This week, I’m going to talk about exactly what it is I’ve been doing these past two months. My major goal was to set up a data collection system for Nuru. We had decided that phones were the way to go, but which phones, and what system still needed to be answered.
For the phone search, I went to Kisii, the nearest city (>100,000 people) and I selected a few phones that met my needs. #1 Internet (GPRS) #2 Java (ability to install software) #3 Camera #4 Not a “China Phone” (a brand-less or knock-off phone) #5 Were not way more expensive than other options. I found four phones which met those criteria: Nokia 2330, Nokia 1680, Samsung C-450, Samsung E2120.
I bought four identical SIM cards, and began the testing. The field testing consisted of their ability to connect to the Internet, a subjective test of their user interface, and their ability to connect to the web at all the important locations in our area (and some intentionally low reception spots). I also did an aesthetic survey, showing them to those who would be using them and asking them which they would prefer. I installed the relevant software on them (or tried to; Opera Mini 4.2 for Google Forms, and SMS:Frontline Forms). I then gave points for features above and beyond the 5 listed above (e.g. one had a microSD slot, some had video cameras, etc). I was concerned that my arbitrary weights and criteria would leave me without a conclusion. But I was mistaken, as the Nokia 1680 blew the others out of the water; it was the best in almost every single category.
Having selected a phone, I began to test the platforms and do a cost comparison. The major competitors were Frontline SMS (the leading SMS-based data collection tool) and Google Forms (a web-based data collection tool). The capabilities of Google Forms far exceeded Frontline. The major benefit of Google is that it’s “in the cloud,” that is, it’s stored “on the Internet” and not on any one physical computer. Which means that a computer crash (as happened to me yesterday) wouldn’t destroy any data and that data would be accessible in more than one place. The other factor was operating costs, and with Google Forms, they’d be about one hundred times lower than with Frontline. Data is super cheap, about a dime per MB, and form submission is about 5KB; SMS would cost $0.03 each form.
Then came the process of looking for a place to buy the phones. I went with Philip to Nairobi and looked for a place to buy phones. I went around the city with Philip’s son, Laurent, who knew the city well. After realizing that the Nairobi phone book was a joke (I gave up after 0 of 15 numbers connected to a business in the category I was calling), and that all the retailers in Nairobi were getting their phones from the Safaricom dealer, we went there. The phone retailed at 3000/- (~$40) and they gave us a whole 50/- discount when they heard we were buying 100 pieces. So we kept looking. My friend from Stanford put us in contact with people from the University of Naibori, and they tipped us off to a wholesaler. And they cut us a deal: 2850/= per piece. And since we were new customers, they agreed to ship them for free.
After this, I had to figure out the best configuration for the phones. For network, I decided to start with Safaricom as it had the clear majority of the local business already. I wanted people to keep their phone numbers and not have to change networks on top of all this new phone business. I realized after buying 100 phones, that the browser I installed (Opera 4.2) would crash fairly regularly when it viewed a Sheet. While not a critical operation (our main goal was data input), it would still allow us to do a great deal more if we could view the data that was submitted. So I started scrambling. And eventually found a link to an older version of Opera (2.0) that worked and wouldn’t crash and ran faster on the poor under-powered phones.
While Google Forms can work with any Google account, it works even better with a Google Apps deployment because you can securely record who submits data. Our amazing tech guy Will Kerr registered http://nuru.co.ke for us, and linked Google Apps to it. I created a very basic Google Site to allow for easy navigation of available forms. I then registered accounts for all the staff and started the tedious task of programming phones.
I know there’s probably a way to clone phones using USB, but after hours of trying, I had failed completely; USB is technically unsupported on the phone and I’d have to hack it, even assuming I could find the right cord (which is no easy task). I decided that doing it one by one was the only option. So I used the phones’ email client to access their new @nuru.co.ke addresses, installed Opera, added a bookmark to the Nuru site (thank God for bit.ly!), and for kicks, added the Nuru logo as the phone’s wallpaper. Wash, rinse, repeat… a hundred times. I got really good towards the end. In my personal time trials, my record was 8 minutes, 9 seconds for a complete setup. I eventually realized it was mindless enough work that I could learn about God, ethics and human nature simultaneously by listening to the Brothers Karamazov on MP3 while programming the phones.
Last of all was training. I trained and trained and trained. For most of them, they had never used a computer before this. And now they had to grapple with concepts like “megabytes” and “internet.” It was slow going, but rewarding. My lesson was once interrupted by a spreading wave of fake shutter sounds, as one person figured out the camera and then spread the knowledge like a virus. I’ve heard numerous accounts of people Googling and Wikipediaing late into the night about topics as varied as “rich people,” and “mountains.” These people had never even had access to a public library, and now they have all the world’s knowledge open to them. So training took a very long time with each group. And even after hours and hours of training, people are still uncomfortable with the phones. And this is why I’ll be spending a few more weeks on training when I come back.
In addition to training the Kenyan staff, I started showing the Foundation Team about the phones and about Google Forms. And I experimented with the Forms myself. I found that there is incredible power to be unlocked in Forms, particularly in the ability to link different Google Sheets and display the output in an HTML table (essentially it can pretend like it’s a database). This means that we can have all of our data on our handsets.
So that’s what I did the last two months. It’s been a really incredible experience. Looking back on it, I’m excited about the state of the world. This would have been impossible even last year. Google has invented a system for easily collecting data. Safaricom wanted to make money selling phone internet access. Years ago, Opera made a way to view web pages on a phone. Nokia built a high-quality phone at a low price. And I, with no special experience or programming knowledge, am able to piece together these technologies to make a cheap, real-time data collection system.
What’s so exciting to me is that I’m nothing special; anyone could have done this, and done it cheap. It only cost Nuru the cost of my time to set this thing up, $30 per year to register a domain, and $40 per phone with about $1 per month in data charges. There are no development costs, no programmers to be paid, no server costs. Anywhere there is GPRS internet (which is quickly becoming everywhere), this kind of system could be set up. This blog isn’t patented. So please, rip off this idea! And encourage others to, also!