Interview with Alan Phillips on ending launch on warning

An Introduction to No Launch on Warning
Robin Collins interviewed  Dr. Alan Phillips in Hamilton, Ontario, June 2003

Note that the phrases “the paper” or “my paper” below refer to the following papers:

The first is “No Launch on Warning” by Alan Phillips, available at: or by email from the or from Physicians for Global Survival.

A shorter version was made in collaboration with Steven Starr, called “Change Launch on Warning”. It is on several web sites including the CNANW page:

We have made several revisions and the latest is called “Alter Launch on Warning” and it will be placed on the CNANW web site.

Q1. How might an accidental nuclear war happen?

There have been many mistakes in the operation of nuclear deterrence — “mishaps”, as I’ve called them in my paper “20 Mishaps That Might Have Started an Accidental Nuclear War”. One was when all communications from the strategic headquarters to the radar stations quit. This was due to a very bad piece of design by which all the routes went through one relay station. There was a fire there which put that relay station out of action. The general in charge thought this must be deliberate sabotage or a nuclear strike at the distant early warning stations. That went to a very high level of alert. Bombers were actually at the point of taking off but didn’t.

Q2. Do the kind of problems that cause such incidents happen on a regular basis?

No, I think regularity is definitely not what you get in these weird accidents. The whole system is very complicated and nobody knows where in it some defect will show up, like the independent phone lines all going through one relay station. Very irregular. The dangers now are probably different. That was in the days of bombers and they would take off and it would probably be hours before the bombers would arrive. Now it is minutes, with rockets, and the big danger is of the satellite and radar warnings giving a false warning. It’s difficult to see just how that might happen but radar errors have been among the weirdest causes of alerts. One time they want on a high alert because the long distance radar for the first time picked up signals reflected from the moon. As soon as they realized what the range was, it clearly wasn’t an attack, but this actually put them on a high alert in the very early days. What might make them do so now is different — a computer glitch with just coincidentally a satellite glitch. Satellites have picked up heat that people have interpreted as a launch when they were due to an oil rig fire or something like that.

Q3. What’s the most recent known incident that almost resulted in a nuclear missile launch?

There’s a famous one, but the word “known” in your question is an important one, because we assume most silly accidents that happen to the military they keep quiet about – both to keep the public content, and to maintain the honour of their own unit, the fiction that their own unit never makes mistakes. So the known mishaps will probably be a small fraction of the mishaps that actually occurred. The last one that I know about, and that the general public knows about if they choose to read the stuff, was in 1995 on a day in January. The Norwegians, in cooperation with an American research team, were launching a rocket to explore the upper atmosphere regarding the aurora. They had told every country in the vicinity that they were launching this great rocket, which was an American rocket, on such-and-such a day at such-and-such a time. That message somehow didn’t get to the Russian military although it got to everyone else. When the Russian early warning people saw this rocket go up, it was interpreted as a possible attack on Moscow. The nuclear suitcase was handed to the president (who was then Yeltsin), switched on, ready to go. They had about eight minutes to decide if this was a real attack, if they were to “Launch on Warning”, that is to say, launch before any detonation. It was a very anxious few minutes. Fortunately, the radar people interpreted the track correctly and saw that it was not going to land in Russia. They still didn’t know it wasn’t a nuclear bomb.

A thing that worried me about that episode was: suppose the guidance mechanism of the rocket had malfunctioned, and actually directed it towards Moscow. Then it seems “the balloon” would have gone up: they would have launched their missiles at the United States. And that would most certainly have started a full scale nuclear war.

So the whole concept is very dangerous – launch on warning, launch before an actual detonation occurs.

Q4. How likely is an accidental nuclear war to happen… as compared to an intentional nuclear war?

At the present time, I think an accidental war is far more likely than an intentional one. An intentional one could happen I suppose if the Russian situation became desperate and the head of state really lost his wits as a result of the stresses there. It is conceivable that they might start one. It seems to me almost inconceivable that the Americans would start a war and obliterate Russia, because of world opinion. It might be that they might wish to, still; but at present they are trying to maintain friendly relations so I think an intentional start of a war is most unlikely. It’s impossible to estimate how likely it is for an accidental war to occur but the mechanism for it happening is right there with the reliance on early warning by satellites and by radar, with the intention to launch retaliation even before the attack arrives. That’s the big danger point. It could happen. It hasn’t happened for 40 years while this system has been in operation. It seems rather a miracle that it hasn’t happened already.

Q5. Can we calculate the odds of an accidental nuclear war happening?

No. The short answer to that is simply that we can’t. The fact that it hasn’t happened in 40 years is remarkable, and some people, including General Lee Butler, have said that it is miraculous. It does suggest, but it doesn’t prove, that the probability isn’t high. But you can’t calculate the actual probability of an entirely random unexpected type of accident that may be completely unforeseen. Plus usually there must be, surely, a few coincidences to cloud the situation so that the human judgment turns out to be wrong.

I’ve put in my papers a calculation that is purely an example. I’ve taken what I consider is a very conservative estimate of the possibility of a threat conference coming to the wrong conclusion — that is, concluding that a threat is real when in fact it’s not. We know that there have been thousands of threat conferences, and for one period when we had an actual count of the number of threat conferences. It turned out to be something over a hundred a year — maybe a few hundred a year. This was in some period late in the 1970s. I did my example calculation (which is not an attempt to calculate an actual probability) by taking one percent of all the threat conferences in a year to be wrong. Suppose just once in a year they might make a mistake, with a probability of only one in a hundred. So most years they would obviously make no mistake. Well if you take that risk of one percent of disaster per year for thirty years, it comes out to this: that the world had a 3 to 1 chance of surviving, a 1 in 4 chance of disaster. Now, that’s bad. It’s not so bad as an even chance, but it’s not the sort of thing one should accept. If you take a revolver and hold it to your head, and have one chamber loaded, as in Russian roulette, and you pull the trigger once, you have a one in six chance of death; you have five to one chance of surviving. That is safer than what they did, on my assumption which, as I said, I believe is conservative. Then you have the equal or greater risk of the Russians making the mistake, so that would be like two pulls on the trigger.

Q6. How does “No Launch on Warning” differ from “de-alerting”?

Yes, that’s a very important distinction. Briefly…the two are quite different.

Giving up the policy of launch on warning doesn’t require any de-alerting. The weapons and crews can be and should be just as alert as they are when they are going to do launch on warning. The only difference between “no launch on warning” (RLOAD, Retaliatory Launch Only After Detonation) and “launch on warning” is that the retaliation has to be delayed until at least one nuclear detonation has happened. But they have to be fully alert and ready to fire as soon as a detonation has happened. Now, I’m not saying this is a good thing. I’m not saying that deterrence and retaliation are acceptable, but that is the way the Russians and the Americans are set up now – if they are attacked, they will do a devastating retaliation. That’s not likely to change, but taking the risk of retaliating before a single weapon detonates is a real risk that could be avoided.

De-alerting is quite different and it is that there is an enforced delay between the decision to fire and the actual launch taking place. This can be done by a variety of methods of which the most certain is removing warheads from the delivery vehicles or rockets or whatever, and storing them a distance away so that it is physically impossible for them to launch without a delay of several hours. That would be excellent — it would completely abolish accidental war due to false warnings. It would greatly delay the start of a war if there were political reasons believed to be strong enough to start one. So it would be an excellent thing.

The fact that deterrence is believed to be essential by the governments and militaries on both sides, as long as there are nuclear bombs at all, means that it would be very difficult to arrange de-alerting. The whole systems in the two countries are different: the rockets have different launch programs, and different launchers; the warheads behave differently. It’s very difficult to see just how the timed delay from the order to launch to the actual launch could be made the same on both sides. If it were not the same, the military would argue that the side that could launch first might indeed launch first, and launch several salvos before the other side could get started, destroying their capacity to retaliate. If that were to happen — if that were believed to be possible by one side — then deterrence has failed. So you need equal times. That would take a great deal of expert consultation to arrange. People would doubt that it was done properly. If they indeed did manage to agree that there was a method, they would have to have a written convention to assure them that it would be so, and they would have to have inspections and verification the whole time. They would have to be certain that no one was cheating. This would at least involve several years of planning, and execution systems would have to be changed. Then it would require a legal document, perhaps a full scale treaty. A treaty would have to be passed by parliament on both sides, and they might just turn it down after years of hard work had made a feasible system. So for one thing, it may be impossible to get a feasible system for de-alerting. And if they did get a feasible system, a whim of one side’s politics might prevent it being put into operation.

RLOAD is simply a policy change which would require minimal changes to the systems — just changes in the standing orders and in the procedure for a launch, so that the launch took place a few minutes after the decision — which would be a decision made while the rockets are in flight — to launch retaliation if, and only if, a nuclear detonation was detected at the expected time of arrival.

Q7. Is “RLOAD” easier to implement in terms of existing deterrence doctrine, then?

I could take your question two ways. One is “would the technical changes be easier?” and the other “would the effect on deterrence be less: which would leave deterrence more fully in operation?” [Q: If you would answer it from both perspectives, then maybe we could go over some ground.]

I would say in both of those senses, it’s much easier to change the policy from “launch on warning” to “no launch on warning”, than to “de-alert” in an acceptable way. The sort of changes that would be necessary would be to alter the sequence of the firing routine so that it stopped when everything was ready to go except the key finally being turned. The key would be turned only when the bomb alarm signal came through. The bomb alarm signals would have to be rerouted so that that they went not only to the command centres but to the silos themselves in case a first salvo destroyed the command centres. Not very likely, but possible. One of the plans to overcome deterrence was to make what was known as a “disarming first strike” and put the enemy’s weapons completely out of action. The first salvos of a nuclear first strike would be directed at the command and control centres and at the weapon silos. That would be the only way that “no launch on warning” could prevent retaliation. It would be if they could put everything out of action in one go. That is hardly conceivable because the silos are spread over thousands of kilometres in each country so that both the launch site and the target each have a variation of the order of one or a few thousand kilometres; and there are several command centres. The rockets when they are in full flight travel at eight kilometres a second, so it takes a few minutes to go a thousand kilometres. They would have to get this simultaneous to all the launch sites and all command centres. They would all have to arrive at their targets within a few seconds — within the time it take for the final turn of the key, which is only seconds. To get the salvo landing simultaneously on all the targets, the different rockets would have to be fired at different times, strictly coordinated with one another, over a period of some two or three minutes. Any errors in the time of launch would make an equal error in the time of arrival. It’s most unlikely that all that could happen perfectly, and again the actual time of flight could not be predicted perfectly so the head of state on the presumed attacking side could hardly be sure that her first salvo could destroy all the retaliatory capacity. I should mention here that we believe that there are something over a thousand warheads on each side available for launch on warning. A hundred warheads landing on their targets has been deemed the number which would put a country completely out of action. For a head of state to be confident that nearly all the enemy silos would be put out of action isn’t a sane piece of optimism. Only an insane head of state could do it. So retaliation is not significantly affected. Therefore deterrence is not significantly affected.

As regards de-alerting, I would say that deterrence might be severely hampered because it is so difficult to ensure that everything would be delayed the same length of time; and it seems to me there is considerable possibility of cheating. Of course it doesn’t apply at all to a country like Israel because it has no nuclear-armed enemies at present. Between India and Pakistan the flight time is so short that launch on warning is hardly feasible.

Q8: Do you think militaries mistakenly believe that abandonment of launch on warning will affect their deterrence capability?

I can’t answer that question directly, but there are a lot of things that bear on it. One is that nobody yet, until my papers, has made the point that “no launch on warning” doesn’t need verification and doesn’t need to be symmetrical. The idea of getting symmetry in that might be a problem; the idea of getting verification might be a big problem because there might be secret ways of giving orders that couldn’t easily be verified. So there might be a problem there, but the main point is that nobody has asked them to do “no launch on warning” that isn’t symmetrical. That may well be the big worry. It’s a pity because the people and bodies that have considered this, like the Brookings Institute and Canberra Commission, Bruce Blair himself with his colleagues at the Centre for Defense Information — they have all said “with reciprocity”. My big point is that it doesn’t need reciprocity. The military — and here it could be fixed modes of thinking, and the inertia of a big organization — were given the task of making a retaliatory system that would work, that the enemy would see was guaranteed to work, so that the enemy knew that if they fired rockets at the United States, they would receive retaliation back that would be so big as to be utterly unacceptable. So they would never attack. Making retaliation certain was their big task and they had to do it right; and with military thoroughness they did it right.

The risks of launch on warning were mentioned in discussions with political advisors and civil servants in the 1960s and I’ve quoted some of that in my first paper on the subject. But this discussion didn’t seem to affect the military who had their task of making retaliation appear to the enemy to be inevitable … and they did it. If they could be persuaded that NO L-o-W could safely be unsymmetrical, that it could be done unilaterally, they might not resist it so much. Their method of thinking that retaliation is so important — is in fact vital to the safety of their country — makes them very unwilling to do anything which even might diminish deterrence. This is a curious gap in the thinking of some military. An example of this is Bruce Blair quoting with apparent approval a Russian training manual for launch site people that it is essential to operate quickly and faultlessly in the event of a warning so that retaliation is launched before the enemy missiles arrive. They are told that “they have badly failed their country” if they fail to retaliate before the incoming missiles have put them out of action. The point is that once there are enemy rockets on the way, their country is finished. The retaliation doesn’t save their country from anything. It does destroy the enemy country. The idea that they have failed their country is not one of a failure that makes physical harm worse, it’s just one that insults their military attitude, their military responsibility. They have failed in what they took as a responsibility; but it has made no difference to their country. That is a curious blank point in the thinking that I have seen in writings from both sides.

Q9: Why do you say that reciprocity with “No Launch on Warning” is not important?

Because not responding to a false alarm does not put that side at any disadvantage.

If one side is at “no launch on warning”, then no false alarm from a supposed but non-existent attack from the other side, will trigger a response. Presumably both sides are roughly equal in their capacity for making mistakes, and so it approximately halves the risk. The system is a bit unsymmetrical now with the Russian submarine fleet in very doubtful operational readiness, and the fact that some of their radar stations have been taken over by the other countries that the Soviet Union divided into. So the situation is not entirely symmetrical but if the Americans gave up launch on warning, a thing like the Norwegian rocket event, working the other way around with the United States getting a warning, they wouldn’t launch until the time had passed when a detonation should have taken place, and no detonation would take place so they wouldn’t launch. And the world would be saved. If they made the change, the risk is certainly reduced. You can’t say nowadays that it would be halved because the risk on the Russian side may be greater.

Q10: Is the U.S. nuclear posture review telling the truth when it says the Americans are “not on hair trigger alert”?

I guess it’s probably a true statement intended to deceive. “Hair trigger alert” is not a technical term as far as I know. It’s not in the official dictionary of military terminology. I’m not that familiar with firearms, but in the old days of officer training corps before World War II — I was in that corps and we used Lee Enfield rifles which had a two-stage pull to the trigger. You were ready to fire, you were aiming and you pulled half-way — first pressure I think it was called. Then you had to make quite a firm pressure to actually fire the rifle. That was certainly not hair-trigger. My understanding of hair-trigger is that it is the kind of thing in a duck shoot where you don’t have that first pull at all. The trigger is very light and as soon as you see your target and get it in your sights, you want to fire instantly. That’s where hair-trigger comes in. There is no way that a launch of a nuclear rocket from the United States can be called “hair-trigger” because it involves a threat conference, a conference with the President, an order coming through to fire, and then procedure for firing which itself takes several seconds. In that sense, it is perfectly possible and not strictly lying for the military to say they are not on hair trigger alert. The peace movement and many important bodies like the Canberra Commission have used the term hair trigger alert, and I think it was ill-advised to bring that term into it at all. So there’s the situation: the military wants us to think they are not in danger of firing when they shouldn’t fire. So they can reasonably say they are not on hair trigger alert; but that’s not what the peace movement means they complain that the weapons are on hair trigger alert — they mean they might go off for a very slight cause.

Q11: How would you compare the U.S.-Russia stand-off with the nuclear weapons danger posed by India-Pakistan?

That’s a really important comparison to make. The probability of either one of the two pairs of adversaries making a mistake — I’m not going to estimate which are the more reliable, which are the more cautious, or anything like that — but the big difference is this. If India or Pakistan have a nuclear war, with the nuclear weapons in the tens or possibly in the low hundreds, they’re going to do enormous damage to those two countries. They could destroy most of the cities, perhaps all of the cities of Pakistan and a good portion of the cities of India — the most terrible human disaster to ever have been caused by peoples’ activities. But the effect on the rest of the world would be minor. There would be a certain increase in radioactive contamination all around but you wouldn’t notice it until the statistics were done — statistics of long-term cancer mortality, and so on. You could detect it with Geiger counters but it wouldn’t be an immediate disaster.

If the Russians and the Americans had a nuclear war the effects would be on the whole world. As well as destroying those two countries so that they were completely non-functional with millions of injured and dying survivors, and uninjured survivors dying of starvation (or of disease, because the medical facilities would be put out of action). And there would probably be nuclear winter. The nuclear winter issue was quite falsely sidelined by deliberate tactics of American administrations at the times that Carl Sagan published on the subject. But it is quite clear that this is a real risk and an expected result of the detonation of more than a few hundred nuclear weapons. Tens of thousands of weapons would be involved in a war between the United States and Russia, so there would be something like nuclear winter. How dense is a matter of speculation and calculation, and in Sagan’s last book it was made very clear that this would be severe on any reasonable assumption of what would happen in any war between Russia and the United States. Nuclear winter would be severe certainly over the whole of the Northern Hemisphere. That would mean no crops for a couple of years. All sorts of other effects like increased ultraviolet light and so on, but what would matter for the survival of the human race, the human species, would be whether that crossed the equator, and that is a thing which up to the last of my reading on the subject (which is the 1990 Sagan book) was uncertain from a meteorological point of view. But there was a feeling that quite probably it would extend to the Southern Hemisphere as well. There would almost certainly be frosts in the tropics that would kill all the indigenous vegetation. There are other crops brought by colonists — corn and wheat for example — that might survive. But basically their food supply would be gone and of course there would be no foreign aid coming in. There would be widespread famine, and there is the outside possibility that the survivors after a few generations would gradually waste away, and the species might be finished.

Q12: Why are people no longer talking about nuclear winter or its effects… or the conditions that most likely would cause it?

That’s a question of group psychology that I am not qualified to say much about. As regards nuclear winter, I believe there was a deliberate attempt falsely to minimize the atmospheric effects of nuclear war. The newspapers started talking about “nuclear autumn” because certain calculations of Sagan’s needed to be reduced in the severity of the results. But in his second book it was clear that there were other features of them that needed to be increased. The balance came out much about the same. Nuclear winter was widely discussed in the atmospheric science community. I don’t have any doubt that it is really an expected phenomenon, which was minimized very effectively by the American administration. It is interesting to note that in the war against Iraq in 1991, the First Gulf War, there were many oil wells damaged and a lot of smoke went into the atmosphere. It was visible to people who traveled to high altitudes by airplane who could see that the sky was different. The government laboratories were forbidden to publish anything on that subject. So I think that nuclear winter was successfully and falsely sidelined altogether. For the rest, I think because nuclear war between US and Russia hasn’t happened, people stopped thinking about it. But why that is so, and why I go on thinking about it, is psychology and is not clear to me.

Q13: Can you comment on the possible confusion over terminology, in this case the terms “launch on warning”, “launch under attack” and “launch after detonation”.

Yes, that has been a significant nuisance to me because I started off this work and wrote my first long paper with the view that “launch on warning” was a simple concept, which was a launch of retaliation during the flight of the incoming missiles — it seemed quite unambiguous and that was how the whole peace movement was speaking of it. “No launch on warning” meant simply that the military should not have that policy.

Previously, when the military had been asked about launch on warning, sometimes they replied “we’re not at launch on warning, we’re at launch under attack”. And we took it that “launch under attack” was a slightly deceptive term; but they tried to imply that it meant that the attack was definite, for example that they got warning from two independent systems. It still seemed to mean launch during the flight of the missiles. It also rather gave the impression that the launch would be when they knew the missiles had arrived and that they actually had been attacked, rather than that they were going to be attacked. So that confused it a little — perhaps intentionally by the military.

But then when we sent this paper to the military man I mentioned, he picked on that one difference and disregarded the whole purpose of the paper. It was a pity that he did that; it was a pity that I hadn’t clarified that. The clarification seemed from what he said — he answered very briefly and unhelpfully — that launch on warning could mean launch not on what they called “tactical warning”, which means the sight on the computer screen of “missiles” on the way, but launch on “strategic warning” which means including any information from satellites and so on, information from spies, information from satellites on troops movements around silos, and so on. Thus the launch might even be before enemy missiles were launched. That was what the military, or one group of them, seemed to be thinking about with the term launch on warning. That means we should be using “launch under attack”. There’s a difficulty here in that “launch under attack” is defined in a 1970s dictionary of military terms (which is on the web, an enormous dictionary with 100,000 entries) exactly as the peace movement has always thought of “launch on warning”: that is to say a launch while the missiles are in the sky, before any detonation. “Launch on warning” is not mentioned in that dictionary at all. So it was an understandable confusion — unfortunate that we hadn’t got those terms clear in first thinking about this, and all the writings that have been done on it over the years. But it seems possible that we ought to be using “launch under attack” when we talk to government, high civil service levels or military. Either we should specifically deal with “launch under attack” or we should be treating the two terms — launch on warning, and launch under attack, as synonymous.

Q: When the Canberra Commission referred to the situation, they referred to “launch on warning” in their brief paragraph.

They did.

Q: So presumably that term is shared by some officials and some military people.

I think “launch on warning” was used in the military, and this particular retired naval officer, at least, believed that “launch on warning” included a launch before the enemy had launched. But nonetheless the term is not in that dictionary. I searched it myself and I searched it carefully. I did not count all the 100,000 entries but I estimated by the numbers of screenfuls.

Q15: How much time passes between calling a threat conference…and a retaliatory missile being launched and landing on the target’s soil?

The last part of that is easy, and it’s between 25 and 35 minutes as far as I know. That is not secret; it’s just a matter of how fast these things fly to go that particular trajectory.

The missiles can probably be detected within the first five minutes of flight, but the radar probably a bit later. I imagine that the threat conference can be called between the five and ten minute mark from the time when they were launched. The threat conference then has a time perhaps of fifteen minutes before they get in touch with the President. They have already of course been trying to find where the President is, and get him on the phone. So then there will be only a very few minutes left for a conference with the President to obtain his authorisation that retaliation should be done. If it’s launch on warning, the retaliation has be done within the final five minutes at the latest. I believe the final launch steps, when the order to fire is given, would be a matter of a fair number of seconds. Probably not as long as a minute. I don’t know these figures — those are guesses. The total time of flight is well known — it is in the range of thirty minutes. That’s from Russia to the United States, or the other way. Submarines could be in the Atlantic or Pacific Ocean and could get within the range whereby that flight time could get down to fifteen or even ten minutes. Twenty is taken as more likely.

Q16: In any case we’re talking about a very brief period, relatively speaking, when serious risks are happening. To what extent could you say that “no launch on warning” reduces the tension at the threat conference stage such that saner heads may prevail, and so that the wrong decision isn’t made?

I think it would be of considerable benefit for both the threat conference and the final decision by the President, or whoever is making the decision, which in fact is not necessarily the President himself. The stress on that person and on the conference would be significantly less if the policy was “no launch on warning”, because they would know they were not starting the final war of the world on a false alarm. Whether it could be a right decision to retaliate at all is another question because their own country is already destroyed by the initial salvo. There is no actual benefit to anybody in destroying the other country as well. It is not possible to stop the salvo once it has started. And it’s a big one, with thousands of nuclear weapons.

Q17: There’s an ethical question here that is apparent, and I think the peace movement is often concerned that in any discussion of posture changes there might be an indication that a response with nuclear weapons is acceptable. How would you respond to any questions about “no launch on warning” in terms of its making “acceptable” a nuclear response?

“No launch on Warning” is simply a way to make an erroneous response impossible. It doesn’t in any way confirm the idea that deterrence is a good thing, nor that a nuclear response is a good thing. I start from the point that it might be possible to change this one item of government and military policy to make things safer without altering the whole basis that they have worked on for 50 years, which is deterrence. It would be far better if nuclear weapons were abolished, or to a lesser degree if deterrence was abolished. It would be better if there were a long delay before weapons could be launched at all. All of those things would be far better and far more fundamental than what I am proposing, which is simply a change of policy to avoid what happens to be the greatest danger, which is an accidental nuclear war. It isn’t intended to support the idea of deterrence (and in my paper I have said that, and I’ve said it more than once). I can see how people would look at it as accepting deterrence. I think we have to accept the fact that deterrence is the doctrine at present. It is most unlikely that we should be able to change that at a stroke. I believe it is possible for us to change “launch on warning” to “no launch on warning” at a stroke.

Q18: If implemented, would your proposal reduce reliance on deterrence?

It is a positive thing. It would be very encouraging to know that the military is open to considering a change to make the world safer from a nuclear war. I don’t think “RLOAD” would be a concrete step towards it. De-alerting would be a concrete step towards not relying on nuclear weapons. No launch on warning doesn’t make that step, but it would be the very important psychological step that the military and the governments of those two countries are willing to consider changes. That would be at the very least a great boost to the morale to all the people in the world who don’t want a nuclear war.

Q19: Is No Launch on Warning a necessary or likely first step towards de-alerting?

I don’t think it is necessarily a step towards anything, except a step that will have reduced a risk enormously – reduced the greatest risk virtually to zero. As far as that is concerned, it isn’t a concrete step towards anything else. But it is a first step well worth taking because it means the world is that much safer — a lot safer — for the years of work we have to do to get rid of nuclear weapons altogether, or even to get de-alerting in action.

Q20: Why, in the documentation that has come out in the peace community and in think tanks, has there been discussion by analysts of de-alerting, but very little discussion of abandoning Launch on Warning?

I don’t think they have clearly separated the two ideas. Either of them makes the launch a little later, but actually the two are quite different. No Launch on Warning doesn’t mean any reduction of the alert status, either of the missiles’ launching apparatus or the crews. Everybody has be entirely alert and able to get ready to fire within the twenty minutes or so that they know a missile is on the way. The firing difference is only a matter of a few minutes between the time they would fire just before the missiles arrive, or they wait until the first missile arrives and they receive the indication of an explosion. Then they fire. They’ve got to be fully alert. In those few minutes they’ve got to make everything ready for a launch and it is just the moment of launch that is altered.

De-alerting is quite a different thing. It is introducing a forced delay between the order to launch and the actual start of the process of launching. That of course couldn’t be compatible with “launch on warning” but it is not the same as the “no launch on warning” that I am proposing. De-alerting is much more complicated process because there has to be a fixed delay — usually a mechanical delay of some sort, like moving the warheads or moving components of the warheads.

With No Launch on Warning, if one side does it, the risk is approximately halved. If both sides do it, then that particular risk is down to zero. And it doesn’t depend on one side knowing what the other side has done; it doesn’t depend on trusting them; it’s just a benefit to both if even one side does it. It gives that side no disadvantage. So I think it only requires more careful thinking by the people concerned, who simply haven’t looked into the details of the matter, to see that those two things are quite different.

Q21: How easy would it be for RLOAD to be implemented?

I don’t think it would be difficult to be implemented compared to many of the difficult things that have been done in setting up the nuclear system altogether. It would be trivial compared to that. There would need to be a change in standing orders, a change in the launch routine that might be minor. The only material thing which I see that needs to be altered is that the bomb alarms which are at present in every city and around every big military base — the bomb alarms would have to signal direct to the launch crews as well as to the command posts in case all the command posts were destroyed in the first salvo. That would be an expense, but it wouldn’t be a significant expense compared with the expenses of nuclear weapons on the whole. It could be a simple thing, it could be written down on paper as to what has be done. The change in orders could be effected in a day perhaps. The change in the wiring might take a few weeks but it should be done in way less than a year.

Q22: If No Launch on Warning is unverifiable, how do we know it is being implemented, and does it matter?

The answer to that question is in the second part of it: it doesn’t matter. If we all knew that both sides were working at No Launch on Warning, then the world would feel more comfortable. But if both sides are at No Launch on Warning, then the world is in fact safer. I am proposing that this does not need verification, that there is no need to know. In fact, the tiny defect to deterrence that proponents of deterrence may maintain is significant, would, if anything, be less if the enemy did not know for sure that you were at RLOAD. If anything, verifying the RLOAD status would be a disadvantage rather than an advantage. The only advantage to it would be to make the other countries of the world feel more comfortable. That’s a real advantage but it doesn’t alter the chance of an accidental nuclear war happening or not happening, which is what I am concerned about.

Q23: Is it possible that RLOAD is already in place?

I wish!! I have often thought of a sort of dream that there is a very, very secret standing order on both sides that on no account, on no orders whatever, should a nuclear weapon ever be launched. It would be wonderful if there were such a standing order on both sides, and it would be entirely reasonable. I am afraid though that it is unlikely.

Q24: What’s the best way that the proposal that you’re making can become a reality if we can’t confirm that it’s happening, if military circles might not even want to say that they’re willing to change the status of their nuclear weapons? What’s the best way for people in the peace movement to ensure that this is in the think-tanks and military circles so that it might be considered?

The very last paragraph in our recent paper tries to deal with that. If in spite of all those arguments the military were to insist that they have to keep launch on Warning, it is for the head of state to weigh the danger of that, against the danger of an intended nuclear war. In fact he has to weigh the real danger of an accidental nuclear war which we know has nearly happened a number of times in the past (the Norwegian rocket incident being the most recent and one of the worst ones), against a slight hypothetical danger that deterrence might fail because of the change from the present LoW to a future RLOAD. It is only a very remote possibility that that would really affect deterrence. It would pretty well depend on the head of state of the attacking country being on the border of insanity. So there is a very remote chance that this would have any effect on deterrence. If the military don’t see that point, the head of state has the responsibility to decide. In my opinion every government in the world which has the responsibility for the safety of its own people, has to realize that freedom from a nuclear war between Russia and the United States is essential for the safety of the people of every country in the world. Therefore every government has the responsibility to press the United States and Russia to give up this dangerous posture of launch on warning.

Q25: What should the peace movement be doing on this issue?

I think there are two ways that might get a little progress towards RLOAD. One is that if the general public was made to realize that there is still a danger of an accidental nuclear war from a false warning, whatever the political situation may be, and this danger of a false warning could trigger launch on warning, and end the world. The other way is to work through government committees, civil service committees, army committees, to persuade the people who have the powers to change these things, to realize there is this great safety action which could be taken without altering their whole policy.

Q26: Is there a special role for Canada?

Of governments foreign to the United States, we probably have more influence on the United States than anyone; and of Western governments ours may well be more trusted by the Russians than any other. So it may be that Canada does have a special place.

Q27: Launch on Warning policy was intended to deter a disarming first strike. Did the militaries not think false warnings were a serious problem?

False warning was considered very carefully. There were huge arguments about it. We have found some of the documentation of that on the web. One letter I remember was from Robert McNamara when he was Secretary of Defense for Jack Kennedy. He said that LoW was so unsafe that never while he was Secretary of Defense and Kennedy was President, never would the US adopt LoW. Shortly after that there was a memo from the head of the Air Force to the President saying that they were ready to fire within 30 seconds with the new missiles which were just installed, which must mean they were thinking of Launch on Warning. There was that sort of argument going on. There was discussion about missile defence in the 60s and the people who were in favour of that were against LOW. It was regarded as a big ethical issue. Quite rightly. But it seems to me that the military put their weapons into an LoW posture regardless of the discussion that was going on. I don’t know that there has been any great change now. There have been gradual improvements in the accuracy of weapons and in the technology. There has been the improvement that now much of the early warning is “dual” — that is satellite as well as long-range radar. If both of them give the same signal, it is a much more sure warning than if only one does.

That brings the point up which is that there is greater danger at the present time — because the Russians have a good deal of their equipment out of action or in the hands of states which were part of the Soviet Union and not part of Russia. So they haven’t got the dual coverage on all the avenues that the attack could come along. That adds to the danger that they might be jumpy and apt to fire unnecessarily.

But really LoW has always been a danger. During the Cold War I suppose there was a serious possibility that one side would attempt a disarming first strike or simply a full scale extermination attack, whereas now that doesn’t seem probable. Perhaps the relative probabilities of disaster due to accident, and disaster due to evil intent, has changed.

Q28: Launch on Warning has presumably always been irresponsible policy. Has the political situation with the end of the Cold War made it more clear, or should it have made it more clear, that there is an opportunity to change that policy?

Yes, I think it has, Robin. People are not seriously concerned that the two sides are going to hit each other intentionally. That risk, whether correctly or not, has been seen as greatly diminished. It seems that during the Cold War people were half expecting the Russians to attack North America at any time. Whether that was a realistic risk or not I doubt. But certainly people don’t think of that now. If you are thinking of a risk, the risk that comes to mind is the risk of a war on an entirely false warning. It is more relevant now to say we can abolish that.

End of interview, June 2003, Hamilton Ontario.
Interview with Dr. Alan Phillips conducted by Robin Collins.

[updated January 21, 2007]