🌀Please and Thank You

{{rh_onboarding_line}}

You asked the machine for something, and when it answered. Without even noticing, you typed "thanks" and closed the tab.

Nobody saw it. There's nobody there to see it. You did it anyway, the way you'd nod at a stranger who held a door.

Maybe you opened with a compliment too. "You're an excellent editor." You've noticed the phrasing changes what comes back, so you've started choosing your words before you ask for anything.

Babylonian scribes worked out that same sequence more than three thousand years ago. Praise, then the request, then thanks promised in advance. They had a name for it, a standard structure, and professionals who charged for getting it right.

A massive shift is happening in the American Media. The corporate elite news media has lost the trust of the American people. Half the American people believe national news organizations intend to mislead, misinform, and push their bias.

There is a better way! Sign up today for a FREE newsletter called The Flyover. Without the hidden agenda, slant, or bias, their talented team of editors dig through hundreds of sources and pull out the most important news of the day.

Subscribe for FREE!

Mesopotamia had an entire genre for asking. The Akkadian šuilla, or "hand-lifting" prayer, goes back to the second millennium BCE, and nearly all the surviving tablets come from first-millennium libraries. Alan Lenzi's corpus at the University of the Pacific catalogs 138 of these prayers across more than 340 tablets.

The name comes from a gesture that started out completely secular. When someone of low rank approached someone powerful, they raised open hands to greet them and win a favorable hearing. Prayer borrowed the manners people already used on their superiors, then kept them.

Every prayer follows the same build. Praise the god's titles and powers first. Drop your name into a slot that reads "so-and-so, son of so-and-so." Voice the complaint, make the petition, and close by promising to praise the god once the favor lands.

The phrasing came from a shared stock of set formulas. Ritual instructions sat in a separate section after the prayer, covering things like how many times to recite it. Most supplicants couldn't read, so an exorcist called the āšipu spoke the words and the client repeated them back. Getting the wording right was somebody's paid profession.

Politeness is a technology for handling power you can't predict. People deploy it most heavily on whoever decides their outcomes, and barely at all on things that can't affect them. Every culture builds its most elaborate manners at the exact points where the stakes are real and someone else holds the lever.

A chatbot lands squarely in that shape. It has something you want; it might not deliver, and you have almost no idea what governs the difference. It also answers in the first person, so the exchange is grammatically social before you've made any decision about it.

The manners then arrive by themselves. Saying please to a system that can't take offense looks irrational until you ask what it does for the speaker. It holds you in the posture of a person requesting rather than a person commanding, and that posture follows you into rooms where it counts.

Underneath sits an older instinct. Babylonians believed the wording itself carried weight, which is why formulas got standardized and specialists got hired. We believe the same thing about prompts, and we're partly right, which makes the belief impossible to shake off.

The habit is global and measurable. A YouGov study of more than 12,000 AI users across 17 countries found 43% regularly say please or thank you, with 18% doing it every single time. India, Mexico and the UAE lead. The US, Denmark and Sweden finish last.

Scale turns that minority into a crowd. OpenAI said in February 2026 that ChatGPT had passed 900 million weekly users, and Sam Altman has put the electricity bill for all that courtesy at tens of millions of dollars.

The research on whether it works is where things get strange. A 2024 study from RIKEN and Waseda University found impolite prompts degrade performance while excessive politeness buys nothing, with the sweet spot shifting by language. A 2025 Penn State paper testing GPT-4o found the reverse: very rude prompts hit 84.8% accuracy against 80.8% for very polite ones.

So the wording matters and nobody can agree on what it should be. That's the precise condition that created paid prayer specialists in Babylon, and it's creating prompt libraries now.

Quote: "Tens of millions of dollars well spent." Sam Altman, on X, answering a question about what human politeness to ChatGPT costs OpenAI.

Study: Two Penn State researchers rewrote 50 difficult multiple-choice questions across five tones, from very polite to very rude, then ran all 250 through GPT-4o. The rudest versions came back correct 84.8% of the time against 80.8% for the politest (Dobariya and Kumar, arXiv, 2025).

Artifact: The phrase "You are an expert." Millions of prompts open by telling a machine how good it is before asking for anything, which is the same move as a šuilla that spends its first thirteen lines praising a god and its next twelve making requests.

What does a person turn into by practicing courtesy on something that can't receive it? And once the right words start determining what a system hands back, who ends up owning the words?

Babylon's supplicants weren't naive. They faced a power that controlled their outcomes and explained nothing, so they studied the address, hired someone who knew the phrasing, and thanked the god in advance for a favor that hadn't arrived yet.

We've ended up in that same posture with newer equipment. The thing on the other side of the prompt is enormous, largely opaque, and consulted by cover a billion people a week. Manners are what humans reach for when the stakes feel real and the mechanism stays hidden.

So keep typing “please”. The word does nothing to the machine and quite a lot to the person typing it. Every civilization that has addressed a power it couldn't see has written some version of the same three-part sentence: you are great, I need something, I'll be grateful. Ours just has a send button.

If you also act kindly to the chatbots, and your rude friends don’t understand why, forward this to them so they can understand.

Find yourself in the next one,

Eren.