iiinova — where your digital self keeps evolving

Greg Faletto @gregoryfaletto.com · 01/10/2026

Hello! I made a skill for R package development. It's a multiagent system running an end-to-end workflow for converting a GitHub issue into a PR solving the issue, with subagents for planning, implementation, and review. github.com/gregfaletto/... 🧵

github.com

Greg Faletto @gregoryfaletto.com · 21/07/2026

I think the audience is just that bad at recognizing it

031

In fairness, people using creative definitions of "reading" a book to make it seem like they read more than they actually do is not exactly a new thing this decade

030

Cool, fair enough. My main objection is to "it's unfair to ask CA homeowners to pay their fair share in property taxes [in the form of repealing Prop 13]"

030

I'm in the same boat as you and I'm certainly not going to tell you you're wrong. And when it comes to NIMBYs, certainly I hope their political project fails.

Whoa Prop 13 was repealed? That's awesome news! I admit I am surprised people aren't selling their homes, there were lots of complaints that if Prop 13 were repealed people would be forced to sell their homes because they couldn't afford the property taxes, but I guess it didn't play out that way.

230

I'm not saying CA homeowners are fools for not selling their houses today, I'm saying that it's unreasonable to claim they are middle class smol beans whose home equity doesn't count as wealth and they can't afford to pay their fair share of property taxes if Prop 13 were repealed

240

I refuse your premise of a very specific guy who fears investing, doesn't know about FDIC limits, and goes on to be screwed over by the first bank run in over a century where FDIC insurance is insufficient to bail out all of the victims

180

Maybe I typed it wrong the first time, now I'm getting $3000 per month lol but same idea

000

Yeah and there's no financial pressure on them to move and we're both happy right?

What are you talking about? Of course you can keep 900 grand in a bank. Spread it across four if you're really worried about FDIC limits. But yes it would probably make more sense to invest a lot of it

170

I agree that it's not a great scenario that housing costs rise so rapidly in CA and it creates no-win situations. I'm sympathetic to people who put down roots and don't want to be priced out of the home under their feet. My solution would be to build way more housing to match demand.

210

At a safe withdrawal rate of 4% you could use that money to pay $3300 per month in rent in perpetuity, that's pretty good living in LA. ($3300 in today's dollars, that safe withdrawal rate accounts for inflation so you could increase that amount every year with inflation.)

3240

For many people, if you offered them the deal of "you can have $900,000 but you're going to have to move out of your house" they would jump at the life-changing opportunity and go on to tell their grandchildren about the day they signed the deal

7260

Issues · gregfaletto/fetwfePackage

Please let me know if you have any questions or feedback, and feel free to create an issue at github.com/gregfaletto/... if you notice any bugs or want a new feature.

github.com

Implementation of fused extended two-way fixed effects (Faletto 2024) - Issues · gregfaletto/fetwfePackage

000

Finally, the package also supports {broom} functions like `tidy`, `glance`, and `augment` for easier interactions with other tidyverse packages.

Documentation snippet for the fetwfe package titled 'Tidy output with broom', from the blog post at https://gregoryfaletto.com/2026/05/23/now-on-cran-fetwfe-version-1-10-0/. It demonstrates tidyverse integration using tidy(), glance(), and augment() functions, featuring a console summary table for ATT and cohorts 2 through 4.

100

Cluster-robust standard errors are also now implemented. It is also worth noting that this update fixes errors in the previous package versions in the standard error calculations. This affects both point estimates and p-values.

Documentation snippet for the fetwfe package titled 'Experimental cluster-robust standard errors', from the blog post at https://gregoryfaletto.com/2026/05/23/now-on-cran-fetwfe-version-1-10-0/. It demonstrates using the se_type = 'cluster' option to calculate a unit-clustered sandwich estimator, complete with example R code and console output.

100

There are also expanded features for statistical inference. P values and confidence intervals are available for all estimated treatment effects, and there's a "selected" binary column indicating whether the cohort ATTs were estimated as exactly 0--which serves as an asymptotic 100% CI.

Documentation snippet for the fetwfe package titled 'Testing whether an effect is zero', from the blog post at https://gregoryfaletto.com/2026/05/23/now-on-cran-fetwfe-version-1-10-0/. It explains the 'selected' flag property using a console output table for 'result$catt_df' across Cohorts 2, 3, and 4.

100

What's new in version 1.10.0 compared to the previous CRAN release: first of all, the package now provides convenience functions for event study estimates and plots.

Blog post excerpt (from https://gregoryfaletto.com/2026/05/23/now-on-cran-fetwfe-version-1-10-0/) for the fetwfe package titled 'Event-study estimates and plots'. It describes the eventStudy() and plot() functions, including an example output table showing event-time estimates, standard errors, confidence intervals, and p-values.

Event-study estimates plot showing pooled ATT over event time (t−r) from 0 to 4. Black dots with vertical error bars indicate estimates and confidence intervals; a dashed horizontal line marks 0. All estimates are below zero: approximately −0.75 at time 0, −1.03 at time 1, −1.28 at time 2, −0.34 at time 3, and −0.84 at time 4. The most negative estimate occurs at time 2, while time 3 is closest to zero. Error bars are widest at times 1–2 and narrowest at time 3.

100

FETWFE is a middle ground. It regularizes the coefficients in the ETWFE regression towards each other, so that if e.g. within a cohort the treatment effect is the same in time periods t and t + 1, we only estimate one TE for those two time periods, preserving unbiasedness with better efficiency.

$Screenshot from Faletto, Gregory. "Fused extended two-way fixed effects for difference-in-differences with staggered adoptions." arXiv preprint arXiv:2312.05985 (2023). Figure 1: A line graph visualizing connections between estimated marginal average treatment effect terms (tau_rt) across time t (horizontal axis, 1 to 6) and cohorts r (vertical axis, 2 to 5). Details of the grid and connections are described in the accompanying text. Figure 1: Visualization of which of the estimated marginal average treatment effect terms \hat{\tau}_{rt} from regression (1.5) (which estimate the average treatment effects \tau_{ATT}(r,t) from Equation 1.1) we penalize towards each other in the FETWFE penalty (5.1). In this setting, T = 6 and R = {2, ..., 5}. The horizontal axis depicts time and the vertical axis depicts cohorts. FETWFE works well under an assumption that the linked treatment effects tend to be close together, and at least some of them are exactly equal. See further details in Section 5.$

100

This works, and is unbiased. But maybe all of these separate TEs are overkill, and some of them might actually be equal (like the ATT for the same cohort in adjacent time periods). If so, ETWFE is inefficient (if you speak econometrics) or has a poor bias/variance tradeoff (if you speak ML).

100

Two-way fixed effects, the two-way mundlak regression, and difference-in-differences estimators - Empirical Economics

One solution for this, extended two-way fixed effects (Wooldridge 2025), solves this problem with an extended regression that estimates a separate regression coefficient for the treatment effect for each cohort at each time since starting treatment. link.springer.com/article/10.1...

link.springer.com

I derive a result on the equivalence between the two-way fixed effects (TWFE) estimator and an estimator obtained from a pooled ordinary least squares regression that includes unit-specific time avera...

100

Briefly, under staggered adoptions, the two-way fixed effects estimator is biased for the average treatment effect on the treated units. It turns out that estimating a single coefficient for the treatment effect on all cohorts at all times doesn't aggregate into a sensible estimate of the ATT.

Screenshot of an excerpt from Chapter 18 of: Hansen, Bruce. Econometrics. Princeton University Press, 2022.

"See Exercise 18.1.) Thus the above regression is identical to the two-way fixed effects regression Y_it = theta D_it + u_i + v_t + epsilon_it (18.3) where u_i is a restaurant fixed effect and v_t is a time fixed effect. The simplest method to implement this"

100

fetwfe: Fused Extended Two-Way Fixed Effects

Version 1.10 of {fetwfe}, my package for difference-in-differences under staggered adoptions, is now available on CRAN: cran.r-project.org/web/packages... Blog post: gregoryfaletto.com/2026/05/23/n... Brief explainer on the package, and description of the updates, in the thread below.

cran.r-project.org

Calculates the fused extended two-way fixed effects (FETWFE) estimator for unbiased and efficient estimation of difference-in-differences in panel data with staggered treatment adoption. This estimato...

120

Greg Faletto @gregoryfaletto.com · 23/05/2026

who would have thought the "performative allyship" of the late 2010s was not, in fact, the worst it could get

0263

I'm the guy who would have guessed this and I'm not surprised at all

Greg Faletto @gregoryfaletto.com · 23/05/2026

It's very funny to run the "this is no big deal actually" playbook about solving a famous open problem that the smartest people in the world had very much been trying to resolve for many years

061

Greg Faletto @gregoryfaletto.com · 19/05/2026

They tacitly condoned Trump firing Comey, obstructing Mueller, the Old Post Office, etc. House Dems played a huge role in ratifying that Trump is a special birthday boy who is allowed to do crimes.

200

Greg Faletto @gregoryfaletto.com · 19/05/2026

Honestly I think the behavior of house dems from January - September 2019 is hugely underrated here. Dems campaigned and won on holding Trump accountable. They broke their promise and did nothing until Ukraine almost halfway through their term.

210

Greg Faletto @gregoryfaletto.com · 18/05/2026

yes. experienced programmers used to talk about how nobody understood machine code or assembly anymore and that made them worse programmers. they weren't wrong that what a swe was expected to know had changed, and that it wasn't all positive, but that doesn't mean it was a bad thing on net

030

from the bottom of their hearts they do not think a democratic trifecta can have any more bite than the last one. i suppose they think they (and/or the supreme court) could rein in newsom with a trifecta. presumably if you asked their opinion of spanberger or pritzker they would not be familiar.

1170

Whoever wants to come in the tent has an open invitation in November. If they don't want Democrats working on their behalf, we should stop second-guessing them and let them have more of what they want. That means farmers, labor unions, rural areas, small business owners, etc.

070

Yes 100%. Democrats are *way* overdue to go back to basics of representing the interests of their districts and voters instead of endlessly throwing them under the bus in favor of long-gone constituencies.

1131

I think “citing” a paper that *literally doesn’t exist* is a pretty serious ethical violation that should be punished pretty harshly, regardless of LLMs. I think a minor typo or single incorrect word in a reference is no big deal (also regardless of LLMs).

080

Thanks, I missed the first name being off. That’s not how I read the announcement and I’d certainly take that to be unreasonable, I hope that’s not how it’s implemented.

Greg Faletto @gregoryfaletto.com · 16/05/2026

The cool thing is, if it were a 6-3 liberal Supreme Court making decisions they sharply disagreed with that appeared unconstitutional, they’d for sure be totally cool with it

1170

That’s not a hallucinated citation. There’s a real paper of that name by those authors. I’m not sure if you’re aware, but LLMs have been generating “citations” of papers that literally do not exist. If you typed the purported title and authors into google, nothing would come up, unlike this paper.

140

Greg Faletto @gregoryfaletto.com · 16/05/2026

This policy as stated wouldn’t ban you over a dead link, but you should check (all of) your references enough to know that they exist. Or do you have a different failure mode in mind?

120

Greg Faletto @gregoryfaletto.com · 16/05/2026

Him not having fully wrapped his head around how much money he has would honestly explain a lot

(Hell think about someone in 2021 trying to predict the next 5 years.)

020

Think about talking to someone in 2014 who was trying to predict the important developments of the next ten years. You’re allowed to give them feedback but not give any substantive new information. You’d just tell them not to waste their time bothering, right?

120

Very common analytical error to think “obviously the thing we’re preoccupied with in the last X time periods is going to be the main thing for the next 100X.”

140

Wow, very interesting, thanks! For sure an interesting antecedent.