On 1 December, nature decided to drop 30cm of rain in a 24-hour period over Chennai. It was the heaviest downpour Chennai experienced in a century, and the city and its people were haplessly unprepared for the record deluge.
I live close to a suburb that has the Tamil word for lake (Eri) in its name, so flooding is not uncommon because water tends to have this terribly annoying habit of nodding its head to gravity at every chance it gets.
Yet, the story of the latest flood is not a simple linear narrative of callous government officials, poor urban planning and marvellous self-organizing social media-driven help networks. As with most things in India, it’s complicated.
In September, the local folks in the Public Works Department decided to do something uncommon. Their job. They desilted every storm drain, elevated every street and, amazingly enough, collaborated with another government department, the Tamil Nadu Electricity Board.
Almost every junction box was raised to keep it above water level in case of floods. These were model local government employees, that kind that most people in India assume do not exist.
On 30 November, when that famous unending shower began, I was, as I usually am, in an airport. On my way back from the Chennai airport after having landed quite turbulently on runway No. 2, which in 24 hours time would be inundated by the Adyar river over which it was built, I asked my Ola cab driver how long it had been raining.
Since the morning, he said with great annoyance. Although, more than the rain, he seemed more irritated by the fact that there was no surge pricing in effect despite the shortage of cabs at the airport. This lack of real-time insight is quite relevant to our story, but more on that later.
I reached home to find knee-deep water levels just threatening to enter the house but, crucially, functioning power and broadband Internet. A less fortunate neighbour already had knee-deep water in his ground floor, so he took his family and moved to a relative’s place in an apartment nearby. Then we lost power.
But this was perhaps the best possible experience one could have had as a ground-floor resident in Chennai on 1 December. If one lived in the suburbs abutting the Adyar or Cooum rivers or in the suburbs that used to be lakes before greedy real estate developers (and greedier citizens like us) turned them into urban jungles, it was likely a nightmare.
Water levels in areas like Jafferkhanpet and Ekkaduthaangal rose to 6 feet in a very short span of time. Six feet of water on the ground floor essentially means that you have lost pretty much everything in it. Furniture, electronics and even cars. Almost all families in these areas quickly moved up to the first floor, and in places like Mudichur, there was ankle-deep water on the first floor! Houses without first floors meant that entire families were stranded on roof tops and terraces.
Power of open source, open data
But on the night of 1 December, the world outside Chennai had only two sources of information about the city, and it wasn’t TV, radio or any public warning system. It was people sending frantic SOS messages over SMS, WhatsApp, Facebook or Twitter. The second source was non-resident Indians and people not in Chennai who had folks in Chennai they couldn’t reach. By late evening, a lot of mobile towers had run out of inverter battery power and phones desperately trying for that one bar of signal drained themselves pretty quickly.
So, complete radio silence from someone you know, it turns out, is information too, albeit a dangerous double-edged sword. So, that was when the deluge of rescue requests started hitting social media, and Sowmya Rao, a New Delhi-based lawyer with Chennai roots, decided to do something about it.
In a few hours, there were practically thousands of “stranded family in Velachery, need rescue”, “1,000 people in Jafferkhanpet about to drown in 8 feet of water, need rescue right away” posts started surfacing. And since the flooding was quite asymmetrical, there were many from less flooded parts of the city who quickly started responding to these rescue requests as local aid volunteers.
With help from a bunch of techies in New York and Bengaluru, Rao quickly put together a Google Docs spreadsheet that consolidated all these rescue requests and also contact details of volunteers who were willing to help. This seemed like a sensible thing to do because there was absolutely no guarantee that the person who posted the rescue request would somehow automatically see a relevant local volunteer willing to help. So, the idea of centralizing these requests and offers made sense.
But pretty soon, there were hundreds trying to edit that sheet and it was getting out of hand, which is when Karthik Balakrishnan from Bengaluru put together a simple search interface and a submission form in front of the spreadsheet and hosted the whole thing onchennairains.org. And all of this happened within a matter of a few hours.
Soon, there were other kinds of requests as the next day rolled around and in addition to rescue, there were requests for water, food and medical support. The number of volunteers on the site was now up to about 25 and it was nearly evenly split between people in India and those abroad.
Changes made to the website were tracked and managed using GitHub and there was even a smooth IT style offshore-onsite handover every 12 hours that just emerged out of nowhere.
And this is really the power of open data and open source. The idea that someone who was making 500 food packets for people in need can do so without having to nail down ahead of time who needed the food. All they had to do was make that information public and it could be matched up with an actual need somewhere.
Uber for aid
And the fact that an entire functional search interface powered by a shared spreadsheet behind the scenes cloud be hacked together overnight using open source components like Twitter’s Bootstrap is something unique to our times.
It was also absolutely fascinating to watch these groups self-assemble. At its best, the site felt like an Uber for aid. Those who needed aid could add their request and those who could offer it could be matched and dispatched. And all of this could be coordinated largely by people not even in Chennai!
One set of techies mapped incoming requests onto Google Maps, which gave us a better sense of where the hotspots were—Jafferkhanpet, Ashok Nagar, West Mambalam, Srinagar Colony—and so on. It was also very easy to try out new processes and dump them if they didn’t work. We tried assigning people to specific areas, but after a couple of days, we realized that some of the areas were largely back to normal, so it didn’t make sense anymore.
There were just so many high-five moments. An elderly man in West Mambalam who needed oxygen cylinders managed to get them on time thanks to an NRI finding a source for them on our site. A mortuary van was arranged for someone who had passed away while the area was still flooded. A source for 200 blankets was matched with a colony of retired policemen and medicines were directed to an orphanage that was hosting an entire nearby colony that was washed away by the Cooum river.
But about 48 hours in, we ran into problems. We had, by now, divided ourselves into three teams. Social Media Trawlers, who would monitor the @chennairainsorg Twitter handle and Facebook page, specific hashtags we had popularized (like #chennairains) and make those entries into our master spreadsheet; on-the-ground volunteers who were covering specific suburbs in Chennai; and operations folks who would take these aid requests, find the closest on-the-ground volunteer and coordinate with them.
The first problem was social media noise. Well-meaning slacktivists who wanted to help but couldn’t be bothered with exercising anything more than hitting the RT or share button kept re-sharing the same old aid requests. An example—“An old-age home on ECR has 60 starving old people who need food!! Right now!! Urgent”. Problem was, the home was quite well served by two local aid agencies from the moment the flood hit.
Another example—“Pregnant lady in Mudichur in chest-deep water with escaped crocodiles nearby!!”. The sheer human interest in a nine-month pregnant lady made the story grow with every RT, share and forward.
Except, the army had already airlifted her to safety on day one and she had given birth quite normally, but that didn’t stop a lot of people from wanting to still play a part in rescuing her by the cunning use of a smartphone, thumb and forefinger. Despite the Madras Crocodile Bank Trust reassuring everyone that absolutely none of their reptiles had escaped, they had, nonetheless swum their way into WhatsApp forwards.
So, the very first real-time reorganization we were forced to do was to put in place a verification army. It was a group of outstation folks (since they had power, landline telephones and a complete lack of knee-deep water in their homes) who would call the numbers being shared on social media to verify if the problem or need actually still existed.
We made a call to the old-age home. The exasperated lady there said (and I paraphrase)—“We are getting food from two aid agencies. Do you have anything else to offer? Like gobi manchurian?”
And this was indicative of a larger problem. Social media was amplifying aid requests from a small part of the city. The wealthier, social media-savvy side. There were several parts, particularly in north Chennai that were in scarily bad shape and had no one tweeting or Facebooking on their behalf. Social media, we realized, was a middle-class-centric echo chamber.
The second problem was NRIs. Since my phone number was listed as one of the on-the-ground volunteers taking care of Anna Nagar, I (and several others on the volunteer list) received, on an average about 20 phone calls at night from people in the US, UK and Australia.
There are parts of Chennai, like West Mambalam and Ashok Nagar, which have a disproportionately large number of old folks whose children live abroad, and the moment they realized that there was only one tick on their WhatsApp messages to them, they hit the panic button and started calling every number they knew in Chennai, and stressed an already flaky mobile network.
A deeper problem
This was my favourite—A lady from Phoenix, Arizona, called me and said: “My parents are in 12th avenue, Ashok Nagar (which was under 6 feet of water and only the army had access till almost Saturday), so can you patch me through to the government rescue officer in charge of 12th avenue?” I almost wanted to ask her if she wanted fries with that. But frivolity aside, there is a deeper problem here.
Our collective need for information was actually more than the need for aid. NRIs simply had not heard back from their old folks in West Mambalam and Ashok Nagar for 24 hours, and that was enough for them to assume that escaped crocodiles that had mutated thanks to submerged electrical wiring were on their way to devour their parents.
And together, they were calling any number they could find, often on-the-ground volunteers whose phone battery lives were precious in a city that was largely dark and wasting their time with personal “go check at this address” requests. One of our volunteers even made it through to one of these places in hip-deep water only to find two perfectly fine old folks who asked the chap to go get them some fresh milk if possible.
I can understand how anyone would feel sitting 10,000km away and not knowing what is happening to their folks back at home and the only thing you see on the news media is looped footage of a collapsing bridge and boats being operated on what used to be streets.
They were getting either amplified bad news or no news, so they simply trusted every WhatsApp rumour that came their way. And oh boy, did those rumours keep coming. Lakes were breaching all the time, El Nino was threatening more cyclones and Nasa had shifted focus from Sanskrit to Tamil.
But something heartening from this entire disaster was the sheer scale of volunteering. The rains had barely stopped when literally hundreds of cars with food and other aid had left Bengaluru for Chennai. In many relief camps, there were volunteers competing to supply food, blankets or sleeping mats.
But at the same time, our exercise with crowdsourcing also taught us some hard lessons about the challenges of efficient distribution when there was very unreliable and skewed on-the-ground information. I went to relief camps with blankets and found everyone had two blankets already. At the same time, there were SUVs from Bengaluru roaming the city with chapatis but had no idea where to deliver.
Making a difference
It was also incredible how 200 strangers (yes, that was the final size of the chennairains.org online volunteer team) literally formed an organization where people just took up roles that emerged in real time without any petty infighting.
Every night, we would discuss and decide what the focus was going to be for the next day. No more food aid, we decided on day three when clearly, blankets and medicine became a bigger priority.
The tools we used, such as Twitter, Google Docs and Slack, give a distributed team the unprecedented ability to spontaneously assemble and get things done. And GitHub helps us build new tech without having to invent it from scratch.
The source code behind chennairains is open source and is available for reuse by anyone wanting to run a crowdsourced relief operation the next time this happens anywhere.
But along the way, as we pat ourselves on the back for making a difference, for being that information broker that cut the time between the need for aid and the provider of aid, we realized a few sobering things.
I was at a relief centre in Aminjikarai, delivering milk powder to a camp filled with kids when this lady told me “It’s only when disaster strikes that you middle-class folks pay attention to us. Our lives are not any better when there is no disaster. But now it’s a bonanza with you guys competing to give us stuff.”
All the pedantic analysis of uncontrolled building on marshland areas, climate change and uncoordinated government departments aside, the uncomfortable truth this incredibly huge army of volunteers is likely to conveniently ignore is that the poor in India live in conditions that no civilized society should tolerate.
The slums on the banks of the Cooum have always been mosquito-infested hellholes and now they are mosquito-infested sludge-filled hellholes.
Another realization was the fundamental asymmetry of delivered aid, at least in the first few days. As much as we wanted to direct those cars from Bengaluru to north Chennai, no one had a clue who needed the aid or whether the roads to Kargil Colony in Tiruvottiyur were even motorable.
We also realized that in the entire group of 200, there was not a single public health expert or even someone with experience in relief operations. It was a bunch of really passionate folks figuring things out on their own, and while definitely making a difference, could have done so much more if governments and institutional experts in relief were less sceptical about social media and its ability to connect people.
We had no access to information from the National Disaster Response Force in terms of where they were operating, who they had already rescued and where food packets were being air-dropped.
And this brings me to my point about open data and surge pricing. Self-correcting, efficient relief operations need open data, as much as Ola and Uber need better quality demand and supply data to price their cabs meaningfully.