Re: Ontology of disaster management

Craig,

Thanks for your, and Gavin's, comments.

I come from an Info Studies angle and I'm not a practitioner in  
emergency management so my comments are limited to the general  
development of ontologies and their design.

I think the reason ontologies for the response phase have flourished,  
whether formally in the US or more loosely in Sahana or other systems,  
is that the community of interest in the response phase is relatively  
localized and/or circumscribed: it's generally relief agencies and  
their operatives.  This community can thus reach agreement about the  
elements and attributes of their domain.  An expressive ontology of  
resilience is harder because it involves many different institutions,  
from local communities to national planning agencies. It is  
questionable whether islands of uniformity exist that could be  
exploited to derive a normalized description of these domains.  I  
think the US effort is developing a relatively thorough ontology of  
response phase communications and subsequently resource management,  
but resilience will be a much bigger ask.

An "ontology of chaos" is relatively easy because anything of value  
will be accepted in those circumstances, even ad hoc, on-the-fly  
information systems.

 From my perspective, I welcome Paola's insistence on testing.  Agile  
iterative development, user testing, all the trappings of  
user-centered design, can only bring the ontology in line with the  
community of interest.

Quentin Halliday



Quoting C H <craighubleyca@yahoo.com>:

>
> ----------
> SUMMARY
>
> Gavin's right, though Quentin's point about the irony
> of an "ontology of chaos" can't be ignored, and
> suggests more compromises to accomodate long-term data
> integrity and management needs to gain more management
> compliance, cooperation and perspective.  These are
> three different problems.  It's a very social problem:
>
> An ontology that describes situations in which it is
> not possible to manage at all, will be difficult to
> convince long-term managers to impose or even use.
>
> Thus:
>
> A "Disaster Management" ontology that tries to deal
> only with the narrow slice of time and narrow range of
> issues involved in a specifically "disaster" scale or
> "relief" or "response" time frame, will badly fail.  I
> believe the vision of how this ontology should work in
> practice will strongly resemble what Gavin outlines -
> and that this is almost exactly the model that the UN
> Resilient Communities and Cities Initiative pursues.
>
> However, everything is negotiable.  The reasonable
> constraints applied by citizens on data gathering and
> lifestyle inhibitions of risk reduction and also on
> expenditures to prepare for low-probability
> high-impact scenarios cannot simply be ignored.
> Neither can the need to align long-term and short-term
> efforts.
>
> Thus:
>
> Any ontology must work well enough for Response and
> Recovery - this is non-negotiable.  However, it works
> better if it has already been coordinated with longer
> term needs and embodies principled compromises - that
> make no one completely happy.
>
> Improvements to coordination between the phases will
> be more likely to reduce risk than accomodating any
> one phase so far that it reduces compliance in others.
>
> As Gavin Treadgold says:
>> the interconnectedness of disaster impact
>> assessment
>> information suggests that an ontology developed
>> solely for Response,
>> will fail to meet the needs of comprehensive
>> emergency management as
>> it won't allow DIA information to be captured from
>> events, and
>> transferred to future research and planning.
>
> It will also cause numerous culture and coordination
> and conditioning problems among victims and responders
> and managers, due to shifting back and forth between
> one mode of operation and another, and the stress and
> errors this causes.  Resilience minimizes this stress
> and reduces load/stress in Response phase by making it
> more likely that locals can handle their own Response.
>
> Resilience is, in effect, the feedback from the last
> to the first phases and the standards underlying all.
>
> ---------
>
> Gavin and Quentin are debating the most difficult and
> thus most important problem.  Their points deserve
> very deep examination and will change both the mission
> and the name of the project in one way or another.  I
> therefore attempt to combine their views in some way.
>
> I am in absolute agreement with Gavin on all counts
> except a couple of detailed points of language and a
> more serious difference in balance between phases.  I
> believe "Response and Recovery" are MUCH weakened if
> they insist on applying inappropriate data standards
> or practices prior to disasters or fail to coordinate
> or compromise with longer term resilience efforts.
>
> One obvious problem is that shifting over from pre- to
> during- to post- disaster methods and standards is
> costly, confusing, risk-prone and extremely likely to
> fail given the stress people are under.  In a crisis,
> people fall to the level of their conditioning.  So it
> is better to have one robust set of tools, procedures,
> data standards, even if this is slighly less useful in
> the response and recovery phase than another set that
> would have to be imposed by surprise on stressed
> folks.  Admittedly this is not always possible, but
> when it is possible, it should always be preferred.
>
> Another problem is that short-term events are quickly
> forgotten and cannot easily change long-term habits,
> so the motivation to cooperate with prevention efforts
> is less if those efforts aren't seen to be part of the
> ongoing routine effort to reduce other routine risks.
>
> There's a gulf between "management" and "crisis" style
> and habits of decision-making.  The worst time to have
> that turn into a fight, is when a disaster hits.  So
> recognize that cooperation and resistance to pre- and
> post-disaster efforts, and to handoffs either way,
> tend to be partly determined by the degree to which
> the "Response and Recovery" varies from the standard
> routine practices.  Thus the "resilience" approach
> which minimizes these differences and seeks robust
> practices useful across all phases.
>
> For instance, I would rather have a vulnerable persons
> list that meets privacy laws and is accordingly
> limited in scope, but often consulted, often updated,
> consulted long before the disaster strikes for
> planning, and immediately available for use when the
> disaster hits.  This is clearly better than to have a
> specification for a very detailed such list, that can
> only begin to be compiled once a disaster has been
> recognized and authorization to over-ride privacy laws
> is given.  It's possible to keep the data in separate
> "silos" and then combine it only once the disaster is
> imminent, but then it's all the more important to have
> it follow the same gathering discipline and integrity
> tests (why the ontology must be useful in all phases).
>
> That said, there are some who object to this approach.
> I believe Quention's issue of resilience and disaster
> reduction interfering with lifestyles is more a US
> bias with predictable and disastrous outcomes. It is
> at best extremely expensive (rebuilding homes in flood
> plains or storm surge zones to await another
> disaster).
>
> The above however amounts to quibbling.  As Gavin very
> pointedly argues:
>
> The NZ approach appears to be quite robust and the
> only one that is truly dealing with implications of
> the plain fact that information must be gathered
> before, and used after, the so-called "response", and
> that there's feedback between the phases and persons.
>
>
> So a "Disaster Management" ontology that tries to deal
> only with the narrow slice of time and narrow range of
> issues involved in a specifically "disaster" scale or
> "relief" or "response" time frame, will badly fail.  I
> believe the vision of how this ontology should work in
> practice will strongly resemble what Gavin outlines -
> and that this is almost exactly the model that the UN
> Resilient Communities and Cities Initiative pursues.
>
> I think the following statement of his could go in the
> mission:
>
> --- Gavin Treadgold <gt@kestrel.co.nz> wrote:
>> the interconnectedness of disaster impact
>> assessment
>> information suggests that an ontology developed
>> solely for Response,
>> will fail to meet the needs of comprehensive
>> emergency management as
>> it won't allow DIA information to be captured from
>> events, and
>> transferred to future research and planning.
>
> My point regarding compromises and conditioning
> however would change what follows slightly, as we
> should avoid use of non-operational terms like "work"
> or specify what is "non-negotiable" or seem to side
> with one side or another.  I would rewrite this:
>
>> Any ontology must work
> well enough
>> for Response and Recovery - this is non-negotiable.
>> From there, it
>> makes sense then to apply it across all four phases
> and to seek principled compromises to gain compliance.
> Improvements to coordination between the phases will
> be more likely to reduce risk than accomodating any
> one phase so far that it reduces compliance in others.
>
> Or, in more detail, one might argue that case as:
>
> "Any ontology must enable Response and Recovery - this
>  will be how it is judged by most of the public and
> the
>  media.  However, it's ability to enable those phases
>  is radically reduced if it makes no compromises or is
>  in violation of principles or laws that apply both
>  before and after these phases.  If data gathered in a
>  response and recovery effort is not immediately
> useful
>  afterwards, it will fall out of date.  If data that's
>  gathered before is not immediately useful or
> available
>  during the most intense phases, the major opportunity
>  to reduce impact (and gain more cooperation) is lost.
>
>  Compromise costs.  Those who take responsibility for
>
>  just one or two phases of a comprehensive emergency
>  management approach can reasonably be expected to
>  object to choices and compromises made to accomodate
>  constraints that apply in other phases.  However,
>  for every cost one seeks a benefit like reduced risk
>  and failures of coordination are the worst such
> risks.
>  It's better to avoid anyone having strong objections
>  than it is to overcome all objections of those whose
>  responsibilities will be limited to a few phases,
>  because having only one standard is so advantageous."
>
> Another statement worth considering for inclusion in a
> mission statement or paper defining the mission is:
>> community aspects including infrastructure planning,
>> emergency
>> management, hazard management, risk management,
>> urban design and
>> planning and sustainability are all inherently
>> linked and cannot and should not be separated.
>
> It may help to critique other efforts, as Quention is
> seeming to do here:
>> > At present, however, the field of emergency
>> management is
>> > concentrating on developing a description of the
>> Response Phase, an
>> > ontology of chaos as it were,  however ironic that
>> might be.
>
> Gavin says:
>> I don't quite follow this statement?
>
> I think I do.
>
> Assuming Quention means to invoke the terminology used
> in say the ABIDE framework, I believe what this is
> intended to say is that the phase where there is least
> order is the one that is getting to impose its concept
> of order on other phases that are already much more
> constrained and ordered.  This can easily fail and at
> least leads to major clashes and incompatibilities.  I
> believe it is fair to say the the US ER establishment
> has experienced exactly these domestically and abroad.
>
> The "resilience" approach does the opposite, it tries
> to discover what is NOT changing even in a disaster so
> as to maximize the use of existing data at that time,
> and also maximize the usefulness of data gathered in a
> response and recovery effort.  As I've outlined above.
>
> For "chaos" you can reasonably read, as I understand
> ABIDE, "pre-baseline".  That is, the situation in
> which there is no way to assess whether things are
> getting worse or better.  In this situation, you MUST
> act and so an ontology that describes action (only) is
> appropriate, NOT one that attempts to define the
> situation itself beyond a set of inhibitions to
> action.
>
> So a list of types of disaster could help you identify
> types of problems to watch closely for, patterns of
> how the problems get better or (anti-patterns) worse.
> It could authorize certain types of actions or probes
> or data aggregation that would otherwise be forbidden.
> But it would not really help you to "manage" a crisis.
>
> Once a baseline is established, the situation becomes
> "complex", that is, you know if it's getting better or
> worse, just not why.  If you try to develop a plan at
> this stage, you're going to miss things and the plan
> will fail.  Your senses and instincts remain quite
> untrustworthy.  So instead you focus on probing to see
> what specific actions seem to cause less or more good
> or bad things to happen.  This is more like field
> social science epistemics where you lack a control
> group and rely on pattern recognition and heuristics.
> You can do harms reduction, for instance, not develop
> comprehensive treatment plans.  Like field medicine.
>
> Only once there is some agreement on the causality of
> the worst aggravating factors, only once there is some
> probing that establishes what's working and what is
> not working, can we really start to define the "work"
> required long term.
>
> We are dealing in causes now not mere correlations.
> At this point the situation is merely "complicated"
> and can be managed by people with long-term skills.
>
> ABIDE and like approaches are discussed often in US
> circles. It seems to be a response to massive failures
> of management and planning such as Katrina and Iraq.
>
> In both cases, the inability of field personnel to
> take reasonable action or reduce harms (like
> protecting the Baghdad Museum or emptying deserted
> supermarkets of supplies) may have led to a cascading
> series of events (looting, lawlessness, self
> protection societies that become sectarian groups, or
> opportunistic alliances that become insurgent groups
> or crime gangs).  But rules of engagement were being
> "managed", emergency relief supplies "managed", etc..
>
> In the ABIDE terminology, one would say that "chaotic"
> and "complex" situations were thus being approached as
> if they were only "complicated", and attempts to
> "manage" them were made by persons far away who did
> not understand them first, and indeed had no idea of
> how one develops understanding of complex situations.
>
> So Quentin's statement is indeed ironic in that it's
> those who attempt to over-simplify the problem in the
> first place, who are trying to impose their failures
> on the next problem by over-simplifying the ontology.
>
> Elaborating the lingo, what he said can possibly be
> rewritten as follows, though of course in more words:
>
> "Persons skilled in response and recovery often fail
>  to realize that the minimal order they must impose in
>  a genuinely chaotic situation won't be acceptable nor
>  even comprehensible as a goal to the persons they
> must
>  cooperate with to prepare for, and reconstruct after,
>  a disaster.  Accordingly their terms of reference are
>  usually poorly integrated with those of colleagues in
>  positions of long-term responsibility.  Data gathered
>  beforehand by, and left afterwards to, the long-term
>  authority, is useless if it describes distinctions or
>  definitions used in response and recovery but not in
>  long-term planning or routine operations.  Even the
>  victims are likely to demand or seek normalcy or take
>  action to impose it, resulting in 'vigilante'
> effects.
>
>  The chaos of disasters require those who arrive first
>  to establish a baseline understanding of situations.
>  The complexity of response and recovery require then
>  more disciplined probing and experiments to establish
>  what is working to improve those situations - or not.
>
>  Guiding such action and probes under uncertainty is a
>
>  distinct and separately studied phenomena, and what
>  terms are appropriate to describe it may not be
> useful
>  to describe even the most terrifying, complicated and
>  confusing situations that arise in routine emergency
>  management.  In the latter, management is possible.
>
>  An ontology that describes situations in which it is
>  not possible to manage at all, will be difficult to
>  convince long-term managers to impose or even use."
>
> Or just take that last paragraph.  I'll leave it to
> Gavin and Quentin to say if I've understood them or
> not, and if they object to the above elaboration and
> integration of their respective positions.
>
> Craig Hubley
>
>
>
>
> ____________________________________________________________________________________
> Sucker-punch spam with award-winning protection.
> Try the free Yahoo! Mail Beta.
> http://advision.webevents.yahoo.com/mailbeta/features_spam.html
>
>

Received on Wednesday, 20 June 2007 03:57:01 UTC