Notes from building Machli — a digital product for fishermen
At Reliance Jio AI/ML CoE division, I am a part of a team called “AI for India”. Our team’s motivations are towards using technology and AI/ML to build products that can be useful for non-English speaking rural people particularly in the fields of agriculture, healthcare, education and overall life improvement.
We recently worked with Reliance Foundation team, who released an Android based mobile application called “Machli”, which is helpful in the livelihood of fishermen. Let’s take a moment to understand the target users of Machli. According to CMFRI (Central Marine Fisheries Research Institute), there are around 4 million fisherfolk population comprising in 864,550 families. Nearly 61% of families are below poverty line (BPL).[1] India has a coastline of ~7500 kms, almost half of total land border of ~15,200 kms. The coastline is spread across 9 states, and cultural diversity across these states is impressive. Each state has one major language understood by majority of the population. Going from Gujarat to West Bengal, we have Gujarati, Marathi, Konkani, Kannada, Malayalam, Tamil, Telugu, Oriya and Bengali.
Breaking the English exclusivity of digital space : The two levels
If you are reading this article, you have had a smooth transition into digital space. Internet and digital space has potential to be a great leveller to enable opportunity access and removing information asymmetry. But, being able to read and type English has been an unmentioned pre-requisite in the promise of internet. There are two levels of expectations here.
- User should be able to understand English.
- User should be able to read/type English.
Basically, everyone outside tier-1 English educated India (on the left) has to cross two levels to benefit from digital revolution on the right.
If you think about it, the first assumption is separate from second assumption. I understand Punjabi language if someone speaks to me in Punjabi simply because I have grown up around friends from Punjab, but I can’t read/write Punjabi language. Similarly, ‘a fisherman from Tamil Nadu can read/write Tamil’ is stronger assumption than ‘fisherman can understand Tamil’ simply because they have grown up speaking and listening to Tamil. Given that around 61% of our user base is financially disadvantaged, we knew we had to cross over these two levels to build something , and make it accessible and inclusive for fishermen.
Bridging information asymmetry for fishermen
Fishing is an uncertain livelihood. This is because of different reasons. First, typically fishermen go into the ocean for more than a day to catch fish. Ocean conditions are erratic. Some days, the ocean has calm waters. On other days, the wave heights or wind speed can be high and prohibitive or even dangerous for fishing. A lot of stories from local newspapers point out, that a lot of lives of fishermen could have been saved if they knew the ocean state forecast before venturing into the ocean.
Indian government has an autonomous body called Indian National Centre for Oceanic Information Services (INCOIS) which tracks the oceanic weather and activities. Among other things, they release daily weather forecasts for ocean, advance tsunami warnings and potential fishing zones. All the information is published on the website. Our aim is to take the numbers and information from INCOIS and make it available to a fisherman on their mobile device, in a language/medium that they can access and use.
Ocean State Forecast (OSF) tracks the wind speed, wave height and current speed. OSF helps a fisherman decide whether they should venture into the sea. Potential Fishing Zone (PFZ) points fishermen to a location where there’s a high probability of getting fishes. The locations are calculated by analysing satellite data and figuring out the chlorophyll content in the ocean. The fishes go where the food is. And the fishermen go, where the fishes are. PFZ helps a fishermen decide where to go, when they are venturing into the sea.
The result : Machli
Machli is an attempt which combines previous two sections, “Breaking the English exclusivity of digital space : The two levels” and “Bridging information asymmetry for fishermen”.
It is accessible to the fishermen, because it’s in their local language. It does not require fishermen to be able to read their local language. Each screen comes with assistive audio guidelines which tells the fisherman what to do. For example, if we want user to enter their phone number, they’d see the screen below, with an audio telling them to do so.
Second, it provides the OSF and PFZ as discussed in earlier sections which are vital inputs to fisherman to plan their day to day activities. The advisories are in their local language, available both as written text and audio advisory. The advisory is generated across 1200 fish landing centres in India, in all the languages daily. This is important because there are migrant fisherman, e.g. a Tamil fisherman working in Gujarat.
Reliance Foundation has been rolling out Machli gradually to different districts in coastal states, and the response has been positive so far. Over next months, we’ll continue taking feedback from the users and incorporate them into Machli. We will continue to work on such products that bridge the ‘accessibility’ and ‘information asymmetry’ gaps and help people in their livelihoods.
Go ahead, and download Machli from Play store by clicking on this link.
To sign off, here are a bunch of pictures from our campaign on ground.
And yeah, the talented folks that built it.
