Benn Stancil’s Post

i am posting a link to a 2,800 blog post about microsoft excel on linkedin, good god almighty i have hit rock bottom https://lnkd.in/e7ssGkba

Some personal news (clickbait): https://lnkd.in/gbt7fF3k

So 15 years ago, the Dallas Cowboys built a new stadium. It cost about $2 billion dollars.* Fifteen years later, it's one of the biggest and most important stadiums in the country. It's hosted the Super Bowl, the Final Four, concerts, the college football playoffs, Wrestlemania.** It'll host the World Cup. I have no idea if all of this is worth $2 billion or not, but if nothing else, the stadium definitely gets used. OpenAI is apparently spending at least as much, and maybe way more, to train a new model. A new model that will be state of the art for, what 6-12 months? And will be essentially open sourced by Facebook? And will need to get replaced by another multi-billion dollar model next year, that completely obsoletes the old one? It's like every LLM is a car that loses half it's worth the moment it's launched, except instead of costing $20,000, it costs $2 billion. I suppose that could make sense - Apple arguably does the same thing with iOS updates every year - but man, it seems wild. And especially so because the companies that are doing this - that are caught in this sort of trap, of having to build and tear down a new mega-stadium every six months - are now widely seen as the most promising companies in the world. https://lnkd.in/e27Khu6c * In today's dollars. The nominal cost was $1.3 billion or something. ** Apparently, 130,000 people went to Wrestlemania over two days??

So here's my question for Paul Graham: - Elon Musk wasn't a founder of PayPal, wasn't technically a founder of Tesla, and definitely isn't a founder of Twitter.* - According to Paul Graham, Elon Musk is one of the best examples of someone who operates in founder mode. - According to Paul Graham, one of the dangers of founder mode is that managers "who aren't founders will decide they should try to act like founders."* - Huh? Like, if you think the best way to run a company is by having one person make all the decisions, sure, I guess? But why are fOuNdErS the only people who get that privilege? Why did Elon Musk have it? Why can't a "professional CEO," or other executives, or even the manager of a team, do the same thing? I suspect that Paul Graham would say that could, so long as gets results. Which seems like it should be true for founders too - do founder mode if founder mode is good; don't do founder mode if founder mode is bad - but I guess that would ruin the narrative. * Though there were lawsuits, and Elon Musk can legally call himself a founder of Tesla. And I assume he calls himself the founder of X. ** Do you smell what this founder is cooking? Know your role and shut your mouth. https://lnkd.in/eSaJu8XE

"In God we trust; all others must bring data." Like, ok, but what if they bring data that suggests different things? What if my deck says my idea is good and your deck says your idea is good? What do we do then? One of the implicit axioms of the whole "be data-driven" cult is that there is some objective way to look at numbers, and that analysis bends towards some indisputable truth. It assumes that an idea supported with a graph is better than one without, and that if two ideas both have graphs, a sufficiently scientific observer can reconcile them. Which, in some aspirational philisophical sense, maaaybe? Maybe there is a Right and a Wrong, in that one decision will lead to a good thing and one will lead to a bad thing. But practically, that's not how it works. Instead, we end up favoring whoever makes the better argument, with a pretty heavy bias that conflates "data" with "a better argument." It's not a data-driven culture; it's a debate-driven culture, where we debate with different weapons. Which I don't think gets us very far. The "insights" we're looking for aren't hidden in some meeting, and they probably aren't hidden in our data either. They're hidden in experiments: Doing stuff, and seeing what happens. https://lnkd.in/eePrEYqK

Some reviews of this blog post. What can I say, it's just banger after banger here at benn dot substack dot com. https://lnkd.in/evXi3XzR

Just like, from fIrSt PrInCiPlEs or whatever, what would you do if your job was to help people make better decisions, and you had an infinite supply of very fast readers with incredible memories, creative writers, a bunch of mediocre McKinsey consultants? I guess it's possible that your answer could be "I would have the readers carefully study our database schemas, and have the writers write SQL queries, and then the McKinsey consultants would decide if the queries should be a bar graph or a line chart." But it seems much more likely that you'd come up with some plan for the readers to read support tickets or product reviews or just watch all of your customers, and tell you things that they saw that might be interesting for you to know. There are lots of good decisions and "insights" in those sort of observations - and observing things and summarizing things with the rough business intuition of an recent MBA grad is what they're exceptionally good at. Not math, or SQL, or nuanced analytical reasoning about your very particular business. And yet, when data people - who ostensibly exist to help people make better decisions - talk about what they want to do with AI, it's the stuff with SQL and the math. And that seems...weird? Wrong? Low leverage? Sure, analytics workflows are broken; data teams are struggling; all the other marketing scare stories are true enough. But our job isn't to fix those workflows; it's to do the thing those workflows are meant to help us do. Maybe AI doesn't help with any of that, but given the bizarreness of what it is - fast readers with good memories, reasonable writers, and a passable business school student - it seems like we'd be better off trying to figure out how we'd use that skillset to make better decisions, rather than trying to shoehorn those abilities into an analytical assistant that writes some code. https://lnkd.in/etGKHgqW

The second phase of my war with LinkedIn has begun: Become a double agent and infiltrate behind enemy lines by writing the most made-for-LinkedIn, thought-leader thirst trap garbage you could possibly imagine. https://lnkd.in/edAfm9HN

The trap is that sharing a link to a blog post about how LinkedIn is a hermetically sealed chamber of parasocial sales pitches and press releases for personal brands; how everything that happens on here exists in some self-referential loop that is completely detached from the news, our real lives, and the entirety of the human experience that isn't simulated on LinkedIn; and how we all come here because, deep down, we want to be popular somewhere else, but this social media salt mine are the only one that will have us - the trap is sharing that link is actually the most LinkedIn thing you could possibly do. https://lnkd.in/ehqN_73x

I have a LinkedIn profile; therefore, my ambition is to have more LinkedIn followers. https://lnkd.in/eTvqPB6w