chris muscarella

“Courage is not simply one of the virtues, but the form of every virtue at the testing point." -CS Lewis

Double Sharp, founder/ceo; Field Co, ironman / founder

[chris muscarella @cm]

A milestone on a long journey down—for posterity here. Temperamentally I don't really celebrate wins, the upside to that is non-fatal losses are also taken in stride. Onward.

[chris muscarella @cm]

But let us not forget: parlay time is dangerous

https://t.co/YVuztq6F1d

[chris muscarella @cm]

Proposition people. The only way you find your humans that are game for the most absurd adventures is to parlay.

[Raphaël Doan @raphaeldoan]

Le mathématicien Daniel Litt sur le problème d'Erdős récemment résolu par GPT-5.5 :

"Une autre explication est que la solution exigeait des idées venues de domaines que la plupart des chercheurs travaillant sur ce problème ne connaissaient pas. Si elles sont justes, ces explications devraient nous mettre mal à l’aise. Elles suggèrent que les incitations à la spécialisation et au cloisonnement, si compréhensibles soient-elles, nous ont privés de travaux scientifiques de grande qualité."

Le retour des polymathes à la Leibniz à travers les LLM ? Ce serait une bonne nouvelle.

[chris muscarella @cm]

@tjparker @AeroPress I’m a fan of the AeroPress product. The drops / not being able to ship on time / months delayed / bunch of infrastructure choices… are just dumb for a company that isn’t fashion and is itself “infrastructure”. Probably 30% drop in my NPS since they were acquired…

[chris muscarella @cm]

@pitdesi Sheel, that’s the real world. Brian’s on record saying to dollarize countries / future vision is everyone has a non-custodial wallet and it’s extremely high utility everywhere.

[chris muscarella @cm]

@richardchen39 KYC / AML issues. And tax issues broadly (not triggering / evasion / etc)

[Brent Beshore @BrentBeshore]

"The interviewer, @DouthatNYT, broke down crying during the conversation. And when he did, Sasse laughed. Not unkindly, but the way a man laughs when the heaviness and the lightness of something are both true at the same time and he’s decided not to pretend otherwise."

[chris muscarella @cm]

@karsenthil I would be absolutely delighted!

[Philip Rosedale @philiprosedale]

This is so beautiful. And so important to study. This phenomena (emergent coherence) is central to much of life, and most people don't even imagine it could happen.

[chris muscarella @cm]

@Hadley Same with things like chat. As a small team we're dumping WhatsApp logs into CRM / ML loops and kind of waiting for agent native things like Ando before just jumping into Slack

[chris muscarella @cm]

@Hadley One of the funny things about evaluating tools for teams these days... sometimes helpful to vibe code something over a couple of weeks just to get a better sense of what you actually want before choosing a 3rd party system... we're doing that with our CRM

[chris muscarella @cm]

@Hadley @mszepien @Benioff Also interesting that CRM as system of record (when good) is less about screen interfaces than most other SaaS applications I can think of—and you see people like @markitecht pushing that hard at Day AI

[chris muscarella @cm]

@Hadley @mszepien @Benioff The slightly less salty version of this is something like: would be willing to bet most usage / most revenue at Salesforce is tied to highly customized versions with their own UI / almost no one uses the default

[chris muscarella @cm]

@thegothamgal As an example: are a bunch of debt covenants tied to rent levels or something similar that if triggered would cause a cascade across a real estate portfolio?

Or the simple frame of: what are the top rational reasons for owners of NYC real estate to behave this way?

[chris muscarella @cm]

@thegothamgal Joanne, actually curious to go a click or two deeper in a “follow the money” way… is there a structural reason why buildings stay empty / other than a cartel-like action to keep rents at a certain level?

[chris muscarella @cm]

@justinmassa What surprised you most? And did you spend time trying to make a recursive loop on new things to add etc?

Haven’t taken a crack yet at making my own index / crawler for links out of things like bookmarks but excited about it

[chris muscarella @cm]

Dream come true: have built libraries of papers and semi well catalogued digital exhaust for 20+ years waiting for this

[Andrej Karpathy @karpathy]

LLM Knowledge Bases

Something I'm finding very useful recently: using LLMs to build personal knowledge bases for various topics of research interest. In this way, a large fraction of my recent token throughput is going less into manipulating code, and more into manipulating knowledge (stored as markdown and images). The latest LLMs are quite good at it. So:

Data ingest:
I index source documents (articles, papers, repos, datasets, images, etc.) into a raw/ directory, then I use an LLM to incrementally "compile" a wiki, which is just a collection of .md files in a directory structure. The wiki includes summaries of all the data in raw/, backlinks, and then it categorizes data into concepts, writes articles for them, and links them all. To convert web articles into .md files I like to use the Obsidian Web Clipper extension, and then I also use a hotkey to download all the related images to local so that my LLM can easily reference them.

IDE:
I use Obsidian as the IDE "frontend" where I can view the raw data, the the compiled wiki, and the derived visualizations. Important to note that the LLM writes and maintains all of the data of the wiki, I rarely touch it directly. I've played with a few Obsidian plugins to render and view data in other ways (e.g. Marp for slides).

Q&A:
Where things get interesting is that once your wiki is big enough (e.g. mine on some recent research is ~100 articles and ~400K words), you can ask your LLM agent all kinds of complex questions against the wiki, and it will go off, research the answers, etc. I thought I had to reach for fancy RAG, but the LLM has been pretty good about auto-maintaining index files and brief summaries of all the documents and it reads all the important related data fairly easily at this ~small scale.

Output:
Instead of getting answers in text/terminal, I like to have it render markdown files for me, or slide shows (Marp format), or matplotlib images, all of which I then view again in Obsidian. You can imagine many other visual output formats depending on the query. Often, I end up "filing" the outputs back into the wiki to enhance it for further queries. So my own explorations and queries always "add up" in the knowledge base.

Linting:
I've run some LLM "health checks" over the wiki to e.g. find inconsistent data, impute missing data (with web searchers), find interesting connections for new article candidates, etc., to incrementally clean up the wiki and enhance its overall data integrity. The LLMs are quite good at suggesting further questions to ask and look into.

Extra tools:
I find myself developing additional tools to process the data, e.g. I vibe coded a small and naive search engine over the wiki, which I both use directly (in a web ui), but more often I want to hand it off to an LLM via CLI as a tool for larger queries.

Further explorations:
As the repo grows, the natural desire is to also think about synthetic data generation + finetuning to have your LLM "know" the data in its weights instead of just context windows.

TLDR: raw data from a given number of sources is collected, then compiled by an LLM into a .md wiki, then operated on by various CLIs by the LLM to do Q&A and to incrementally enhance the wiki, and all of it viewable in Obsidian. You rarely ever write or edit the wiki manually, it's the domain of the LLM. I think there is room here for an incredible new product instead of a hacky collection of scripts.

[chris muscarella @cm]

@fromedome there

[chris muscarella @cm]

And I'd also argue that the real power in brand equity takes time and durability to build— this chart from 7 Powers (Hamilton Helmer) shows life stage of business where it's developed...

[chris muscarella @cm]

@fromedome and how we managed to conflate rented distribution and some lipstick with “brand” which you’d actually be able to measure over time as margin structure, organic customers, and borrowing the Buffett quote of not needing a prayer session to raise prices