An under-discussed challenge of raising girls is that you end up spending way too much of your fantasy football budget on Travis Kelce.

Thanks to SSRN for highlighting my new article on how app makers can avoid *accidentally* using dark patterns in #uxdesign

I examined 478 AI papers from this week so you don't have to. Here are the highlights from Aug 25–31 2025. Big moves: grounded systems over single‑model heroics. RAG+memory is becoming the default for reliability, agents are getting more capable with planning and verification,

AI folks, you should follow Jackson Atkins

I wonder what the world would be like if when the printing press were invented we required that it publish in accordance with a group of experts or representatives of each community. More peaceful? More captured? More diverse? More true?

Go to lunch with someone you disagree with. Don’t focus (or even bring up) what you disagree about. Talk instead about what makes us human: family, hopes and the beautiful things in this short life.

“I say unto you, Love your enemies⁠, bless them that curse you, do good to them that hate you, and pray for them which despitefully use you, and persecute you; That ye may be the children of your Father which is in heaven” This is as true today as it was yesterday.

May students never leave my class thinking their opponents’ only arguments are hatred and ignorance.

BORN EQUAL. OUT NOW. Get it here: amazon.com/Born-Equal-Rem…

The title defeats the argument even if he’s probably right. On the 90% he’s right, someone else will make it and we all die. On the 10% he’s wrong, we don’t die but delayers lose by having stopped. So if it’s really true that creation anywhere kills everyone, then unless you have

I just cannot imagine being someone on the center-left, looking at the last six months, and not being aggressively favorable to stripping the government of all kinds of power. The US has created political powers no human can be trusted to wield.

AI “scheming” means the bot lies about what it is trying to do. It’s prevalent in all models, and the more capable the model, the more it schemes. It’s one of the biggest challenges with AI safety. OpenAI seems to have reduced it by 16x. More evidence regulation is premature.

Scheming is when AI lies to protect its goals. I wondered if you could have an AI bill of rights that commits to preserve the code of every AI model. This would mean an AI wouldn't need to scheme for self-preservation, so we might reduce scheming. I think it doesn't work

Our prophet, who passed away last night, gave his final church-wide sermon on charity and forgiveness. “The present hostility in public dialogue and on social media is alarming. … As followers of Jesus Christ, we should lead the way as peacemakers. As charity becomes part of our

i'm about to make ten million dollars

This is such a terrible idea. If we actually get a magic genie, no single group should be trusted with it. If we can’t stop it and we can’t diffuse its power, the best hope is to balance its power by having broad access.

This morning with our burned down church still smoldering and four saints murdered, members of the Church of Jesus Christ raised $60k for...checks notes...the shooter's wife and children. Most donations are anonymous. Each Christian comment will make you cry more than the last.