Andy Smith
MIne-action specialist

 
Land Release - a reduction in standards


'Land Released' without being cleared used to be called 'Area Reduction'. That was the process by which deminers went to a reported mined area and decided which parts of it needed to be cleared. Sometimes it is immediately obvious that parts of an area do not need to be cleared. Sometimes that becomes obvious as the work is conducted. Land not cleared was recorded as 'reduced'. The Land Release concept extends this to include releasing areas without even going to them, which is not quite as irresponsible as it sounds because the original records of mined areas are often very imprecise and unreliable. So, under Land Release, areas can be 'released' without conducting any demining at all. This avoids conducting any demining activities in areas that were wrongly recorded as being hazardous, but also invites error because there is no requirement to check whether there really is a problem there. Up to now, the only way to check has been to search some parts of the area in a process known as Technical Survey which is intended to narrow down the area that is hazardous and find out what hazards are present. The new Land Release way to check is to conduct some kind of process in the area that raises confidence that there is no explosive hazard without actually wasting time and resources by thoroughly searching any part of the area. Can a process like that prove there is nothing there? No, but a vague requirement to 'raise confidence' is a lot easier to achieve than proving the area is safe. It can also be quicker than normal Technical Survey methods and so make managers look efficient.
But it is an erosion of standards for which the end-users of the land must pay.

Let's put this in context....

Back in the bad old days of the 1990s everyone had to clear 99.6% of mines and explosive hazards where they worked. This was because an old British Brigadier General who was working for the UN had once publicly said that 99.6% was the best that was achievable. It seems that he invented the figure because neither he nor anyone else has since been able to explain how it was calculated. 99.6% may sound like a lot, but in a barrier minefield there are often many thousands of mines. Professional deminers must try to clear them all because they will be the first people to step on the land, and because they will bear the consequences of any mines ‘found’ later. The commercial companies involved in writing the first IMAS hated 99.6% because if they found more than 250 items and those conducting QA found a mine, that was still acceptable. The commercials (led by Chris Pearce, a founder of Minetech) did not like this because it meant that cowboy competitors could get away with working badly and so undercut them to get contracts. Minetech led the IMAS founders (of which I was one) to agree that 99.6% was rather silly. We came up with the definition of clearance that required the removal of all mines and explosive hazards to an agreed depth. From the first (albeit almost hidden in IMAS 07.10) there was a definition of clearance that stated under the heading “Clearance requirement”:

“The target of humanitarian demining is the identification and removal or destruction of all mine and UXO hazards from a specified area to a specified depth.”

If that does not sound like a formal “definition”, it is repeated in the IMAS Glossary (04.10) under the heading “clearance” as: ‘in the context of mine action, the term refers to ..... tasks or actions to ensure the removal and/or the destruction of all mine and ERW hazards from a specified area to the specified depth.

For manual deminers this merely reflected what they did their best to achieve anyway.

It has survived as a definition, albeit slightly revised in 2009 to now read, “The aim of humanitarian demining is the identification and removal or destruction of all mine and ERW hazards from a specified area to a specified depth”. The word “target” has changed to “aim” and the acronym “UXO” has changed to “ERW”. These are minor revisions that were intended to introduce the “consistent” use of the words, so the changes did not need the approval of the IMAS Review Board.

The inclusion of the words “or destruction” allowed the use of machines that could reliably destroy all mines and ERW with mechanical implements. I could not see how that could ever be achieved, but had to accept that it might one day be possible because the mechanical lobby at GICHD was too powerful to oppose unless it really mattered. All decisions made by the IMAS Review Board should be by consensus, so it is important that individuals choose their battles and go with the flow whenever they can. With other battles in hand, I went with the flow.

In 2011, I used the demining googlegroup forum to challenge the mechanical lobby’s irresponsible use of a single pass of a Minewolf machine as “clearance” in Sudan (it left behind many mines and ALL other explosive ordnance and this was approved by the UN team there). This led the temporary head of UNMAS to manoeuvre me into a position where I had to leave the IMAS Review Board. It seems that UNMAS was then championing a reduction in a standard that is the foundation of humanitarian mine action – that is, changing what “clearance” actually means.

Now I have concern that the ill-defined Land Release IMAS will lead to a further reduction in the clearance requirement. Land Release standards probably have to be ill-defined, because area-reduction (during technical survey or at the start of a clearance task) has always relied on the informed risk assessments made in the context by the people who will be the first to take risk on the land that is not cleared – and the first to take the consequences if they get it wrong. Defining a simple way to do this that could be applied in all contexts around the world is simply impossible. But that does not stop people trying. It has always been a major failing of the IMAS Board, UNMAS and the GICHD that those heading the organisations see their role as one of imposing orders and systems on the people doing the job rather than serving the people doing the job. There is a balance between these two goals, but not an easy one.

Land Release is effectively being used to hide a double standard. Land that has been ‘cleared’ by searching and removing all explosive hazards to a depth and land that has been ‘released’ by using procedures on the land that we know cannot achieve real clearance over a single square metre. A case could be made for a double standard but it would be hard to justify, so no case has been made: instead it has simply been imposed under an ‘efficiency’ smokescreen.

When I agreed the wording of the first Land Release standards (meeting with HALO and Guy Rhodes and Håvard Bach for the GiHAD in Scotland during 2009), one major concern was that those tasked with improving figures would simply cancel land recorded as hazardous without taking responsibility for the consequences. We went to some lengths to ensure that land could only be released after “a credible and well-documented process has been fully implemented”.

I suggested including the line, “those prepared to release the land should be prepared to walk over it or traverse it in vehicles (depending on the type of hazard that was found in the area or that is suspected of being present)”. I wanted this to be a “requirement” (“shall) but others agreed to put it in as “guidance” (“should”). As “guidance” it appeared under section 7, Land Release criteria in the 2009 version of IMAS 08.21 (which was then part of the Land Release series).

The line required a level of confidence that should have made those who might release land unwisely think carefully.

But that requirement evaporated after I resigned. On 1st March 2013, a restructured set of Land Release IMAS was released. Under the restructuring, IMAS 08.21 became IMAS 07.11 and under section 7 the text remained identical except for the guidance that those who were willing to release the land should be prepared to use it. That line was simply removed and its removal was not listed amongst the changes recorded in the Amendment record at the end. The failure to record its absence in the amendment record was an error because deciding to omit it was significant enough to have warranted discussion by all Board members. But it seems likely that no-one noticed, or possibly no-one cared.

In both versions of the Land Release IMAS you will find the following line:

“A confidence rating system should be considered to allow quantification of survey information to facilitate improved decision making. The information needs to be reliable.”

So what is a confidence rating system? Amongst other things, today it is a way of predicting the probability that the use of a machine over an area without any detonations means that the area is hazard free and can be released. There is no machine that reliably initiates or destroys all pressure operated devices or any other ERW.  There is no finite variation of the context variables that occur in demining tasks around the world. There is no single level of “tolerable residual risk” that can be applied across all post-conflict societies. With so many variables that cannot be either controlled or predicted, creating a universal confidence rating system that can be confidently applied BEFORE any clearance is conducted in any task area is an impossible task. But still the race is on to decide a way that will allow the industry to release lots more land without clearing it regardless of the consequences for the end-users (whose safety is nominally the reason for doing any work at all, of course). They want to create a universal confidence rating system that can be applied in any task area and accelerate Land Release regardless of any inevitable impact on end-user safety. It is true that the current ways of reducing land are not readily measurable or replicable – but no imposed pseudo-quantitative confidence system that compresses variables and relies on incomplete statistical analysis would be any more reliable. Worse, it would move responsibility for error away from the field person who would still have to take the consequences of all errors.

Most professional NGO demining groups have always released land without clearing it when it is obviously safe to do so. Commercials are sometimes contracted to clear the whole area and so have to, although I can think of more than a few cases where they did not because there really was no need. Who makes the decision to reduce land without clearing it? the man on the ground. Has he always been right? Almost, and I think this is because he has known that he and his organisation must take the consequences if he is wrong. The idea that someone could impose a confidence grading system that would prevent him taking responsibility for that decision is deeply disturbing. But the challenge of creating one is obviously attractive, even to those who recognise that any system could only become even theoretically “reliable” after iterative revisions that took account of failures... Revising a system as experience is gained seems realistically pragmatic – but in this case each failure would only be discovered at the cost of civilian limbs and lives.

I was surprised when Norwegian Peoples’Aid (NPA) publishing just such an unreliable confidence grading system with Håvard Bach’s 2014 article in the JMU Journal   www.jmu.edu/cisr/journal/18.1/focus/bach.shtml. I have worked with NPA here and there and yes, we have always reduced areas whenever it was safe to do so. But we have not had some pseudo-scientific system to apply. “Pseudo” because Bach’s system is based on guesses and blatant bias. And I wonder whether NPA realise that his “method” moves the Norwegians away from “Humanitarian Demining” according to its definition in IMAS?

As example of bias is Bach’s repetition of his long-held opinion that the use of one dog gives 95% confidence of finding any targets. I first heard this guess ten years ago – and argued that it was wildly optimistic. Ironically, NPA’s own Global Training Centre for Mine Detection Dogs (GTC) in Bosnia confirmed my opinion when I worked with their handlers to write the SOPs for NPA dog use in Jordan. They confirmed that using one dog is not appropriate because the use of two dogs provides a continual QA check on the other. As soon as a dog stops performing, it is known and the dog is replaced in the search by another dog. The non-performer is immediately sent for reinforcement training/retraining. With only one dog, no one knows when it is not performing so it continues to be used and the confidence in the quality of the search drops dramatically. It is also bad for the dog’s motivation because it knows that it doesn’t have to perform in order to be allowed to keep “playing” (and they really want to “play” because they want to be with their handler). This explains why each handler has two dogs, so allowing a dog’s performance to be “off” without unduly disrupting output. The Bosnian handlers I worked with wanted three dogs, so guaranteeing long hours of work each day. If you use one dog, the quality of the search could be so low that it found nothing – 0% – and you would never know. NPA’s own GTC would confirm this.

An example of pseudo scientific blag in Bach’s paper is the assignment of a percentage to the clearance capability of machines. He writes, for example, that flails and tillers clear 40-80% of hazards. Now I am not fond of using them for clearance (as the UN programme in Sudan knows) but they have their uses. However, to confidently assign any kind of percentage to their ability to initiate devices is impossible without knowing the context (ground, vegetation, slope) and the range of explosive hazards present. If you look at their performance in test beds sown with fake pressure mines, it is higher that 80%. If you look at their performance in the field, it can be as low as 0% depending on the hazards present and other context variables. The pretence that you can look at past performance records and derive a reliable figure that would allow you to predict performance at a new task is simply a convenient lie.

And behind all this lies the notion that I could be 40% confident that there is nothing there. That really doesn’t work when you look at a suspected hazardous area. If 40% only refers to a process by which four out of ten mines would be initiated, I have to ask why? If the context and the mines meant that six could be “processed” and not initiated, why not all of them? Or, having bought into the senseless notion that I have 40% confidence, do I only have to clear 40% of the land? If so, is that any 40% rather than the predictable hotspots within a suspect area? This kind of statistical fluff really is not meaningful. Fortunately, the real field people in NPA do know that. (NPA are affectionately known as Nice People About, and generally are.)

Moving to the NPA policy, tucked away in the “Endnotes” of Håvard Bach’s paper is the admission that “this process is... not applicable when searching for other ERW”. He refers to the Mine Ban and Cluster Munition treaties earlier in the paper, using them to justify ignoring anything not covered by those “conventions”. The clearance requirement of Humanitarian Demining is not defined in those treaties: it is only ever defined in the IMAS – and the IMAS says all ERW. So Bach is taking NPA away from the IMAS and away from the definition of “humanitarian” by declaring that Norwegian demining is only about meeting the obligations of treaties, not about protecting end-users from the explosive remnants of war. I wonder whether NPA’s leaders know this? Is it actually their policy – or just the opinion of their “chief technical advisor of operational methods”? Do they know that explosive devices other than mines cause far more civilian casualties than mines? They may not because for many years all civilian casualties of explosive devices were recorded among the mine-victim statistics (which was convenient because high numbers were useful for campaign purposes). Disaggregating those records retrospectively is not possible because the actual device involved in an injury was very seldom recorded. But still, it is a fact that in most post-conflict scenarios ERW other than mines cause far more civilian injuries than mines. Despite this, Bach would have NPA leave them behind.

The NPA ‘Scalable Technical Survey’ system described in Bach’s paper is so ill-thought out and under-researched that it is dangerous. It has apparently been pushed within the organisation for more than two years but it is worth emphasising that no NPA field people have actually used it. Thankfully, the Nice People About in the field know better than to rely on a system that is simply inflated non-quantitative opinion masquerading dangerously as the application of science.

This shows that it is not only UNMAS and GICHD who see their role as one of imposing orders and systems on the people doing the job rather than serving those people. I myself am old enough to like deskwork when I can get it, but not yet old enough to promote illusory efficiency at the expense of safety.

 



BACK