Ofcourse, MIS is a reflection of policy and there's directionality. But overtime & innocuously it dictates the conversation, governance and bureaucracy of the program they administer.
While, policy goes through feedback cycles with states, IT requirements almost never.
+
At root, it's because monitoring is often mistaken as governance at top as top can often only monitor.
But IT can do much more: it can help ease paperwork, question redundant but entrenched workflows, mitigate risks etc.
But for that the requirements need to be bottom-up.
+
Not to mistake it for more human centered design etc etc.
Asking the frontline how better to make a form by giving them a voice is a start but I am more hinting towards letting them decide if the form should be built at all and if something else should be prioritized.
+
Should IT invest dev-time in a module that helps reduce the administrative paperwork of a rural engineer so he can spend more time on field inspecting roads
OR
Launch a new app with geo-tagged photographs, linked with ML to catch dereliction of duty by engineers from Delhi.
+
Often in trying to catch the mischievous minority, the administrative burden of the majority dutiful functionaries increases.
More the burden, more humanly difficult it is to follow all your duties.
So your new app might actually increase your problem & not decrease it.
The distance between the developer and the frontline, both physically and in hierarchy is huge. Even if the requirements seep up, they'll be second to what a bureaucrat in Delhi has requested.
Keeping are ears to ground is not enough - smth more systemic is required.
fin.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
Last year, we were hiring for Ministry of Rural Development’s Data & Insights Unit to use data science to responsibly improve rural-dev service delivery across schemes.
Hiring similar posts again (last tweet).
First, here's what the wonderful team did in <12 months:
(1/n)
PMGSY (Rural Roads): We have an AI based warning system which flags suspicious payments comparing images of road inspections with payments being made to contractors for maintenance.
But do warnings work?
(2/n)
To answer that, a state-wide AB test is live to check the impact of warnings (intended and unintended) and critically examine algo-human interactions. We’ll be combining qual + quant research methods to answer that. Very exciting results pouring in.