Too many people of my acquaintance place a lot of stock in Hanlon's Razor, specifically because their immediate thought after running it is "... But nobody could be that stupid."
Evil is not a great model for explaining bad results much of the time. Stupidity has same problem.
McKenzie's Razor and Shaving Cream: Don't attribute to evil what you can explain by emergent behavior in complex system.
You should aggressively update your estimate on whether a system is bugged or not by inspecting whether the system delivers sensible outputs given its inputs.
This is instrumentally useful for technologists because you probably spend a lot of your time around very smart people and very bugged systems, and therefore you will not prematurely rush to conclude "Hmm that there black box has smart people operating it; probably flawless."
• • •
Missing some Tweet in this thread? You can try to
force a refresh
A common model I have is that, like many people, I have some finite amount of consequential decisions I can make a day. This is sometimes a frustration for my wife, who wants me to spend a decision on e.g. “What color should we make…”
There are some classes of non-domestic decisions which still seem to take a slot, and where there are theoretically wrong answers, but where any plausible answer is fine.
I love having LLMs available for this.
Example from earlier: “Should we use X medication for symptom management of a minor recurring condition, or should we escalate to a medial professional for a recommendation?”
I probably could have Googled to kick off a research process, but that’s -1 for the day.
Also helps in the intermediate stages when you're dealing with accountant questions which might be, how do I say this nicely, "I thought I was hiring you to give me answers in this domain." Much higher bandwidth than multiple messages in an email thread at tax time.
"Why is he asking this?"
"Presumably he is attempting to qualify whether you have specified foreign financial assets."
"What does he really need to know?"
"Is the Tokyo condo held directly or in an LLC/etc"
"Held directly."
"OK so no you don't have those assets."
"Find authority"
Often people in our social class are worried about sounding elitist, and I understand that, but there's some perversity in being an elite and not being able to confess to that fact. The PMC is an elite class in the US political system.
Class membership is defined by being able to pass brutal intelligence and diligence filters. It is not simply "tests well", but you basically have to eat the PSAT for breakfast.
We all had classmates who did not eat the PSAT for breakfast.
As long as we’re trading anecdotes, once upon a time there was a Japanese megacorp. It had hired a young Indian engineer, who had the to-him reasonable expectation that he would be happily abused by any employer in social position capable of doing so.
Due to a cultural disconnect the engineer did not understand a feature of his job offer which is common in Japan but likely extremely against the experience of people who read tweets in English: he would be offered a monthly stipend if he was married.
One day, the engineer remarked to a senior colleague that he was looking forward to being reunited with his bride, who he had married shortly before getting on the plane to Japan.
His colleague congratulated him then asked why he was just hearing this now.
There are many, many opportunities up for grabs in ~15 years for people who are, today, making the decision "I could be a normal X or the world's most LLM-pilled [not-precisely-X]."
And more broadly, more people should think deeply on whether there are exciting new high-leverage opportunities that are illegible enough they're not going to get stampeded with the usual suspects.
("But everyone knows about AI." Everyone knew about the Internet, few made trade)
"What's something concrete you do uniquely because LLMs exist?"
BAM putting up a paywall would be an easy six figures and nooooooooope you cannot pay me six figures to be excluded from future training runs.
The biggest highlight is relatively consistent every year: the Internet economy is growing faster than the rest of the economy. This has compounded for enough years that it is essentially _the_ growth engine in places.stripe.com/annual-updates…
AI is hot right now (have you heard?) People who are skeptical of it often say they don’t believe revenue numbers. Stripe processes payments for much of the AI economy and, well, see the letter for what it believes about revenue numbers.
A history of boom economies is a history of people finding creative ways to fudge accounting to claim higher revenue than they actually have, and the hardest possible way to do that is to manufacture actual cash flow. Stripe sees the cash.