Jessica Leight

@leightjessica.bsky.social

Development economist, @Yale / @UniofOxford / @MIT. Asso editor, REHO + CER. Editorial board member, BMC Public Health + PloS One. Structural transformation, agriculture, gender, IPV. Views my own

As part of a recent discussion around agentic coding, I developed my "top 10" in agentic coding applications, and thought I would share it here. (Note these are not ranked!) 1) Re-organizing project folders! No more floating, disorganized, mislabeled files!

A fresh take on challenges for funders drowning in a higher volume of applications: at Wellcome trust (where they report 50-100% ⬆️) - if they receive applications above predetermined cap (unclear if it will be public), then they'll randomly select applications to move forward

This one hit home (+ not in a good way) expansion of high-speed internet in Uruguay *worsens* child development indicators. Why? Parents spend more time online, engage in more high-risk activities. from @kari_colombo @EFailache in JDE

BildBild

As a counterpoint to all the posts describing an incredibly detailed + convolute agentic coding set-up: your regular reminder that if you are an applied economist, you probably don't need any of that detail (at least to start)

How many seminars + individual acknowledgments do development RCT papers report? Around 5 on average in a field journal and closer to 10 in top-5-journals, largely following the gradient you would expect . . .

Bild

What does disclosure of IRB approval look like in a recent sample of published development articles (2025-2026)? Not great, tbh. Only about 2/3 disclose IRB approval in full text, for papers published in top general interest journals + JDE

Bild

I like to index my sleep disruption based on how successful I am in using my basic coffee maker in the morning. This morning I removed the basket for grounds to dump it in the trash, and left the entire reusable basket there and just put grounds directly in coffee maker.

This morning I opened the laptop to check email, realized I needed some additional (simple) robustness checks for a paper, opened Claude and entered my prompt; then my toddler climbed on my lap + I cuddled himwhile answering simple Qs so the agent could do tedious coding