CGNet Swara | News & Blog | What's new and exciting with Swara…

Last weekend I was in Abu Dhabi attending the New York University (Abu Dhabi) Hackathon. Over 50 students from countries across the Arab world, including Palestine, Syria, Egypt and of course the UAE along with team leaders from across the world (yours truly included) came together for one weekend to put their brains and skills into a pressure cooker with some coffee, more than a little nicotine and the fantastic food provided on campus to see what revolutionary technological solutions we could come up with to move the world forward.

The event opened with panel discussions where panels of experts answered questions from the participants. On the panel discussing hacking (i.e. developing, modifying, tinkering, not breaking and entering) for social good panel, I was joined by Will Pate from Random Hacks of Kindness, Jay Bhalla from the World Bank and David Hutchful from the Grameen Foundation. One of the key discussion items was the importance of field testing and getting to know the users when developing technology for social causes. Since impacting society through technology requires great user adoption across a wide spectrum of society, social technology projects typically require much more effort in the design phase. For one, unlike commercial projects where the customer has a clearly visible vested interest in defining requirements in a structured manner, social projects typically cater to a user base that is unaware of the benefits and therefore still needs mobilizing. In such scenarios, a lot of guesswork goes into designing the initial implementation, much of which is discarded as soon as field testing begins. The consensus on the panel was that the best strategy is to “release early, fail fast, fail cheap and keep getting up until you get it right”

After the panel discussions we proceeded to individual presentations by many of the team leaders. I particularly liked the sessions on Python development (Mohammed Khatib) and HTML5 (Jeremy Johnstone), since they were immediately usable. David Hutchful’s session on J2ME development for feature phones was also great.

I presented Swara and demonstrated the platform using a local SIM card. Despite the local provider (Etisalat) refusing to forward DTMF tones during the demo (even though I tested it beforehand, this is an instance of the famous Demo SNAFU phenomenon that I will write about one day) the participants enjoyed the session and I got several very well thought out questions from the students, going beyond technology and grasping the idea of community and networking.

Slides

Post dinner, we got down to the business of pitching ideas and splitting off into teams. This is always the second most exciting part of any Hackathon (or similar code jam event). This is where everyone goes wild, throwing figurative noodles on the walls of everyone else’s psyche, just to see what sticks. The criteria were the usual suspects, innovation, impact potential and sustainability.

Several ideas were pitched, from the simple social network or media sharing website to crowdsourced police surveillance using smartphones.

Once the ideas were pitched the team leaders helped refine them and build teams around the most promising themes. Usually, Hackathon pitches are an overcomplicated expression of a relatively simple idea. Its possible to gauge the relative experience of the participants just by hearing the pitches. More experienced participants usually realize that its far more productive to pitch a concept and leave most of the implementation details vague during the pitch, since as the teams are formed and the team members discover each others interests and abilities, implementation details change. This is all the more underscored by the fact that its far easier to alter the pitch to the available skill set than to learn a fresh set of skills in the short time available. The goal is to get to a prototype using whatever is available. Elegance can come later, hopefully with funding.

The pitches at the NYU Hackathon reflected the fact that many of the participants were students, including several sophomores. The underlying ideas were strong, but in several places, the implementation pitched was unnecessarily complex. This was quickly remedied by the team leaders who brought their own experience to the table and helped the teams to simplify the projects down to implementable size.

Coding began in earnest next morning. The team I was working with (Abdul Sartawi from Palestine, Hassan Mousa from the UAE and Saleem Adele from Jordan) was looking to develop a smartphone app that would use the camera in combination with an OCR to enable blind people to read. Refining this further, we expanded the target user base to include illiterate people and added an IVR interface, which we planned to build using the Swara platform.

2 days of code, arguments, near team break-ups, coffee and very little sleep later, Qare was ready to test on D-Day (Demo Day) morning.

Qare is a 2 part system, one for content creation and aggregation and one for dissemination. The collection piece consists of an Android app, which we deployed on Saleem’s Samsung Galaxy tablet. The app is an OCR, that uses the camera to pick up printed text and convert it to a digital form. We then use a Text to Speech (TTS) engine to read out the digital text, creating an audio file.

This audio file is in turn uploaded to Qare, using the Swara-based admin portal. Once published, the file becomes available for playback on the web or on the IVR.

We modified the Swara IVR to remove the recording piece and added categories to the main menu. We also re-recorded the prompts in Arabic (Thank you, Sana Odeh ).

Callers use the IVR to navigate to a particular title. Once there, the user can either listen to a short summary, or choose to have the file sent over via MMS.

This sort of service could be used by individuals or organizations to host books either for free or at a nominal price, making a world of knowledge available to people who would otherwise have to spend years learning to read before they could access it.

The team presented the application to perfection during the final showdown and the demo went perfectly. No winnings this time, but as the judges told us later, we all felt like winners that evening.