Saturday, March 22, 2008

Crowd In Control --new article

Crowd In Control: Decision-Making During Disasters


Information dissemination/control and decision-making are two primary and interrelated activities any response must confront during a disaster. How information is shared/managed and decisions are made will shape every other aspect of relief and recovery. These important areas must be decided upon before a disaster if a team or organization is going to be able to respond effectively to the many unknowables during a crisis. The Aftershock Action Alliance model uses a decentralized and autonomous approach to both information and decisions which is at odds with most Government and Private emergency management plans. We believe that there are very good reasons for bucking the trend of centralization and information control which will in turn lead to a more effective relief/recovery effort and promote community empowerment and active resistance.

Information is a key component of decision-making and in fact is nearly impossible to imagine action without some information. There are a number of theories around how information is gathered, disseminated and analyzed. We will not go into the theories of what information is but accept the etymology of the word, “to give form” to something. Information allows us to form in our mind what something is. For a disaster there can be many types of information but one thing is clear that there is often only incomplete idea what is actually happening. This is a result of many factors like the surprise and speed of a disaster and its interruption of normal flows of information (e.g. knocking out phone lines). Disasters are not information free zones but the context of events is often disrupted to such an extent to make comprehension difficult. For example, the power may go out in our home and we look out and the street is black. We do not know if the entire city, county or state is without power. We may also not know what caused the power outage or how long it will be out. The normal attempts to seek information may be thwarted by not having access to phones, Internet, television or radio. We must decide “what is happening”, form the context for us to decide what action is to follow. If it is a blown fuse then we will search out our circuit box in the basement, if the state is blacked out then going to the fuse box is foolish.

Obviously the more correct information we have the better we will be able to decide and plan action that is sync with the events. If we have little contextual information or our information is in error our decisions and thus action will be severely impaired. That is why every organization dedicated to responding to disasters puts such a premium on information collection and management. In this way Aftershock is no different than government agencies or NGO's. Where we differ from these other organizations is in our belief in the old computer hacker motto, “Information wants to be Free”.

Government/NGO organizations spend thousands of dollars to create encrypted and scrambled communication devices so their communications (and thus information) can be private. They also use a variety of scientific and military jargon, code and abbreviations to make their communications unintelligible for the average person. What is telling is the Emergency Alert Service's (what used to be the buzzing EBS) own guidelines never mention providing “information” to the general public but only “instructions”. The motivations (and thus the information they are based on) for these instructions are kept secret from the public. This is the position of the most “public” of the emergency management agencies. The government and most private relief agencies believe that the public can not understand the information coming in and out of disaster areas. One would assume this is based on some sort of research but this isn't the case. Even a cursory glance at research from sociology, psychology and even disaster studies demonstrates that ordinary citizens can assimilate vast amounts of data and “give form to” reasonable scenarios based on what they've learned. One study done by the Santa Fe Institute, gave information about a group of tornadoes that hit Kansas in 1962, to a group of ten emergency managers and ten civilian survivors of tornadoes from New Mexico. They found that on 7 factors (e.g. Estimated number of causalities) both groups did similarly. In fact the ten civilians had greater agreement internally then the “experts”. So why won't the government give the public information it needs? One rationale for withholding information from the public is the great fear of “mass panic”. As we have written elsewhere this is mostly a myth and there is over 20 years of research to back us up. In the rare occasions that panic does occur it generally is in very specific situations. These situations are characterized by lack of familiarity with the environment, over-crowding, lack of access to the normal social networks and expert over-reaction. Even under some of these situations, panics do not occur. It seems the fear of panic is wholly over-blown and not a good excuse for withholding information from the public. Controlling information does allow organizations to exert control over groups that do not have access to information. Psychology experiments have suggested that people over-estimate the value of secret of information. For example, a psychology experiment at the University of Wisconsin in 1999, allowed long-time gamblers to peak at the one card at anytime during a standard five card draw poker game. Those that peaked, bet more even if they had a loosing hand and those that were not allowed to peak bet less even if they had a good hand. The effect on the poker game was large despite the little statistical significance of looking at one random card. What this showed is that people who get secret information, even when it is not important, will over estimate the value of that information and so will everyone else even if they have no idea what that information is and whether it will be useful for their actions. So you can see how the fact that the government has access to information the rest of us don't have allows them to exert a greater control over our actions regardless of the usefulness of that information. This approach also breeds a foolish arrogance in the part of those in possession of such supposed knowledge. This is probably the real reason governments spend so much time and resources “protecting” the public from panic by withholding important information.

Aftershock believes information should be shared and be public. We are dedicated to finding ways of getting information to the public so people can better form an opinion on what is going on. Understanding that there are many unknowns during a disaster and that normal communications systems will undoubtedly be impaired, Aftershock relies on the “many eyes, big voice” model of information sharing. Aftershock believes that people are reasonably capable of providing useful information about the key issues regarding a disaster. This of course goes against popular thought about the unreliability of eyewitness testimony. Eyewitness reliability has been questioned in courts and in classroom experiments for decades. The assumption is that people make very poor judges of actual facts and misinterpret and add to things they witness. While that is true, it is not the whole truth. Witnesses are very good at recalling and explaining certain facts (e.g. gender) and poor at others (e.g. Colors), it depends on what they are experiencing. Groups tend to increase the accuracy of eye-witness testimonies. Research suggests that 4 people witnessing a car accident can get over 90% of the “relevant” details correctly. The question is whether eye witness information about disasters is any better or worse than in car crashes. We do not know the answer and as far as we can tell there is no reliable research on that question. What we do know is that the more eyes you have, the more likely you will get accurate information. In_____ influential work The Wisdom of Crowds he shows that over a 140 years of research in biology, sociology, psychology and information systems suggest that larger groups of people can make better decisions and accumulate more useful and accurate information than smaller cadres of experts. There is some precedents for this in our experiences. At mass mobilizations decentralized information gathering systems like Indymedia website1 are superior (in terms of speed and accuracy) in relaying crucial information like arrests and injuries when compared to official government agencies (e.g. the police department) or corporate media (e.g. Local news). During Katrina amateur radio buffs provided a better description of the damaged areas and the type of flooding in New Orleans than FEMA or any other organization. Aftershock seeks to maximize the number of people that can participate in the information sharing during an emergency. The larger number of reporters or collectors of data ensure that widely inaccurate information is weeded out. Wikipedia according to a Nature article (December 2005) is as accurate as the Encyclopedia Britanica when experts compared randomly selected articles. We are confident that the information collected by a multitude will be accurate and useful if we can get enough people to share information. The question is how during a disaster do we allow regular people to share the information they have. Aftershock is working on a number of decentralized models in which people can bring information and report what is going on, even when communication disruptions and other obstacles are in place. We are all familiar with the “telephone game”, where information is passed on from one person down a chain and by the end the it is completely changed. This demonstrates the inherent problem of second-hand sources, which the Government, media and NGOs almost entirely rely on for their data. We want to allow individuals that have first hand information to be able to share it, like wikipedia and indymedia, in an uncensored way. We believe that if there is substantial first-hand information available, the need to use possibly inaccurate second-hand sources will be greatly diminished. This will also allow the most up-to-date information since it will not have to be managed before being released. We all know that during a crisis timely information is crucial.

Being able to accurately provide first-hand information to the public will require a big and simple voice. That means, we have to come up with ways to take decentralized reports and expand their reach as quickly as possible. Broadcast radio seems ideal for this type of timely dissemination. Relatively simple technology would allow an am/fm or shortwave broadcast to reach a large number of people spread out geographically. Radio technology is relatively efficient and the power requirement would be minimal and could be kept operational with a generator or alternative energy resources. Radio has the advantage of allowing anyone with a radio or within hearing distance of a radio to obtain information simultaneously. A crisis center, or any other meeting place, could with one battery/crank/solar radio provide information to hundreds of people at once without using precious resources or labor. To collect information the use of small digital recorders and other devices could reduce the time of writing or reproducing information and could easily and quickly be disseminated by a transmitter. Radio broadcasts also allow people to obtain information while doing other things (e.g. first aid), requiring no focal point and minimal attention drain. It also allows people to take their information, if they have a radio, and be mobile with it. To be truly useful, the information will need to be in languages used by the folks in the area. First hand accounts, will aid this by allowing participation by various segments of the population. It also makes literacy a non-issue. Psychologists have done studies on freshman students (one might assume they are a bit more literate than some general populations) and found they can retain 50% more information from audio resources than written resources in the same amount of time. This maximizing of information load could be crucial for people during a crisis and involved in decision-making.

We all know that government agencies and relief organizations use a top down decision model , often called a command & control model. The problems with this model should be obvious to anyone who studies disasters. C&C models tend to be slow reacting, when time is of the essence, and often lack appropriate flexibility. Flexibility is key when dealing with disasters because of the large amount of unknowables. C&C models do not promote individual and/or small group initiative. They put a primacy on analysis over on-the-ground knowledge. C&C models are much better at reacting to predictable events where people have been trained for specialized actions. Much more could be said about the problems of C&C models (e.g. overlapping authority) and management systems in general. For more information on the problems of C&C models see our other article (___________).

Aftershock believes in a participatory and horizontal (P&H) model. The P&H model allows as many people as possible to participate in decision-making. Participation is useful during disasters because all researchers agree that being actively involved in decision making can actually reduce the harmful psychological effects of emergencies on ordinary people. This beneficial psychological effect is not the only reason to use a participatory model. As _____ pointed out crowds are often smarter than the smartest person/people in them. Crows have an eerie ability to know more than the individuals that make it up. This has to do with the aggregation of information and the ingrained desire for mutual aid/cooperation. Since all the participants are deeply invested in coming up with the best solution to their own problems they will minimize damaging decisions and maximize decisions that can aid them in reaching their goals. Recent research in self-organizing models suggest an emergence (moving from simple to complex) occurs when certain principles are met. These crucial principles are: feedback, size and egalitarian power-sharing. Feedback and size are related, if the size of a group gets too big feedback systems are less effective. Research shows that humans can reasonably “read” others in groups of about 60 and when groups get bigger the feedback system breaks down. When feedback is no longer working all sorts of negative characteristics manifest: aggression, suspicion, gossip, isolation and so on. Size also effects how coordinated a group can be. One can think of a flock of geese, they can engage in complicated aerial maneuvers in flocks of about 20 but bats can coordinate their actions in the thousands. If you double the size of a flock (geese or bats) their coordination drops substantially and they will almost invariably break into smaller groups.

The P&H model must allow for groups that get too big to break into smaller groups so that the feedback system and coordination can continue. Flocks, herds, ant hill, bee hives and so on show that key to maintaining complex behavior (and high degrees of coordination) is egalitarian power sharing. While it is true many social animals, wolves and baboons, exist in rigid hierarchies, their ability to be flexible is seriously compromised. Biologists have noted that in times of environmental changes wolf packs and baboon troops actually become less hierarchical. This probably serves a variety of reasons that allow these groups to maximize their adaptability.

Egalitarian groups are more adaptable because they can break up and reform. One can think of a flock of birds when attacked by a predator like a hawk. The hawk will swoop in and the flock disbands and then quickly reforms, over and over again. This both tires and confuses the hawk and minimizes the damage it can do to the flock as a whole and increases the survivability of each individual. The fact that any faction has the same ability to decide how and when to reform with the other splinter groups allows it to be very adaptive. This also plays out during disasters. There is a myth that disasters are just emergencies writ large. Some disaster experts, like NYPD chief Kelly, believe disasters are “like huge car accidents”. They believe that disasters are simply a matter of scale not quality and this can be a costly mistake (though common of C&C models). Hierarchical groups, with power concentrated in a minority of individuals, allow for greater disruptions. What C&C models often don't take in account is disruption among their own system. If 10% of professional emergency responders can not get to the disaster site or report to duty how will that effect the chain of command. The loss of an individual or a group of individuals is not evenly distributed among the whole and thus can have devastating effects at both end s of the chain. There is the problem of decapitation of course, when the leadership is unable to communicate its commands to the subordinates but equally problematic is when subordinates are not around to carry out the tasks of commanders. During Katrina and 9/11 we saw this play out with devastating effect. In 9/11 the Office of Emergency Management was in the World Trade Center and thus commanders could not get together and send out commands. In Katrina over 50% of patrolmen did not report to duty so commands went unheeded. Sharing power, like sharing information, allows for more balanced decisions to emerge. Extreme errors in judgment can be checked if there is a horizontal structure that might be missed with a few commanders calling the shots. Sharing power allows actions to be more timely since the center of decision making is with the group. In C&C models groups must await orders before acting even if they know what to do, this delay can be catastrophic (eg helicopters being grounded during Katrina). Sharing power also increases investment in the decision, creating greater cohesion and coordination. When people are not invested in a decision other influences must be brought to bare to ensure compliance like training, retribution and reward. These types of influences may be difficult to muster during a disaster especially if it is wide spread and unpredicted. People who feel invested in the decision will more likely go out of their way to ensure its success even at sacrifice to their own immediate needs or concerns. We believe consensus is probably the best way to ensure maximum input, participation and investment in the decision making process. Many believe that consensus is not effective is large and diverse groups but research doesn't support this proposition. Experiments conducted at Stanford suggest that diversity has little impact on a groups ability to reach consensus regarding concrete goal-directed actions. Groups of students were divided in two groups after taking a battery of personality tests and surveys. Group A was made up of people with similar profiles while Group B was purposefully made up with people to maximize diversity. Each group had to use consensus to work their way through a digital maze in a certain period of time, if they succeeded they receive a monetary reward. While both groups were able to succeeded at the task, Group B out performed Group A every time in terms of speed and accuracy (while reporting it a more difficult experience). So consensus may be filled with arguing and frustration it doesn't seem to negatively impact the actual decisions being made. Another common criticism is that consensus takes too long. While it is true that a dictatorship is the most efficient decision making in regards to time (assuming the dictator has an efficient way to disseminate their commands) most other decision models are not significantly faster than consensus. According to the same Stanford researchers voting takes about 80% of the time as consensus, when they added a majority voting system to the groups. When there is significant time pressure consensus groups can make decisions in a reasonable amount of time. The subjective experience of time is very different between voting and consensus. The Stanford researchers found that the voting groups could accurately assess the amount of time their decisions took, while the consensus groups invariably believed their overestimated the time it took to make decisions. This may be the result of the reported frustration of consensus building. This will be a stumbling block for the P&H model in that many people will at first be resistant to consensus (despite its practical benefits) and exposure to the process will not necessarily engender satisfaction with this mode of making decisions. We believe this problem does not outweigh the benefits and we will need to do more research and thinking about how to reduce the frustration with consensus decision making.

The P&H model also has the added benefit of best utilizing the skills and knowledge of an ad hoc group. It is difficult for C&C groups to effectively gage and utilize the abilities of volunteers and thus prefer to use only known subordinates (trained professionals). That is why there is so little effective inter-agency cooperation between both government agencies and NGO relief groups. By creating and utilizing an open P&H model individuals can more easily share their abilities with the group. A decentralized system allows for greater and more timely feedback loops to allow groups to adjust to their strengths and weaknesses. This can be crucial in the ever shifting landscape of a disaster. Individuals in a group can quickly replace someone who is not able to do the job and switch roles more fluidly than in a C&C system that doesn't allow for such transitions.

Additionally, the Aftershock model allows for greater community empowerment. People who have the experience of direct decision making during a crisis are less likely to abdicate that collective power to authorities. Joint decision making also creates important social bonds between individuals and groups that can be used to network knowledge and resources inside the community. We believe by supporting and participating in community self organization efforts we can foster greater possibilities of active resistance. Active resistance can take the part of protecting a community from destructive and exploitive recovery schemes and/or be used to challenge generalizable oppression. A successful community response to disasters can be a powerful tool in creating greater autonomy and strengthening the interconnectedness of dependent and fractured neighborhoods.

1Indymedia is made up of non-specialists who run a web-site where anyone can post information about a protest.