Tragic L for the stats community that this beats out "Generalized Linear Model" on google

I guess this is how I learn a factoid means a fact, and also not a fact Thanks I hate it 

Grant Brisbee@grantbrisbee.bsky.social · 6d ago

This is a good question! A factlet is a little fact (and not a real word ... yet). A factoid is false or exaggerated fact. Join the factlet revolution.

at least my one-year-old is cutting 4 teeth during deadline week

hello there P = NP is solved thanx to my close friend akhil for asking about it and my other close friend fable for working during the world cup final P=NP N=P/P N=1

these scammers just aren’t as philanthropic as me

Tyler Burch@tylerjamesburch.com · 2w ago

Didn’t really expect this to work but sure

Didn’t really expect this to work but sure

The @RedSox REFUSE to lose! TWELVE straight wins for Boston!

Please avoid all salads, including word.

I've got a complex relationship with ai

in the last 3 months the only walkable coffee shop and ice cream shop from my home have both closed and I don't know what I did to deserve this

Apologies to anyone who has to occupy a 4 foot radius around me for the next month, but these cucumbers aren’t going to pickle themselves and smelling like garlic and vinegar is just the cost of doing business, baby.

Went to the Noah Kahancert. Got sad. 10/10.

Tyler Burch@tylerjamesburch.com · 4w ago

pumped to watch the sad new england boy sing sad new england songs in new england tonight

pumped to watch the sad new england boy sing sad new england songs in new england tonight

Well doc, I guess my abandonment issues probably started when I opened the Taco Bell app and nothing was under two dollars.

Alright claude, this has gone too far

happy fourth i guess

And by soon I mean like 4 hours later lol tylerjamesburch.com/blog/tooling...

Letting Agents Drive Your Interactive Sessions with tmux

A workflow I’ve been using lately - pointing agents at a live REPL through tmux, with a file-redirect trick that makes it reliable.

tylerjamesburch.com

Tyler Burch@tylerjamesburch.com · last mo.

So I figured out that you can use tmux send-keys to have an agent backdoor into an interactive session. This is super nice if you want to avoid reloading data on every turn and also reduces the amount of scrap scripts they pile up. Will write a short blog post soon.

And by soon I mean like 4 hours later lol tylerjamesburch.com/blog/tooling...

Letting Agents Drive Your Interactive Sessions with tmux

A workflow I’ve been using lately - pointing agents at a live REPL through tmux, with a file-redirect trick that makes it reliable.

tylerjamesburch.com

Tyler Burch@tylerjamesburch.com · last mo.

So I figured out that you can use tmux send-keys to have an agent backdoor into an interactive session. This is super nice if you want to avoid reloading data on every turn and also reduces the amount of scrap scripts they pile up. Will write a short blog post soon.

So I figured out that you can use tmux send-keys to have an agent backdoor into an interactive session. This is super nice if you want to avoid reloading data on every turn and also reduces the amount of scrap scripts they pile up. Will write a short blog post soon.

asking an LLM, that thinks in tokens, how many letters are in "strawberry" is like asking a person on the street, who thinks mainly in words, how many ASCII values in "supercalifragilisticexpialidocious" are prime numbers i won't answer that question correctly without tools (pen and paper?) either

The thing I'm learning about agent-based development is that you need to do something like: - Let it implement the thing - Let a new agent review the thing - Let another new agent review it again - Repeat until it's happy with it's work - Finally, look at it yourself

Frankly I’m not really sure what this has to do with language acquisition, Duolingo.

Grant me the serenity to accept the problems where I cannot offload my understanding, courage to let the LLM go burrrr on the problems where I can, and the wisdom to know the difference

The more I do it, the more it feels like there's some fundamental misalignment between LLMs and statistical work. Maybe it's over-abstraction leading to making it harder to turn knobs and experiment? Maybe the harnesses are oriented toward shipping code and not sitting on a problem?

The better the models get, the more I seem to have to scream "DON'T FORGET YAGNI" and "come on, KISS!" at them.

No company has ever been more relatable than Dunkin giving up on the “healthy” protein menu strategy and hard pivoting directly into putting Oreos on everything.

Quick question, does gpt 5.5 still talk to me like it’s trying to prove to me that it’s read every single book on computer programming? “We need to leverage a shim to reduce the risk surface area” got it

In case you were wondering how this performed, near-perfect temps meant no weather effect, leaving just the random-walk estimate. Top-3 average was 2:02:30, ~0.7 posterior SDs faster than the 2:05:10 median, inside the 95% interval (1:57:55–2:12:29).

Tyler Burch@tylerjamesburch.com · 4mo ago

ased on the attire on the Green Line skewing more Hokas and On Cloudmonsters than a typical workday, I can confirm today is the Boston Marathon. Just uploaded a new blog post looking at how much of finish-time variation is attributable to weather. tylerjamesburch.com/blog/statist...

gotta cash these in somewhere

ased on the attire on the Green Line skewing more Hokas and On Cloudmonsters than a typical workday, I can confirm today is the Boston Marathon. Just uploaded a new blog post looking at how much of finish-time variation is attributable to weather. tylerjamesburch.com/blog/statist...

Weather Effects on Boston Marathon Times

How much did the 1976 ‘Run for the Hoses’ actually slow the field?

tylerjamesburch.com