
Juliana Pinillos
5.1K posts

Juliana Pinillos
@JPinillosp
Econ PhD Candidate at @USC. Environment+urban economics. 🇨🇴 Colombiana. RT≠ Endorsement



Everyone gives lists and long essays on how to use Claude Code or Codex. I only use Codex these days as it’s superior. Plus, there’s very little barrier to entry for the Codex App. You don’t need to learn 1000 skills or download clunky dashboards to get started. Keep it simple: 1. Install some key plugins like Slack and Gmail and now “Computer Use”. 2. For slides and docs, open them in the app and interact with them directly. You can point to or annotate the parts you want changed and send that as instructions to the agent. You never leave the app that’s the beauty, even for web browsing. 3. Feel free to just talk to it and just ask it to make skills for you for repeated workflows. No need to overdo it. Keep it really simple. Let the agent do the work. You are there to steer it skillfully. The way I see it. You have an F1 car. Your job is to drive it. You can go 100% autopilot (they are that good now), but you’ll have most fun when you’re steering it by your instructions as it’ll feel interactive and you’ll feel more in control. That’s it really! PS - Use a speech to text app. I use monologue. You can use whatever you like most, It helps a lot!

I used to have this view until I was convinced otherwise by a conversation with Abhijit Banerjee. His point was that we should try to alleviate poverty — RCTs are one way — until we figure out the big questions of macro development — eg industrialization. An analogy: research on palliative care until we figure out a cure for terminal diseases. It is not either or…

it's getting harder to show up for work each day knowing there are unemployed men in Colombia with 3 girlfriends doing motorcycle wheelies at 2 am


Proposal: grad students should get a budget for compute



1/ Final countdown for @lacea2025 in Recife hosted by @ufpeoficial @pimes_em. If you submitted a paper, you can check the Final program here in the Oxford abstracts virtual.oxfordabstracts.com/event/75037/pr… A pdf will be in the page soon








Meet the 2024 cohort of Wrigley Institute Graduate Fellows, representing 11 disciplines from across the natural sciences, social sciences, and humanities at @USCDornsife. Learn more about each fellow: dornsife-wrigley.usc.edu/news/meet-the-…

@prisonrodeo @jfeldman_epi @PNASNews ^2 in a formula gives an interaction not a squared term lm(body_mass_g ~ (sex + year)^2, data = palmerpenguins::penguins) lm(body_mass_g ~ sex * year, data = palmerpenguins::penguins) lm(body_mass_g ~ sex + year + sex:year, data = palmerpenguins::penguins) are all equivalent







