On Localized Information

2026.11.01

CLXXVI

[The Digital Ocean; Slopthropic's S-1; PR Tourist Trap; The Only Hacker House on the East Coast; Distribution is the Only Moat is Bullshit; Dune & Skill Monopolies; BJJ Cults; Impressive Institutions; VC Blacklist; The Repository of Knowledge]

Thesis: Localized & institutional information can sometimes be more powerful than what is available on the net.

[The Digital Ocean]

I've been born and raised in a digital world; still, I think the way it contrasts a hyper local, offline world is beyond fascinating.

It seems like all the open sourced info that is made available out there is enough, but there is a weird inverse effect: when you make info public, it sometimes becomes less valuable and less useful.

We're sifting through this endless sea of data; it's easy to fall in and drown in the nonsense when you're looking for something specific.

I don't actually think that this world kills the importance of the institution with it’s monopoly on some sorts of information and skill; I think it leaves it intact in many ways, possibly even amplifying it in some cases.

Loose post on information flow, jumping from Anthropic's S-1 to Puerto Rico to Boston hacker houses to nonsense vc claims to Dune, BJJ, and modern institutions that are valuable.

[Slopthropic's S-1]

Anthropic is going public and it's s-1 is missing all of the information you would need to make a decision as to whether or not you would actually buy it.

An S-1 is an oh so public document that should tell you everything you need to know about a company before an IPO; granted, this is a draft, so the information will come out eventually. But it is fascinating that what they don't say is more interesting than what they do say.

They have over $430B in uncancellable compute commits, but it's not at all clear when each of these is due.

This is critical information, because if the spend curve is flatish* next year, say $100B is due in 2027, the firm may very well face a solvency risk then or mid '28 depending on how much revenue they make and, of course, the cost of their other expenses.

To be clear, a solvency risk is really, really bad. If they owe someone more money than they have, they will be forced to take financing at terrible terms or go bankrupt. There is a non zero chance that the legal teams of the companies they owe money to could gut them for their IP and talent.

But, if you're in leadership at Anthropuc, who cares! You could already sell a bunch of your equity by then.

While we're here, I'll point out two interesting items in the s-1:

(1) there's an argument that gross margins are misleading due to the way they're recognizing revenue. If a channel partner sells $1 of inference but only pays Anthropic $.80, Antrhopic recognizes $1 as revenue and takes the $.20 out as an operating expense. This is potentially misleading because a lot of operating expenses shrink as revenue grows, but this one cannot! This part of margins won't naturally expand unless they renegotiate with the resellers, and this is a material part of their expenses. [Source]

(2) 5-7% of revenue comes from competitors who are distilling (read: copying) their models, and they don't know how to defend against it. The hand that giveth taketh - imagine them losing $1B in revenue from the attacks stopping, and then losing more as people swapped to the providers who copied them.

Either way, yeah, $16B in one year is a lot of money! But it's also exactly preciselyt the information they want you to see.

_{*$430B commits as of EOY 2025, sub $10B spent H1 '26, assume (no default in that period | material omission), let's say 2x in H2, leaves $400B spread over 27, 28, 29, 30, as the oft reported timeline on these is through end of 2030}

[PR Tourist Trap]

My best friends surprised me with a trip to Puerto Rico last weekend. It was a blast.

I was going to dinner with II, & she wanted scallops and I wanted a ribeye. So, I asked Claude Code to find a restaurant within walking distance, meeting those parameters.

It 'broke up old San Juan into 4 quadrants and sent out sub agents to check every menu'.

This is, on the surface, cool! And it found a place.

But it was a fucking overpriced tourist trap. The information it needed was out there to find what I thought I wanted, but it wasn't actually what I wanted.

RR found a local diner sort of thing where we had lunch and it was 10x better and less than half the price than the tourist trap. She is good at finding it by vibes, and it was also one of two restaurants that a local told us to go to when I asked him.

So, two points:

1. The wealth of optionality in our digital age allowed me to over index on two variables that didn't actually matter that much and were possibly even misleading

2. Asking a local is always safe, where if we would have tried finding the same 'local spot online', we'd have found a spot that people say is ‘local’ on the internet, which is terrifically easy to pervert

Yes, so much info is online; but given that, it's easier to ask the wrong questions.

[The Only Hacker House on the East Coast]

Myself & a contingent from C House went to a Halloween party at another hacker house we hadn't heard of in Cambridge.

There were quite a few people at the party; age gap between oldest and youngest attendees (18 to 45?), along with the lack of alcohol barring wine that was empty an hour in, and the ai generated copy of the evite gave it the air of a networking event.

I spoke to someone who was ostensibly an organizer. He proceeded to claim that his hacker house was the only one on the East Coast. I pointed out another (not ours), and he dismised it by saying the only 'surviving' and 'commercialized one', but didn't really define either.

He proceeded to tell us that this one of his had been around for 6 months and had 3 residents.

I don't like punching down, so no doxxing; his narrative is just very strange to me.

Information flow wise, it's perplexing. Here are anonymous google searches I made for east coast and boston hacker houses, putting ours, Harvard St Commons (C House), as the first result for Boston:

I wonder if the people who hang out there don't use the internet? Or maybe if it's some sort of 'control the narrative' thing, where you just say something and add nebulous qualifiers and then can't be argued against? Depending on his definition of 'commercialized' maybe it's true? That said, by any rational definition, at least half of the companies in our house are 'commercial'.

Maybe it's because a lot of people don't know what a hacker house is? And won't bother to look it up?

[Distribution is the Only Moat is Bullshit]

This guy at the other house actually went on to claim that one of his portco's has a moat because they hired a John Deere enterprise sales rep.

This is such a strange and confusing claim that I can hardly begin to unpack succinctly, but I'll try:

• A moat is supposed to be a thing that makes it harder for competitors to copy you

• Hiring an in industry enterprise sales rep is simply an expensive thing that can get you sales - they tend to bring contacts with them and can, resultantly, get deals much more quickly

• Anyone with money can hire one of these people?

Does this mean the claim normalizes to money is a moat? If you spend it on acquiring customers maybe? If so then why doesn't everyone who spends money to acquire customers keep them forever and have a moat?

I guess if this claim has the effect of making founders think about getting revenue more than raising capital, maybe it's useful? But does it have that effect if the way he says you need to get customers is to drop half a milli and a ton of time hiring a dude who will sell your product for you? Doesn’t that mean you have to raise to have a moat?

Maybe it's something about ostensibly having credibility, speaking with confidence, and using words that signal you're in the 'in group' that passes as information nowadays? The claim is probably congruent with shit drifting around on Twitter maybe? I don't really know.

But I do know that this isn’t valuable information. Really, it’s a misleading hazard.

[Dune & Skill Monopolies]

A mark of our digitized, global society seems to be a decreasing awe for institutions, and a correlated lack of value put behind mastery. Definitely a double edged sword.

I'm reading the second half of the Dune books (finished God Emperor, halfway through Heretics) and one hallmark of them are the intense cults of specialists.

The economics of the universe revolve around, of course, the wonder substance Spice Melange, but also humans who are really, really good at something. (altering details to be congruent with book 1, if you catch my drift.)

As an example, one faction, Ix, has a monopoly on digital technology; a fight scene included a drone, and this led to the immediate and correct conclusion that Ix was involved - no one else could even dream to make such devices!

The Tliexu can regrow dead people from their cells (Gholas) and have shapeshifting face dancers, the Guild controls space travel, the Bene Gesserit have something of a monopoly on religion and are always in demand for negotiation purposes, the Suk Doctors are basically the only ones you want practicing medicine, the Mentats are master strategists and human computers; I'm sure the list goes on and on.

The shape of the world is that groups can control significant economic power by controlling the flow of deeply specialized people. If you want a competent doctor, you MUST do business with the Suk.

These sorts of monopolies still exist in real life, but people are calling them into question.

I wonder how much of that questioning is founded in reality and how much of that is missing the point that these monopolies are built on cults of intense practice. Repeating tweets and making untrue claims is not a substitute for being the beneficiary of internalizing institutional knowledge through intense years of practice and cultivation.

[BJJ Cults]

This makes me think of what I'm told jiu jitsu was like 25 years ago. It was offensive and dishonorable to leave your gym and share the techniques you learned their elsewhere.

However, as first training videos starting coming up for sale, and then people started posting techniques online, this massively changed. Advantages can't really be garnered from this sort of secrecy and hoarding information as much anymore. It's open source.

That said, you still have to intensely train and practice to get good at it; that part hasn't changed at all.

And beyond that, there is SO much information out there that half of it is sifting through it to the important stuff.

This wednesday, our professor showed us a sweep you can do when someone starts standing out of your guard. Then, he proceeded to show us how the standing opponent can legally break your neck when you do it, and told us to never use the sweep because of how dangerous it was.

Anyone can learn this sweep online. And people who are good at it are incentivized to show it to you online without explaining the risks, because they wouldn't feel liable for it.

But a good, savant coach who understands the risk might warn you about it.

The information online just feels more cheap and less useful than the hidden secrets. Or maybe it's the value of knowing which information to reveal when?

[Impressive Institutions]

We still do have a cultural fascination with institutions that inject know how into the people that go through them. Here are some that I've seen in practice that seem to be relatively strong heuristics:

• Navy Seals, SAS, and perhaps a few more are terrifyingly efficient & assymetrically powerful military forces

• Palantir employees or quantitative trader are savant engineers and math minds

• Bain & BCG consultants being quite good at strategy. (This is based off each of the people I know who work their being competent, not ever having worked with the firm!)

But then there are a collection of these that I think are bullshit heuristics, too. It's hard to impress me with the elite school you went to or dropped out of. It's also hard to impress me by having been a banker.

Then some are hit and miss: expensive lawyers impress me based on their skill level & clarity of communication. VCs, now, are actually on average not competent in my mind. MBA (getting it in the last 10 years, different if earlier) doesn't add anything, and oftentimes subtracts.

[VC Blacklist]

C House has an unofficial VC blacklist. I'd never publish it online; having it out there somehow makes it less credible.

Rather, it's something we share with all of our residents. Which should make it more terrifying, as the 90 or so residents have raised over $250M in the last 3 years.

The accumulation of Institutional knowledge like this is fascinating to me.

How can you accumulate useful data internally and distribute it to your community in an efficient way?

Yeah, it's a bit exclusionary to not spread the info everywhere, but I think it also marks the information as a loaded gun. You wouldn't leave the loaded gun on a table in your living room, you'd only give it to people who you believe would use it right.

And that doesn't mean that you'd give it to EVERYONE who might use it right; it's not a negative statement about those you don't give it to, it's just a positive statement about those you do give it to.

If you enjoyed this post, please subscribe; I write about philosophy / startups / hacker houses / bootstrapping / finances / AI / adventures / etc each sunday.

[The Repository of Knowledge]

A lot of information is all out there now.

It's harder to know what is useful than it is to find the info.

Maybe the function of the institution becomes more of knowing when to share critical information?

Does sifting through noise become the new institutional information monopoly? Can you ever truly open source that?

Even some of those institutions I mentioned that impress me above will inherently have a surplus of bullshitters come through them.

Is it even possible to have an org that guarantees everyone who gets through is super humanly competent on one axis?

Just some questions for the future. I have ideas, but no concrete answers yet.

Live Deeply,