We are reaching the end of the year, and, as I guess is the case for many of you, I find myself thinking about the events of the year, and reflecting on how they have impacted and changed us.
2010 has been a tumultuous year for us.
Launching Sparkling Logic was certainly the highlight – having an opportunity to work on problems dear to us, with a great friend and co-founder, Carole-Ann, and with a great team of partners in our growing venture, the desire and the ability to approach them in new ways, and with the promise of industry changing products… It’s the beginning of a great story that we will talk more about later.
The year saw the tragedies of the earthquakes in Haiti and Chile. The magnitude of the disaster in Haiti is difficult to imagine, the numbers are staggering, and the realities of the impact on the lives behind those numbers impossible to grasp for us who live so far. Haiti, already a poor country suffering from so many ills, suddenly ceased to function and left millions unable to survive, with no access to basic security, health, food services. I don’t want to bore you with awful details, but those who are curious can check the pictures published by The Boston Globe for Haiti– 24 hours after, 48 hours after, 6 days after, three weeks after. And new pictures would look only marginally better. The quake in Chile was much stronger, but the country was orders of magnitude better prepared to cope with it, and, while it suffered huge damages, it lost relatively few people: the same day, 3 days later, nine days later.
Events such as these tragedies tend to bring to surface both the worst and the best in people. And they do make us pause, and think about what we are about, what we believe in. The earthquakes in Haiti and Chile both taught me once again that human courage and generosity are powerful forces, and that we sometimes have the chance to know and spend time with exceptional humans who display these qualities and illuminate the lives of those around them. They should be cherished and remembered, for they are the best of us. My thoughts go to you, A – hasta siempre.
In Haiti, as is often the case in poor countries, emergency response was difficulty – most of the infrastructure collapsed. But new technologies and access to data and data processing capabilities made emergency response decision making significantly more efficient than in similar situations in the past, saving lives, and giving high hopes for these technologies and approaches to help improve the efficiency of emergency responses in similar situations.
Emergency response in a catastrophe combines the magnitude of the problems to solve with the need to solve them fast, and react rapidly to changing situations. Millions of people in need, people buried under buildings and potentially still alive, aftershocks, etc… Decision making in those conditions is an extremely arduous exercise – decisions need to be made fast, based on partial information and taking into account multi-dimensional constraints (availability of skilled operators, logistics, security, relative urgency, …), they need to be revised almost constantly because of new information, and all that under the pressure that what is being impacted – directly impacted – are human lives.
This field of decision making in emergency response has received some academic focus – mostly in terms of decision analysis. Models have been created to study what factors influence the outcomes of which decisions under which circumstances and they have helped shape field. It is now fairly common to see these models at work in the decisions about the logistics of emergency response – both in terms of preparedness as well as in terms of actual response.
But the situation is a little different in terms of the decisions made on site in terms of what to do where during the emergency response to a catastrophe. Those are different in nature – they take place at the micro-level – discrete human beings are being impacted – and they take place in an environment in which the details that simplifying models discard are key to take into account. It’s not a question of knowing that a segment of people should be sent to a particular type of health service – it’s a question of selecting who between Paul, Peter and Mary to treat first, and selecting where to send him or her, between the local surgeon with no access to antiseptics and the one 3 miles down the unpaved road who may have some.
I have had the opportunity to be involved in only a few catastrophe emergency responses. But let me make two observations drawn from my limited experience with these decisions:
- Incorporating real time local data is crucial in the immediate short term
- Capturing the decision process, the decisions made, and tracking their performance is crucial in the long term
The first point is fairly clear. The decisions need to be made based on good data, and good data, in countries like Haiti with little infrastructure, is necessarily local to where the action is.
Emergency response teams are frequently not local – they often do not even speak the local language and may not have access to basic information such as accurate maps. However, new crowd-sourcing based approaches relying on basic local infrastructure and vast motivated and multi-skilled social networks help alleviate this problem significantly.
In Haiti, while the emergency systems broke down, the cell phone networks continued working, and allowed thousands to communicate reliably through SMS messages. This enabled to quickly (48 hours) set up an SMS-based emergency response system, thanks largely to the work of a team of people connected by Josh Nesbit of FrontlineSMS:Medic.
However, most of the information exchanged through SMS was in Kreyol, not understood by most emergency response teams. To cope with that,
- a system based on Ushahidi was set up to connect volunteer translators located throughout the world who translated the messages, helped geotag them,
- and fed the information to a coordination team
- which sifted through the information made available almost real time to identify actionable information, assess impact and coordinate efforts using the same Ushahidi platform.
The same system was used to crowd-source the refinement of the maps which were extremely imprecise prior to the earthquake. This was done using very simple tools accessible to all, and yet, the result was a very quick progressive enrichment of the details in the maps, enabling emergency teams to efficiently reach their targets.
The platform has also started being used to capture the decisions made and attempt to track their effectiveness through time. That’s a much more difficult task, but one that I believe holds as much promise for the future. It will enable us to form a much better picture of which types of decisions are better suited for which situation based on actual experience. It will foster better models, and result in more efficient emergency response decisions.
To get there, though, we still need to keep working on
- making sure that local data is quickly integrated into the systems,
- providing the right tools to the decision maker to quickly assess the information for decision making,
- making it extremely simple for him or her to capture his or her decision making process as information into the systems,
- identifying how to measure the effectiveness of the decisions,
- and enabling the measurements.
This is quite difficult, but I believe data scientist and decision making specialists can provide the skills needed to get started and extend systems like Ushahidi along these lines. The long term benefits – priceless.
To a great 2011!
Read Carole-Ann’s thoughts here.
Learn more about Decision Management and Sparkling Logic’s SMARTS™ Data-Powered Decision Manager