I could not submit my vote as the time was closed. I am sharing my perspective below after reading all the comments and the original post.
The voting distribution in this poll exposes a challenging dilemma for the movement. On one side, several voters point out that we cannot confidently verify the absolute direction of our impact on the ground, creating a high risk of wasting money or accidentally causing net harm. On the other side, it is rightly noted that building effective field organizations is slow and difficult. If we starve field teams to fund data science, we risk losing the operational vehicles needed to execute better strategies down the line.
It is always good to learn from other movements where similar challenges were faced.
We can see this exact tradeoff by looking at the history of education reform in India, particularly the journey of the NGO Pratham and their Teaching at the Right Level model. For years, the global development community debated whether money should be spent directly on teaching children or on conducting rigorous evaluations to see if the pedagogy actually worked.
If early funders had treated this as a zero-sum choice, the program would have failed. Shifting all funding to pure academic research would have killed the field team, leaving no one to actually teach the children. But shifting all funding to blind scaling would have been just as dangerous. Early versions of the program actually failed to show impact when integrated directly into standard school hours because teachers felt forced to stick to the rigid official curriculum. Had the team just scaled up blindly without real-time data, millions of dollars would have been wasted on an ineffective approach.
Pratham resolved this by turning their active field operations into the research loop itself. They deployed teaching teams to run short, targeted learning camps outside school hours, testing the pedagogy in real time while collecting immediate data on student progress. The field team was not just executing an intervention blindly, and the researchers were not just sitting in distant offices. They integrated the evaluation directly into the daily operational DNA of the field vehicle.
Instead of treating evidence building and field execution as a zero-sum funding tradeoff, the solution is to treat them as an integrated pipeline. The focus should be on shifting how field organizations operate during their initial phases.
Instead of building organizations that optimize blindly for rapid scaling on day one, early field deployment should be designed intentionally as an active data gathering exercise. When a field team treats its initial footprint as a live research engine, the conflict between research and action disappears. The execution itself becomes the primary tool used to eliminate the exact impact uncertainties the community is worried about.
If we do not integrate research directly into the daily operational DNA of our field teams, we will remain permanently stuck in this loop. We will either be driving blind with great execution teams, or map making with no vehicles left to drive.
I could not submit my vote as the time was closed. I am sharing my perspective below after reading all the comments and the original post.
The voting distribution in this poll exposes a challenging dilemma for the movement. On one side, several voters point out that we cannot confidently verify the absolute direction of our impact on the ground, creating a high risk of wasting money or accidentally causing net harm. On the other side, it is rightly noted that building effective field organizations is slow and difficult. If we starve field teams to fund data science, we risk losing the operational vehicles needed to execute better strategies down the line.
It is always good to learn from other movements where similar challenges were faced.
We can see this exact tradeoff by looking at the history of education reform in India, particularly the journey of the NGO Pratham and their Teaching at the Right Level model. For years, the global development community debated whether money should be spent directly on teaching children or on conducting rigorous evaluations to see if the pedagogy actually worked.
If early funders had treated this as a zero-sum choice, the program would have failed. Shifting all funding to pure academic research would have killed the field team, leaving no one to actually teach the children. But shifting all funding to blind scaling would have been just as dangerous. Early versions of the program actually failed to show impact when integrated directly into standard school hours because teachers felt forced to stick to the rigid official curriculum. Had the team just scaled up blindly without real-time data, millions of dollars would have been wasted on an ineffective approach.
Pratham resolved this by turning their active field operations into the research loop itself. They deployed teaching teams to run short, targeted learning camps outside school hours, testing the pedagogy in real time while collecting immediate data on student progress. The field team was not just executing an intervention blindly, and the researchers were not just sitting in distant offices. They integrated the evaluation directly into the daily operational DNA of the field vehicle.
Instead of treating evidence building and field execution as a zero-sum funding tradeoff, the solution is to treat them as an integrated pipeline. The focus should be on shifting how field organizations operate during their initial phases.
Instead of building organizations that optimize blindly for rapid scaling on day one, early field deployment should be designed intentionally as an active data gathering exercise. When a field team treats its initial footprint as a live research engine, the conflict between research and action disappears. The execution itself becomes the primary tool used to eliminate the exact impact uncertainties the community is worried about.
If we do not integrate research directly into the daily operational DNA of our field teams, we will remain permanently stuck in this loop. We will either be driving blind with great execution teams, or map making with no vehicles left to drive.