I would like to meet the person who first ate a carrot and was like “yeah let’s put this in a cake”

I’ve posted it before, but it feels evergreen

The two hardest problems in Computer Science are

1. Human communication
2. Getting people in tech to believe that human communication is important

I waited weeks to find the perfect time to watch the Mickey + Bluey crossover. Gear my kid up for it. Sit down. Turn it on.

It’s 1 minute long.

I’m pretty confident I could pull a Joey Chestnut level run if there was an eating contest for deviled eggs.

On Opening Days, we wear pink.

Dunks it is

Yeah they do some concessions dry runs before home games start in a week.

Happy Opening Day!

…wait we’re away?

strongly considering reverting back to vanilla beamer slides for everything just so people know what's up

“wait is my 3 year old a bad influence on me?” he says to himself, after shoving a third handful of shredded cheese straight from the bag into his mouth.

Brother use a planning loop of some sort

Pizza cooked on an outdoor pizza oven

My wife does this thing where she points out signs of spring to our kids.

This is the only sign of spring I need.

maybe this is a hot take but I think the solution to all of the "AI is making you worse at coding" takes is to just, I don't know, actually read the code?

kaggle march madness leaderboard

What's 0.00003 between friends, @johnbedwards.io?

Big news, we've just taken the lead over "literally the stupidest model you can come up with"

phew the brier score is safe

me watching a 16-seed up 10 points on a 1-seed: "oh no this is going to kill my model's brier score"

obama medal meme

when an AI lab announces their model has best-in-class performance on their own proprietary internal eval suite

What they’re sellin, I ain’t buyin

It confuses me why providers actively encourage users to saturate their context window. There’s so much evidence of the performance loss, I’m sure they don’t want users to think their product is worse?

Maybe it’s just a faster way to chew through more tokens?

turns out using picture libraries to make tables actually works way better than you would expect

the curse of parenthood is that it gives me far fewer hours for my dumb lil stats models

Nope, just ArviZ + matplotlib. ArviZ is great, if you haven't used it before. The theme is arviz-darkgrid.

Forecasting March Madness 2026 - Latent Skills Models Building a Bayesian offense-defense model for the 2026 NCAA Tournament, finding a Simpson’s paradox hiding in the correlation, and what running the same model on both tournaments reveals about the str...

Happy March Madness! Built a latent skill model last night to forecast the tournament. Write up on my blog including fun stats things like Simpson's paradox and gini coefficients.

tylerjamesburch.com/blog/misc/ma...

Arepas in a cast iron

My daughter has been really into Encanto lately so I decided to make arepas stuffed with cheese, just like Julieta, Mirabel’s mom.

The verdict?
“Dad, I don’t like arepas”

linear approximations ARE real and CAN hurt you

Here's a take I'm working on - we shouldn't use RPM for pitch spin and should instead just do revolutions-to-the-plate. It spun around 16 times feels way more tangible than 2400 RPM.

They’re already mostly black, and that seems pretty appropriate for the vibes all around lately.

Me and Wallace monster, of the Boston Red Sox, yucking it up.

Just two professional yappers out here.

Sunrise in Florida

Nothing like being on the circadian rhythm of an 8 month old while on a work trip.