of the planet. Outcomes range from a few thousand dead to everyone dead. That's quite a spectrum.

I chose a middle path, modeling an attack that hits a significant number of targets but doesn't involve all-out war using all weapons. I used Russian warheads in this attack, applying their known weapon yields to the targets most suitable for them. The result is a conservative mainstream scenario.

How many warheads would be required in such an attack? And where would those warheads land?

There's a rich literature on nuclear targeting strategy. People have been talking about this topic for decades. But during all this time - not surprisingly - the targeting lists used by governments have been kept top-secret. This changed in 2015 when the US released a document detailing how it would have attacked the Soviet Union early in the Cold War. It makes for bloody and unsettling reading.

Historically there have been two main nuclear strategies: counter-force and counter-value. With counter-force, the idea is to attack only military and governmental targets. The goal is to destroy the enemy's ability to wage war versus destroying the enemy himself. This doctrine stands in contrast to counter-value, where the strategy is to defeat the enemy by simply annihilating all their cities.

However, as the document makes plain, reality is not kind to these theoretical distinctions. Many military sites are close to cities. Hitting them effectively means hitting the city itself. Further, the definition of a military site is broader than you might think. For instance, a civilian airport can be used by miliary planes. Similarly any port can host a warship as well as a cargo ship. That means all airports and seaports are legitimate primary military targets.

"Government targets" is another broad brush. That translates into hitting not just every bunker, but also every provincial capital and significant government office. Finally, destroying war-fighting ability means destroying economic capability. If you take that logic to its limit it soon becomes necessary to pulverize every significant urban center. And that's exactly what US war planners did.

In the funhouse mirrors world of nuclear strategy, this is perfectly logical. War is, after all, the mechanism by which you defeat your enemy. Incinerating the enemy population is a straightforward path to that goal. Ghenhis Khan would have easily understood nuclear strategy.

Russian and Chinese nuclear plans remain state secrets, but there's no reason to think they are much different than US plans. It isn't likely that Russian and Chinese nuclear forces are targeted to land somewhere in the ocean out of moral scruples. Further, given the arc of history, there isn't a lot of reason to be optimistic that good intentions and mercy will figure prominently in any nuclear exchange. More likely a "use them or lose them" mentality will take hold. And this will be magnified by the chaos and pure terror that will be the hallmarks of any nuclear exchange.

So I chose a strategy modelled on the unclassified documents. Military and command targets are hit in a first wave, followed by waves of attacks on government targets (such as state capitals) and economic infrastructure. The war lasts two hours.

In those 120 minutes every significant military base in the United States is destroyed, as are the largest 100 cities and every state capital. Significant seaports, airports, oil infrastructure and hydroelectric facilities are also struck, per the goal of destroying economic infrastructure.

I chose to be conservative with regard to other tactics. For example, I didn't hit nuclear reactors. A surface-burst on such a reactor would create a monstrous fallout plume that would dwarf anything a mere hydrogen bomb could do. There are 66 such reactors in this country. An attack on them would coat the continent with radioactivity. No one would do that, would they?

I also chose to be modest with regard to after-effects. My fallout estimates are purposively limited. Among other things, I made no attempt to estimate systemic effects from multiple surface-bursts in one spot. If there are 10 100kt surface-bursts (which would happen at some missile silo sites), I model that resulting fallout as resulting from just a single 100kt burst. I also made no attempt to model the levels of radioactivity in the plume itself. I felt such calculations gave the model a false precision. In reality the best we can do with fallout is an approximation. I wanted the simulation to reflect that fact.

Similarly I make no provision for likely outcomes such as ecological collapse, crop failure, social breakdown, runaway disease and mass famine. Netting it out, my simulation might be considered a happy and optimistic best-case scenario.

It's time to mess with LA again. Here's what it looks like after the attack:

This is a simulated Russian attack. So the good news is that the 5Mt Chinese Deng-Feng isn't used. The bad news is that about two dozen other Russian warheads are used instead. That's enough to destroy every economic center in the region, including the major container port in Long Beach. All military bases and major airports are wiped out as well.

To accomplish this the LA basin gets tiled with nuclear explosions. Very few areas escape the direct effects of the blasts and none escapes the overall consequences. Even areas outside the burn rings suffer from the resulting firestorms and radiation.

In this simulation the same attack pattern repeats across the country. In total about 400 targets get hit. However, measured in other ways this is a very light attack. It requires less than 1100 warheads, which translates to about 15% of Russia's total nuclear capability. Things could go far worse.

Up until now humanity hasn't had to endure a full-scale nuclear war. This makes casualty estimates for such a war challenging, given there aren't any empirical models to draw upon. One can theorize, but as with fallout calculations, reality intrudes with complexities and unknowns. Perhaps the best we do is look at historical analogs, combine this with current circumstances, and extrapolate to get an approximate lower limit. And that brings us back to Japan and Little Boy.

Here's Little Boy over Hiroshima:

Hiroshima suffered 140,000 casualties out of a population of 340,000. That's roughly a 40% casualty rate. These numbers are often disputed. but that's a lower-bound.

Using that as a starting point, we first note that Little Boy affected less than half of the city. That is, most of the city was outside the outer ring of the explosion. Many people were able to retreat into those areas. Second, it's worth noting that help poured into Hiroshima immediately after the bombing. Outside the blast zone food and water supplies weren't disrupted. There were still hospitals standing and doctors working. This meant that tens of thousands of injured people were able to get some basic supplies and medical attention. No doubt this saved many lives.

Here's Hiroshima hit by a modern 500kt weapon:

Hiroshima is completely blanketed by blast, heat and radiation. The entire urban area is affected. There are no safe zones or places to retreat.

What's the casualty rate? Taking the Hiroshima data and then mapping it into the known lethality rates of modern weapons, an 80% casualty rate looks reasonable. That means 280,000 dead or wounded. Given the population would be within the burn zone, this is an extremely modest lower-limit, particularly given the fact that this probably isn't the only warhead detonating in the vicinity. Most likely there are multiple local targets and multiple interlocking target rings. Hiroshima is annihilated.

So 80% is our conservative base number. We can now apply this factor to our simulation. Every major urban region in the country suffers an attack equivalent to LA, as does any city over 250,000 people and every state capital. Roughly 75% of the US population lives in these areas. So if we take the 80% Hiroshima casualty rate and apply it to 75% of the population, we get a 60% overall casualty number.

This number doesn't account for casualties from purely military targets. Many of these targets are near cities anyway and so count in the overall totals. More remote bases have smaller numbers whose totals aren't impactful in any case. So we ignored these.

Fallout is a much bigger killer. No doubt millions die as large parts of the country are irradiated. However, given the vagaries of fallout calculations, I don't include calculations from this source. So this lower limit counts immediate bomb damage only and doesn't include subsequent fallout casualties.

For the simulation, I took the 80% factor and applied it on pro-rata basis as each target is hit. Military targets are usually smaller and these get struck first. Thus in the simulation you'll see casualty numbers ramp up slowly at first. Casualties then increase dramatically in the second hour of the war, as major cities are destroyed.

Rolling the numbers up, somewhat more than 180 million American are killed or injured in this war. Again, this is a extremely cautious lower limit that doesn't take fallout into account. Most of these casualties will be deaths, given there will be little medical resources for the millions of severely injured.

Survivors face a critically damaged and hostile environment. Water, food, and fuel supplies dwindle and then vanish. Millions of injured people suffer and die due to the lack of medical infrastructure. Forget about FEMA or aid of any sort, as every other part of the country (and possibly the world) is in the same straits. Absent a functioning government and society, survivors fend for themselves. Mass disease and starvation is likely. Public order probably collapses. The mortality rate in such a post-attack world is truly impossible to estimate.

In a world growing more crowded and more unstable, the existential threat of nuclear war increases with each passing day. Each time there is an international crisis or simply a mistake in command-and-control, humanity takes another gamble. One day the dice will come up with the wrong number. And then the scenario that I outlined above will come true - or something far worse.

However the world remains generally unaware of this danger. The prevailing attitude is that nuclear war is an impossibility or a risk that somehow faded away after the Cold War. Meanwhile the danger builds. Our species is standing at a cliff edge but doesn't even know it. That's a recipe for the final tragedy.