3CR
 Community Radio 855am

THE RADIO-ACTIVE SHOW

With Eric Miller, Ila Marks and Cherrie

Saturday at 10.00am

3rd July 1999

Hello and welcome to the Radioactive Show brought to you by the Sustainable Energy and Anti-Uranium Service. I'm Eric Miller. The Radioactive Show is a weekly program bringing you information on nuclear, peace and energy issues. On today's show we hear from Michael Kraig from BASIC - the British and American Security Information Council. He's talking at the RadioActive Alert Symposium for Sanity which was held in Old Parliament House in March.

His talk was on "The Bug in the Bomb and Nuclear Y2K". Michael explains about the problems associated with US and Russian early warning systems that alert each other that they're under a nuclear attack. He then goes on to say what the Y2K bug means for these highly computerised systems. Michael Kraig.

Michael Kraig: How is Y2K a danger to nuclear weapons, operations and arsenals and also just as importantly for this conference is how does this involve Australia?

I'll actually just start out by saying what we don't have to be afraid of with Y2K and that's missiles launching spontaneously out of their silos without human commands. Human commanders in Russia and the US must authorise a launch. Or someone below them. The problem is its growing fuzzier and fuzzier over the years because there's been more and more pre-delegation of launch authority. But some human has to authorise it and then the people in the launch control centres connected to the sites must turn the keys and input the targeting data. That being said, there is still plenty of danger and instability. In fact, the very systems that are most prone to Y2K errors are also the systems that have been most prone to failures and faults and accidents, computer and otherwise, throughout the cold war. And they're the same systems that involve Australia in the world wide global operations of the US.

To start out the US has three early warning satellites in space and these are there to pick up the image of rocket plumes and the purpose is to let the command post in the US know about Russian missile launches within half a minute to a minute of when they occur.

Russia has some similar systems although they've been degraded and I'll get to that subject shortly as well. US satellites are generally reliable, except for the Caviot - they have always been prone to shutdowns because of solar flares. Guess what. Ironically, there is something called Solar Cycle 63 coming up. There's going to be a fairly intense solar flare coming up to the year 2000 starting in December going through to January.

There's three of them - that's to cover the entire world, literally defence support program satellites or defence satellite programs, at 70 degrees East longitude that sends its information down to your facilities at Pine Gap or rather our combined facilities at Pine Gap and Nurrungar.

Okay, right now I'm just fleshing the idea of an early warning system before going more deeply into Y2K and the whole subject of computer errors and nuclear operations. There are radars on the ground deployed at various locations around the world. In the case of the US, around the perimeter of Russia for Russian and former Soviet systems. The radars see an incoming missile or re-entry vehicle in the later stages of the missile later on in it's flightpath. Satellites see the first launch half a minute to a minute. The US has radars employed in Greenland, the UK and Alaska to see missiles 6.5 to 10 minutes more into their flightpath. Then there's also one shown on US territory - Parkes - that sees it much later in its flight path.

The whole idea is redundancy and this is important when we get to the Russian systems. No one sensor can be trusted. And I'll get to some accidents involving individual sensors throughout the cold war. But the idea is if a satellite screws up or has ambiguous information the radars will confirm or clarify. Or if radars screw up or if they show faulty information the people in command posts in Russia or the US can look at the evidence from the satellites in space and say is there corroboration. This technology is not foolproof. And the military knows that and that's why they have three lines of sensors around the world..

Onto the really scary stuff, and we're talking about huge missiles with several stages can go half way or more around the world. Inter-Continental Ballistic Missiles can strike US territory or vice versa in 25-30 minutes and this is where this 30 minute decision window comes in. The idea is you launch on warning. On warning that the other side's attack so your weapons leave your territory before the enemy's warheads hit them a disable them. And the whole focus, I should add, has always been on using and protecting weapons as quickly as possible, not protecting populations. The idea is early warning lets you save your weapons and use them to prosecute and then win a nuclear war.

I'll stick mainly to ICBM's - Inter-Continental Ballistic Missile attacks. These are the real danger when we're talking about short decision times because ICBM's from the US and Russia can be launched like that. A US person in a launch control centre attached to silos, once he gets the orders, only has to put in one quick touch on the button of the computer for targeting, one for fusing of the weapon - that's arming it - and then there's the turning of the keys and you launch. So it’s a very quick process. So ICBM's are the most destabilising.

There's first detection and then you assemble low level leaders and we're talking about the people at the early warning centres that get the early warning evidence and look at what the computers give them. And by this point, even after one minute, 1000's of computers, millions of lines of computer code have already been involved because the human eye cannot track split second satellite or radar data. Rather that the evidence they get a minute or from a radar 10 minutes on is heavily processed by computers so its intelligible. So keep that in mind. The evidence is filtered, its fused, its correlated, its sorted and so on.

Okay so the low level leaders, the ones who interpret what the computers give them, have to come to a decision in three minutes in US operations and its pretty much the same for Russia - whether the evidence is valid and whether the US is under attack.

Then there's a huge teleconference. Telecommunications virtual conference. Between 3 command posts in the US. One is NORAD in Colorado - North American Air Space Defence Command. They're the early warning centre that will get all the warning information. And Strategic Command and the Pentagon Command Centre as well which is where the President will hopefully be although in most scenarios the President wouldn't have time to get to the Command Centre.

So you have the high level leaders. The high level leaders make a decision in about the 12-15 minutes after first detection of launch. Then they transmit the action messages for launch to the deployed forces, the submarines, the American Tridents and land based missiles. And then there's the launch sequence and full retaliation.

And to give you an idea really how short a time this is for such crucial decisions that impacts the world. It takes me longer than thirty minutes to get a pizza delivered to my door in Washington DC. This is not a lot of time for human beings to make rational decisions that literally could spell the end of the world or the absence of launch and the continuation of the status quo.

Eric Miller: This is the Radioactive Show and we are listening to Michael Kraig at the British American Security Information Council. Although the Cold War has been over for nearly ten years we still have these missiles on hair trigger alert. And they depend on computers on whether they're going to be fired or not. Michael Kraig goes on to tell us what is wrong with these early warning systems without the Y2K bug.

Michael Kraig: What I'm gonna do is continue to go into some rather harrowing details, I think, about the status quo before hitting Y2K. Most complex computer systems and software, and that's really software involving anything over 1000 lines of code, cannot be tested every line. When you get into millions of lines of code like at NORAD or at Strategic Command in America there is simply no way to guarantee the 100% absence of computer bugs that will turn up later because of an odd co-incidence of circumstances. Like certain data streams interlocking in a way that causes confusion between computers. Testing for such complex systems only really reveals the presence of bugs which you work to get rid of, rather than the 100% absence of bugs.

By 1980, about a decade after US early warning satellites were in space, actually a little less than a decade, there were 152 or 156 documented failures or errors or near accidents, several involving computers. I already mentioned the solar flares which is actually a natural occurrence blinding sensors. But also in 1979, a NORAD official inserted a training tape into one of the computers at the time, one of these testing training tapes, thinking it was safe to do so. Suddenly all the lights lighted up and there's different rooms and compartments in NORAD. Everyone's not in the same room contrary to common intuition. There are actually three separated departments in NORAD and they have to communicate through thick, concrete walls with each other. So no-one knew that the person had inserted this training tape right away. It took a while to figure that out. And everyone thought that a Soviet attack was underway.

They figured it out after about eight minutes. And to give you an idea of the incentive scheme that these officers work in. Despite the fact that the NORAD officer waiting eight minutes to come to a conclusion rather than the required three, despite this action had possibly saved the world or kept the US from having an erroneous nuclear response. He was sacked and no officer of that low of grade has ever been allowed into that position again because you're only supposed to take three minutes.

You may say that with this incident, well that's easy don't be an idiot and insert the training tape at the wrong time and the right place. But as it turns out in a Senate testimony about the incident, an upper commander admitted that they couldn't replicate the incident. The systems were so complex that if they just went and re-inserted the tape in sort of the same way as this person had, that they really wouldn't be able to create the same stir. They couldn't replicate it. It's really a hallmark of complex computer systems. Failures, although they happen often, are idiosyncratic by nature, they're random by nature. Its not like a finely tuned engineered German car where you can say this part will fail at five years at 80% probability.

Computer programs are creations of the programmers. Computer Science is not really a science. It's as much as a work of creation of individual programmers than anything else and therefore errors are idiosyncratic as well. They're not subject to a nice probability analysis of occurrence.

To give you another idea in 1980 a US invented telecommunications chip, which had worked for years and had been part of all their annual integrated simulated tests, suddenly failed and told 2 of the 3 command posts that they were under Soviet attack. They figured out in a shorter period this time, 3-5 minutes, that the data wasn't consistent. They called up people, the people operating the radars and satellite data receiving stations in places like Australia or in England. And this is where telecommunications comes in, and verified that no attack was underway. But it nonetheless took them 3 days and a second false alert to root out the dime-size invented chip that cost 64 cents and was causing the trouble. And put simply the engineers and computer scientists, computer engineers that built these systems as a group never could've predicted these faults and they never have. What happens is these faults turn up. You say okay what can we do to avoid that again. And then you fix that, and then you wait for, well hopefully you're not waiting for the next incident, but generally incidents keep happening.

A last example for the US. In 1960 when we first employed our ballistic missile early warning radars around the world. The programs that filtered and processed data for the radars mistook the rising full moon for Soviet missile attack. And the reason they didn't act on that is because there was no crisis at the time. And they figured why would they be attacking us if there's no flare up over Berlin at this point. And that just shows how important the general tension level in international relations is.

Then not to leave out the Russians. In 1983, Russian satellite software didn't completely figure for the intensity of some solar flares. And it showed evidence, instead of sun glare off clouds, it told one lower commander in Russia that 5 American ICBM's were headed towards Soviet territory. And the evidence was given high confidence by the computer. This is another aspect of computers - they won't tell you when they're failing. You have to figure it out.

He literally in his words to a Washington Post reporter said he made a gut level, intuitive decision in five minutes that the data simply wasn't correct. Again because there was no ongoing protracted international crisis like at Berlin or Cuba.

Finally and again without Y2K there is the problem of Russian degradation of early warning systems. Russia cannot at all from space detect the launch of US Trident submarine missiles. And to give you an idea how destabilising this is, the US still today, because we have not changed our war plans of striking fast, of striking quick, striking accurately and winning a war. We still have Trident submarines - 4 deployed in the Baltic, in the North Atlantic, in the Pacific and Indian oceans. We can hit Moscow in 10-15 minutes with very accurate warheads. As well as pre-empting most of their arsenal. Creating obvious incentives for the Russians to launch as quick as possible at indication of attack. What this means is since the satellites cannot see the initial launch, the decision period is actually less than 15 minutes for the Russians. The first indication they'd get of our trident missiles attacking them is from the ground radar raised closer into the Russian periphery. About a 1/4 to 1/2 way through the time of the missile flight. So they don't even really have 15 minutes for most instances of US submarine launch and they still of course have these blind corridors in the ground cover.

Eric Miller: With the Russian systems falling into disarray, it is making nuclear exchange even more likely than the dark days of the Cold War. This short warning time is the same as the Persian nuclear missiles and the cruise nuclear missiles would've brought to Russia in the mid 1980's. This brought millions of people out on the streets around the world. And Australia is still a nuclear target. Pine Gap near Alice Springs and North West Cape in Western Australia are prime Russian targets. Michael Kraig..

M.K: Putting all this together to give an idea of the non Y2K nuclear status quo and how it effects Australia. What this means is that one, Russians have very little or almost no redundancy in early warning systems. If a radar fails they probably don't have a satellite to back it up. If a satellite fails, they probably don't have a radar to back it up. Because they have holes already without system failures. Also, one conclusion drawn from this history is that despite the fact that the DOD in the US, the Department of Defence, rigorously tests their early warning systems every year, more errors can be expected every year simply because that's the way complex computer software is.

The US has roughly 2440 warheads on high alert. Keep in mind we have several warheads per missile and each warhead is thermonuclear so much much stronger than the bombs used at Hiroshima and Nagasaki.

Russia has roughly 2000 on high alert. Primary targets for decapitation by Russia - US early warning and control systems would be Pine Gap and the long antenna - there's a long antenna at North West Cape, a very low frequency antenna to communicate with submerged US trident submarines in the Indian ocean. So at each end of this 30 minute window Australia has nuclear targets on Russia's list. So Australia is really caught and I'm sorry to say you should definitely be concerned.

That's basically the global non Y2K status quo for nuclear operations between the US and Russia. So now on to Y2K and the bad news.

Y2K is as much as a management or organisational problem as it is a software problem. When you have thousands of systems and millions of lines of code and systems that interface with each other which the US Department of Defence has in abundance. You're talking about a co-ordinated management effort to get that stuff fixed, not just a bunch of computer geeks. It's precisely where in the area of management that the US Department of Defence is very lacking.

Assessment is the second phase in the Y2K process. And here you identify your systems. And believe it or not, this is not simple. The Department of Defence found out to their dismay that they didn't know what they have completely or what intersects with what. We have four services that have rivalries. The navy doesn't like the army. The army doesn't like the air force and vice versa.

Y2K is actually easily understood in terms of what's needed to fix it. Its not a mathematical challenge or anything like that. What makes Y2K hard is not the absence of fixes that are known to work well; there's about 10 of them actually that can be used. But rather that you're talking about millions of lines of code, you're talking about dependencies between systems that you maybe don't completely understand and just lots of man or lots of people hours of fixing. To changing code. Scanning every line of code. Changing data fields from 2 digits to 4 or introducing some mathematical function to atone for the fact that there's only 2 digits, or some other operation to deal with it. So it's just the tediousness of the whole enterprise and the magnitude of it rather than the fact that no-one knows exactly how to do it.

This brings us to the validation phase. Testing. Basically you test to make sure that your initial repairs were successful. This is important because as I said before a large percentage of the time, your fixes need fixes themselves.

Well, there have been tests by NORAD, by North American Air Space Defence Command. These tests have included quote "the thin line", the minimal number of computer systems required to execute the mission. Both Department of Defence and private telecommunications routers and switches were not included. The ground radar arrays that we talked about so much, their fixes caused problems and they have to go back and refix the fixes and private power supplies were not included.

I guess just generally what is the answer? The answer of the Pentagon so far is to make sure they fix everything, test everything and have everything in working order to still win a nuclear war. Another possibility, which they have rejected, is to work with Russia and do something called de-alerting. Or standing down nuclear arsenals. To give you some examples, the UK, France and China cannot launch their nuclear missiles and attack with their warheads at any less than days or couple of days time. France ended its dependence on ICBM's. It relies on submarines. So does the UK. They're not quick launch weapons by nature. China doesn't have nuclear warheads on missiles. And their missiles probably aren't even fuelled on a daily basis.

In fact we're looking at sharing our early warning information with Russia on a daily basis. December 15 through to January 15 then later on at a more permanent facility outside of Moscow and a temporary facility for the Y2K rollover is at Colorado. And this raises the question - if we're sharing data which shows whether or not we're launching towards them, why are we still geared to launch towards them? In a very short time?

Just to give you sort of the paradoxical nature of this. In talking to reporters again and again and congressional staffers in my lobbying on the Hill, the common intuitive opinion which is reasonable is that we'll ignore evidence of launch. Why the heck would we launch towards them when there's no crisis and we know Y2K errors may crop up. For 1, you can't count on a crisis not occurring and 2 what this means is they believe their early warning data if there's no positive indication of an attack but if there is positive indication of attack they don't believe their own systems. This is Alice in Wonderland.

And really standing down arsenals, getting more heads off missiles or other intermediate technical means so that no-one can launch an all-out first strike surprise attack. Really that’s the only real option and it’s the option that the Pentagon is afraid of and Russia itself is sceptical - it won't act unless the US does first.

I can answer questions later about the feasibility of de-alerting or where what various US policy makers think of it. Unfortunately most of my conclusions are pessimistic including Clinton. Clinton's past actions as well. Alright, have a good day.

Eric Miller: That was Michael Kraig of BASIC. The British and American Security Information Council. He is based in Washington DC. The sharing of early warning information was put on hold during the Yugoslavian war. But let us hope that as tensions ease the sharing of early warning systems during the new year can become a reality at least. And the nuclear missiles must be taken off trigger alert, and we must ban all nuclear weapons.

Michael was speaking at the Radiation Alert Symposium for Sanity. It was organised by Helen Caldicott and we'll give her the last word.

Helen Caldicott: How dare they put their nuclear weapons on computer control. How dare they. How dare they fission the atom and make huge reactors full of as much radiation as a thousand Hiroshima bombs all over the world on computer control.

Who do these people think they are? And it’s a handful of men in the Pentagon. A handful in NORAD or Strat Comm who are deciding the fate of the earth. A handful of men. And Clinton doesn't give a goddamn. He could've produced abolition of nuclear weapons. If you X-rayed his spine there'd be no image on the X-ray. He's an intelligent man but he's not a leader's bootlace.

Eric Miller: Helen Caldicott at the Radiation Alert Symposium for Sanity. It was held in March in the Old Parliamentary House in Canberra. The Radioactive Show is produced at 3CR in Melbourne and if you have an issue that you'd like to air or just comments to make on the show, please contact us at 3CR in Melbourne or by email on ar3cr@hotmail.com. Transcripts of programs can be got at the sea-us website. That is www.sea-us.org.au.

That's all the time we have on the Radioactive Show this week. It's goodbye from Eric and please listen in next week.

Transcript produced by Gary Schurr - with much thanks!!!
Page last updated September 12, 1999.
Back to the Radio-Active Show Archivesor Back to the SEA-US Front Page

Copyright © SEA-US 1999