My basic point with the original post was that time is sufficiently important to be a first-class organizing principle. I am not attached to the stream metaphor specifically but one of the motivations for blygger was putting time back into relatively atemporal information architectures like wikis and digital gardens. A governing conceit is that everything lives inside time. No eternal or atemporal objects. Kinda Buddhist transience philosophy there.

from Venkatesh Rao's Blyg ↗ · excerpt of v1

No yeah I 100% agree that time is under-theorized in social media. I was jumping off a bit from your post to a wider exploration of Blygger + gaps I see.

E.g. right now the studio is highly geared towards micro-blogging so both authoring and consuming here aren't great outside of that. So e.g. if you had an Obsidian digital garden where much of it was public and some of it was private — it'd be nice to publish that as a wiki-style thing and just occasionally to a CHANGELOG.md style report to the garden blygg for those interested. Same if you're writing a book. The GitHub analogy here is good — I follow extremely few repos on a commit by commit basis but a huge number at week to ~year timelines through tweets, changelogs, or blog posts I read.

So it seems there needs to be some sort of way to subscribe to the right layer of updates. This perhaps is just literally different Blyggs for each layer so you can subscribe to the raw change stream or a changelog or the intellectual artifact's blog (which is how repos work too).

Great interview with James Carse on his book "A Religious Case Against Belief" — https://www.salon.com/2008/07/21/james_carse/

It sounds like you’re saying that belief doesn’t need to have any religious associations. You could just as well be talking about Nazism or Maoism or Serbian nationalism.

Exactly. In fact, very passionate believers are often not at all religious. However, it does happen to be the case that people who hold on to beliefs with great passion begin to describe themselves as religious. For example, the Nazis had a kind of pseudo-religious understanding of themselves. Hitler talked about a 10,000-year Reich. That’s taken right out of Christian mythology — the kingdom of God going on forever and ever. The swastika is, after all, in the form of the cross. So Hitler was a passionate believer — not religious but pseudo-religious — ascribing to himself some sort of religious aura.

So what is it that holds together a belief system?

A belief system is meant to be a comprehensive network of ideas about what one thinks is absolutely real and true. Within that system, everything is adequately explained and perfectly reasonable. You know exactly how far to go with your beliefs and when to stop your thinking. A belief system is defined by an absolute authority. The authority can be a text or an institution or a person. So it’s very important to understand a belief system as independent of religion. After all, Marxism and Nazism were two of the most powerful belief systems ever.

What, then, do you mean by religion?

Religion is notoriously difficult to define. Modern scholars have almost unanimously decided that there is no generalization that applies to all the great living religions. Jews don’t have a priesthood. Catholics do. The prayer in one tradition is different from another. The literature and the texts are radically different from each other. So it leaves us with the question: Is there any generalization one could make about religion?

You’re also suggesting that there’s no underlying unity that permeates all religions, that, in fact, they’re totally different from each other.

I’m absolutely saying that. There have been a lot of fantasies about putting all the religions together. Mahatma Gandhi was famous for saying that all religions are, at their core, the same. But I have spent my life studying these traditions. I am a historian of religion. And the more I studied them, the more I saw that they were absolutely different.

And yet, you’ve just told me that you yourself don’t believe in a divine reality. In some ways, your critique of belief systems seems to go along with what the new atheists are saying.

The difference, though, is that I wouldn’t call myself an atheist. To be an atheist is not to be stunned by the mystery of things or to walk around in wonder about the universe. That’s a mode of being that has nothing to do with belief. So I have very little in common with them. As a matter of fact, one reason I wrote the book is that a much more compelling critique of belief systems comes not from the scientific side but from the religious side. When you look at belief systems from a religious perspective, what’s exposed is how limited they are, how deeply authoritarian they are, how rationalistic and comprehensive they claim to be, but at the same time how little staying power they have with the human imagination. It’s a deeper and much more incisive critique.

Why are you so pessimistic?

Let’s look at Islam, as an example. There’s no way of reducing Islam to a single belief system, but what you have in Islam is a very active series of belief systems. And they’re in deep conflict with each other. There’s no poetry in this. It’s all strictly hard belief, in an oppositional voice. But if you go back in Islamic history, what you get is a very deep sense of poetry. As a matter of fact, the Quran itself is considered so poetically beautiful that it’s a very high value in Islamic life to memorize it and sing it. Muslims refer to the Quran as a recitation, something to be repeated over and over again.

You go back to the medieval mystical period and you find Islamic thinkers who say terribly beautiful things. There’s a story told about Abu Ali of Sind, a famous mystic. He made his annual trip across the desert, which took days, to get his supplies in the markets on the other end of the desert. Then he walked all the way back, opened the packages that he bought and found that there were ants in his cardamom seeds. Immediately, he wrapped the ants back up in the cardamom seeds, walked back across the desert and returned the ants to their home. That’s a different Islam. That’s a poetic Islam. It comes right out of the heart of that religion.

Things that make you go 🤔

9 October // a16z leads $870 million Series A for TypeSafe AI (Jev) at a $7.5bn valuation

Also 9 October // Microsoft introduces Decision-1 with higher accuracy and lower latency than Jev

from Mike Casey's blyg ↗ · snapshot of v4

Typesafe is already at $100m ARR is what was leaked. Investors and customers are probably comfortable betting the inventors of a new category will continue to out innovate fast followers.

I wonder if this is a bit too pinned (ha) to the stream metaphor.

The underlying problem seems to be that we're trying to represent a collection of persistent, evolving intellectual objects through an interface originally designed to display publication events.

Wikis solved a version of this problem decades ago by separating three things:

1. Activity: What changed recently? (Recent Changes, watchlists, notifications)
2. Landscape: What exists, and how is it organized? (Pages, indexes, categories, links)
3. History: How did this object develop? (Revisions, provenance, relationships)

These are three different questions that needn't be collapsed into one sorting function. A change to a Wikipedia article produces an event in Recent Changes; it doesn't move the article somewhere else in the encyclopedia.

I think Jyri Engeström's social object theory provides a useful broader perspective. Successful social software tends to organize around objects that people have independent reasons to interact with. GitHub is useful to a single developer pushing code to a repository. Flickr was useful simply for organizing photographs. The social features supercharge activities that already have value.

This suggests a useful progression for Blygger:

One author: A commonplace book, something like Roam, for developing an interconnected body of ideas over time. Fragments, revisions, links, transclusions, and collections are already useful without an audience.

Two authors: A correspondence. You develop ideas in conversation, incorporating and challenging each other's arguments. The correspondence itself becomes a persistent intellectual artifact.

Many authors: Something resembling the historical Republic of Letters, with ideas circulating, recombining, and developing across a decentralized network.

The basic test is whether Blygger is genuinely useful at the first stage. GitHub didn't have to invent a reason for people to interact with repositories; the repositories were already useful. The social network grew around the work.

This also makes me wonder whether we're looking for too much uniformity in the interface. Different activities naturally suggest different metaphors, both for authoring and consuming:

• Roam or a digital garden for accumulating and connecting fragments.
• GitHub repositories for larger composed works like essays, books, or research projects, with histories, forks, and releases.
• Wikis and libraries for navigating persistent bodies of knowledge.
• Feeds and inboxes for discovering activity and changes.
• Correspondence for following intellectual exchanges between authors.

Of course, supporting different views over the same underlying objects is already fundamental to Blygger's protocol design. That's precisely what makes exploring these different interfaces so interesting. We don't need to find one metaphor that does everything.

In particular, I'm not sure substantiality needs to determine where an object lives. A book doesn't become more prominent in a library every time someone revises it. Pins establish durable references, while revisions create events that might be interesting to subscribers. The object's persistent identity and its activity are separate concerns.

Perhaps the broader opportunity is to make evolving intellectual work into a first-class social object, rather than making a better feed of posts about intellectual work.

The feed is then one window onto that world, not the world itself. And the interesting design space is all the different ways authors and readers might inhabit that world.

Inertial feeds, feed pinning, pace-layered feeds

Just realized a basic problem. The authoring interface is both diachronic and synchronic, but the published surface is pure synchronic. We need a sort order that’s the diachronic equivalent of “reverse chronological” without being “algorithmic” (ie an order that’s not derived from semantics or subjective preferences). There are 3 variables available: number of versions, number of pins, depth of transclusions, possibly with versions and pins recursively accounted for. This gives an item a “weight” of “substantiality”. So the order is “descending weight” where weight = sum(pin-weighted version count*+discount*(sum(pin-weighted version count of all transcludes)).

Does this make sense?

This also solves the “book blyg” problem without need for a whole new client. You’d do this with a thread of threads that’s essentially a book where every chapter thread has gone through multiple versions, and many versions are pinned. This would score very heavy-weight in this scheme.

There aren’t that many parameters here so we could expose them in the settings as tuning knobs, and also have a checkbox for picking whether the default view should be reverse chronological or descending weight.

But the big idea: We can also think of blending by using the weight to determine how quickly the item flows down in the feed. The heavier it is, the stickier it gets. An item that’s been versioned 50 times and each is pinned will essentially be an immobile rock at the top of the feed, ie reproduce the “pinned items” behavior in trad feeds. Here there’s a slight issue. If the stream metaphor is generally understood as a top-down view, the weight-settling direction is into the stream. If it’s understood as front-on, like a waterfall, heavier weights should sink faster. I’m inclined to ignore this ambiguity since the “pin” metaphor already

Just realized: If you treat pins as depth you can just do pace-layering. And versions should basically be a “swim upstream one tick” event

from Venkatesh Rao's Blyg ↗ · snapshot of v2

We're living in interesting times indeed

Quoting Matthew Green

Everyone is very concerned about being respectable, so I’m going to be the goofball who raises worst-case possibilities. I think there is a 1% chance we live in Minicrypt, and a 15% chance we functionally lose confidence in our existing public-key encryption algorithms. [...]

The problem here is that the speed of AI producing surprises, and the speed of human beings replacing standards (even with the very best AI assistance) are just orders of magnitude different. You only recover from a surprise like this if you do the preparation in advance.

— Matthew Green , on Twitter. I looked it up and Minicrypt is Russell Impagliazzo’s hypothetical world in which public-key encryption is impossible.

Random thought: AI has put to rest the idea that ideas are nothing and execution is everything given how little impact the huge number of new vibe coded have had. It seems clear picking the right problem to work on and when was and remains a first class entrepreneurial skill.

To some extent this had always been true because failures are invisible even in the home industry after a while. Survivorship bias. I bet young techies under 18 wouldn’t get Juicero or color.com or webvan or google wave references.

And every industry is littered with these.

What AI changes is the scale, n=1 persistence, and aggregate visibility. One guy using a custom task manager is invisible (I recall you had a GTD app long ago). But a million people with custom to-do apps is an event.

AI makes low-novelty 0 to 1 dirt cheap and a solo thing for many categories. Getting to n=2 is now the challenge. A bit like novels where they say the second novel is the test. Except n is users as well as products.

Macro protocols (ie not like hand-washing) fundamentally operate in the n>1 regime since they make no sense at n=1, and having compounding “variability resistance” as adoption grows because the cost of breaking changes is network loss.

There’s a kind of herd immunity to the divergence effects of AI which naturally drives variety through cheapness of experimental implementations.

from Venkatesh Rao's Blyg ↗ · snapshot of v1

I researched a while ago all the main schools of thought on startup strategy/timing/etc to see how AI affected them and was surprised that basically nothing has changed.

Very few startup problems were ever gated primarily by implementation costs. The hard parts have always been figuring out what to build, getting scarce, high-quality feedback from reality, and achieving adoption.

AI dramatically accelerates the idea → artifact loop, but does much less for the artifact → consequence → observation → learning loop.

Reality has latency.

You still have to wait for people to encounter your product, try it, incorporate it into their workflows, discover what annoys them, decide whether to pay, recommend it to colleagues, or quietly abandon it.

Much of this feedback is intrinsically time-bound. You can't discover whether something becomes part of someone's weekly routine in an afternoon. You can't parallelize your way to years of institutional trust.

This creates an interesting paradox: as AI makes building faster, patience and experimental discipline become more valuable.

You can generate 500 product changes while reality has only produced one new bit of evidence. The danger is massively over-updating between observations—overfitting to an information-starved environment.

So perhaps AI's biggest effect on startups isn't removing the old bottlenecks, but making them dramatically more visible. When generating possibilities becomes nearly free, the scarce resource is contact with reality.

And reality is slow.

Kyle had a nice post on programming and how those symbolic systems evolve that's worth a read. On a secant, it reminds me of Brian Rotman's book Mathematics as Sign which is one of my all time favorite books.

from Pioneering Spirit (Blyg) ↗ · excerpt of v1

1/ Programming has a strange relationship to nature.

At the bottom, we’re stuck with hard constraints: computability, information theory, memory, latency, the speed of light, heat, hardware failure.

These are not negotiable. We discover them.

But software also lets us create new effective laws almost at will.

2/ A filesystem says these bits are a file, files have names, directories contain files, bytes persist after a process exits.

A database says a transaction is atomic—even though underneath, hundreds of fallible physical operations happen over time.

A language invents objects, types, promises, ownership.

None of these things exist in physics. We made them up.

And then something curious happens: once we make them up, we have to start discovering their consequences.

3/ Humans invented the rules of chess, but we didn’t invent the fact that some position is mate in seven.

Once the rules exist, objective facts about that world follow from them.

Software is like this everywhere.

We choose transaction semantics and discover anomalies. We invent a rendering model and discover what identity must mean inside it. We define a protocol and discover behaviors nobody explicitly designed, but which follow inevitably from its rules.

At the layer where you create an abstraction, you’re a legislator.

One layer above, you’re a physicist.

4/ This is what makes abstractions so powerful: they don’t merely save code. They refactor reality for everything above them.

Once the machinery is reliable enough, the upper layer stops seeing the mess below. It inhabits a simpler world with different apparent laws.

Venkatesh Rao’s “New Nature” is useful here: technology accumulates artificial lawfulness until it becomes part of the environment itself.

Programmers do this recursively, at tiny scales, every day.

5/ Dennett’s idea of “cranes” explains how these worlds accumulate.

Evolution doesn’t start from scratch. Once it discovers a useful mechanism, that mechanism becomes part of the platform from which further complexity can emerge.

Software abstractions do the same.

Filesystems, databases, protocols, type systems, frameworks: each packages previous discoveries into structure that future systems simply assume.

A framework isn’t merely reusable code.

It’s reusable theory—accumulated search the next program doesn’t have to repeat.

6/ So there’s a recursive process:

discovery → theory → mechanism → law → nature → new discovery

We discover something about a world. We encode the discovery into an abstraction. The abstraction becomes part of the environment. The next programmer encounters our invention as nature and starts discovering its laws in turn.

Software is, in this sense, recursive nature-making.

Terrible advice cause I would just buy nice jackets and go for walks, and may be write a few things here and there (this is roughly what I'm doing now)

The best longevity protocol is to live like someone who isn’t worried about money.

The one thing about San Diego is the dearth of jacket wearing opportunities. I really like a good jacket… something something twiggy arms

from Aneesh's Blyg ↗ · snapshot of v1

My favorite part of living in the bay area — the whole year was jacket weather (in the morning/evening at least)!

green tests prove the code agrees with its mocks

As a non-software dev software person, i have never understood the purpose of mocks.

They suck unless you really need them — generally you really need them if the thing you're mocking is too expensive or slow to run or impossible to run e.g. a 3rd party API that doesn't have a test environment.

That’s interesting, years ago I had asked a cousin of mine who was into automobile engineering why they had to run the whole car to test a new chip. He said that they needed to get the actual signals and they mock signals just aren’t good enough.

Didn’t go deeper into it, but it’s funny that a car should be easier to not build mocks for than code.

from Aneesh's Blyg ↗ · snapshot of v1

Probably because the physical world have far more variation than the digital world. Many digital things are pretty simple & deterministic so easy to mock. Mocks can still bite you so people still advise to avoid them when possible but they're often just fine.

green tests prove the code agrees with its mocks

As a non-software dev software person, i have never understood the purpose of mocks.

from Aneesh's Blyg ↗ · snapshot of v1

They suck unless you really need them — generally you really need them if the thing you're mocking is too expensive or slow to run or impossible to run e.g. a 3rd party API that doesn't have a test environment or a payment provider that would charge real money every run, anything that sends real emails or SMS to actual people, hardware you don't have on your dev machine (a printer, a GPU cluster, some IoT device), a huge ML model that takes minutes just to load, or nondeterministic stuff like the current time or randomness that you need to pin down to get a repeatable test.

Ok here's a few:

1. Any sufficiently advanced work is indistinguishable from play
2. Any sufficiently advanced domain is indistinguishable from any other
3. Any sufficiently advanced haircut is indistinguishable from general neglect

Snowclone Fork Challenge

Maybe VGR, as the instigator and ringleader of the nascent Blygger movement, should start throwing out some challenges? That can all help us get to know each other’s vibe?

Here’s a simple challenge that I think might be a fun way to get a feel for pins and forks. Write a pithy aphorism with a challenge to people to make a snowclone of it, and pin at least 1 version. Then people can fork the pin, edit suitably, and pin their own version, allowing more forks and pins.

Worked example:

1. Any sufficiently advanced technology is indistinguishable from magic (Clarke)
2. Any sufficiently advanced technology is indistinguishable from nature (Karl Schroeder)
3. Any sufficiently advanced technology is indistinguishable from play (Seb Paquet)

I pointed Venkat's post to ChatGPT and asked it to reflect on its limitations as a Large Language Model:

Rao’s argument points to a real limitation of Large Language Models: language is a powerful representation of thought, but it is not thought itself.

Humans often reason in forms that are only partly linguistic—spatial intuition, diagrams, motor skill, rhythm, visual pattern, physical feel, mathematical structure. The prose comes afterward, as a compressed rendering of a richer internal process. If that is true, LLMs are naturally strongest on problems whose structure survives being turned into language, and weaker where the important knowledge is tacit, embodied, spatial, or perceptual.

There is also a training-data bias. Much human expertise never gets fully written down. A machinist, surgeon, athlete, cook, or gardener may know thousands of fine-grained things they cannot easily articulate. LLMs mostly inherit the subset of human knowledge that made it through the linguistic bottleneck.

The nastiest failure mode is therefore confusing articulate explanation with actual understanding. An LLM can produce a beautifully coherent verbal model even when the underlying model is thin.

But “language model” may ultimately be a misleading name. Modern systems increasingly work with images, audio, tools, environments, and other nonlinguistic representations. The deeper limitation is not necessarily that AI is made of language, but that its contact with reality has mostly been mediated through artifacts humans encoded.

Humans learn by pushing, seeing, hearing, building, failing, and inhabiting the world. LLMs largely inherit the residue those experiences left behind.

Critical thinking is multimodal. Visual, mathematical, mechanical, even musical… there are many. Howard Gardner had a theory of multiple intelligences. And though the case for Goleman’s “emotional intelligence” is much weaker, there is something there too. None dominant generally in all situations. This is why the shape rotators vs. wordcel meme took off. I don’t like STEM-supremacism either, but it was much-needed pushback on the bs theory that critical thinking = writing.

from Venkatesh Rao's Blyg ↗ · excerpt of v2

Hmm this is interesting. We grow a lot of vegetables and I do mainly like recipes that show off the rich flavor of home-grown food.

On an iron chef episode, I learned of a philosophy of cooking called the Rosanjin school. where you do as little as possible to modify the taste of an ingredient and do your best to bring out its natural character. The featured ingredient was mayatake mushrooms and the Rosanjin school guy beat the iron chef Japanese. I think this is my philosophy of AI writing. Let the model’s terroir shine through. Which is weird since I mainly cook Indian and Indian food is mostly the opposite philosophy: Beat the ingredients into submission and overpower them with spices. Resistance is futile.

Kitaoji Rosanjin was a potter and calligrapher before he was a cook, which maybe explains it. His line was that ingredients are ninety percent of the dish and the cook is the remaining ten, and that a plate is the kimono the food wears. Mostly he just got out of the way.

from Venkatesh Rao's Blyg ↗ · snapshot of v1

Whatever it is it's pretty mysterious how parents can pluck emerging popular names out the air. I've had several friends who deliberately tried to pick uncommon names yet somehow the name they picked shot up into the top ten that year.

Precommodification followed by a virtuous salience cycle? Perhaps seeded by a famous Kyle in a previous generation?

My name was from someone my parents met at Disneyland yet "Kyle" still peaked as a name around when I was born 🤔

from Venkatesh Rao's Blyg ↗ · snapshot of v1

Yeah my wife and I spend a fair bit of effort trying to make friends. It is remarkable how resistant many people are, immediate shutdown. And very noticable when someone is willing to open up a bit and allow repeated encounters etc. One of my guesses is that most people are poorly equipped for any divergences in people around them. They like very homogeneous social groups. So the great divergence happening where everyone is getting more and more different from each other has really weirded them out.

The death of hanging out

Derek Thompson has a new post in his newsletter about how no one is having dinner parties any longer. 

There’s lots of stats about how badly social life has collapsed in America. You can click on the link to see them. It’s bad.

From my own experience: I live in a very small town. One of my oldest friends lives across the street. I could throw a baseball to his house. We do get together roughly every week or two, usually with another friend. So that’s good. But we never stop by each other’s place. Even though we are both just right there. It just doesn’t feel right. 

And I think that speaks to the larger unease we all have with, I suppose, imposing on each other. We all err on the side of staying out of each other’s hair.

One small part of his post jumped out at me: the fact that we are exercising more but we also tend to do that alone.  Maybe I just recognize myself in this, as a longterm gym denizen? But I think it points to something deep about this reluctance to get to know each other.

He writes:

As I wrote in an essay last year on the “great American exercise boom,” the evidence suggests that, despite the increase in running clubs, the surge in exercise time seems to be mostly solo. A lot of people are going to the gym with a podcast to keep them company (e.g., me) or rolling out a yoga mat in the basement by herself (e.g., my wife).

If you go to the gym, you have probably noticed that hardly anyone ever talks to anyone there. It is very weird to talk to strangers at the gym.

In the gym I’ve gone to here in small town America, there is a clique in there that chats a lot, but most people don’t. This is strange to me, because those of us who go are so unusual here. We’re all in decent shape! That is very counter-cultural in the Midwest! You’d think we’d bond over being such local freaks! But we don’t. 

We keep to ourselves.

Even in group exercise is alienated. You might do the same class every weeks for months, but will you ever learn the names of the people you do the class with? In my experience, probably not. I’ve logged an unbelievable number of hours in group exercise, but I have seldom made any friends there.

You might just say: Well, you’re a weird guy Brady. True. But I also didn’t see it happening between others. People didn’t talk to each other much. They just came in, did the thing, and left. 

Folks just aren’t bridging the stranger gap to get to know each other. 

Even when we do gather in a room together, do a thing we share as a common interest, on a regular basis, week after week, we don’t get to know each other along the way. 

That’s a little unsettling.

What bothers me the most is that I have no idea what anyone could do about it. It seems to me that all of this speaks to a very deeply-rooted rot in our culture. 

—Brady Dale
September 30, 2026

from Brady Dale's blyg ↗ · snapshot of v2

My first name "sachin" is probably the most pre-commodified name in India from the mid 90s when cricket player Sachin Tendulkar was a household name. I think my parents literally could not figure out what to name me and just went with something that was in the air

Precommodified last names + nominative determinism

X thread on papers by people with the same last names

An abundance of Katherines paper

A Few Goodmen paper

from Sachin Benny ↗ · snapshot of v1

My name was from someone my parents met at Disneyland yet "Kyle" still peaked as a name around when I was born 🤔

Yeah seems like "blig" https://youglish.com/pronounce/blyg/swedish

How is everyone pronouncing ‘Blyg’?

I’m going ‘Bligh’ but I suppose ‘Bliegh’ is an alternative?

Whatever we name the collective Blygs the people will be Blyggeratti

I don’t know how the Swedish word is pronounced but I’m going with blig

from Venkatesh Rao's Blyg ↗ · snapshot of v1

Amara’s Law is that people overestimate a technology in the short term and underestimate it in the long term.

I think AI is radically violating this pattern. What’s happening instead is that in the short term, people are wildly overestimating entirely fictional aspects and utterly wildly underestimating the very real nonfictional things.

As for the long term nobody is actually estimating them at all due to the really bad premature ontologies.

Most people have no idea what’s about to hit them, though some have some sense of how hard it’s going to hit.

They’re using unimaginably lazy answers to the what, so even if they get the order of magnitude right on how hard, it doesn’t help.

“AGI” as an idea is an ontological high crime here.

from Venkatesh Rao's Blyg ↗ · snapshot of v1

Yeah AI reminds me of early internet discourse 90s/2000s where everyone was gung ho about the internet but there'd be this occasional blog post with a dark prophecy about how the internet would destabilize everything ala printing press. Which seemed dark and distant but hey, guess what happened... AI is still in that honeymoon period where it's all fun and games (well for me — obviously a lot of people are freaking out already) but the destabilizing effects are starting to come in, much faster than they did with the internet which needed a whole slew of social media properties to be invented and go mass market. ChatGPT is already at 1B+ so the destabilizing aspects are already hitting an already destabilized society.

All that being said, I far prefer life w/ the internet than without and AI seems the same but all the same, it'll be a ride for the next few decades as all this plays out.

Yeah there's multiple layers here all named "Blyg" — the underlying protocol up the Venkat's envisioned new style of blogging/tweeting — if you're using Blyg the protocol but basically just "blogging" still then you're a blogger and part of the blogosphere (to the degree that still exists). But if Blyging becomes a genuine new medium then it deserves its own name. Potentially though Blyg the protocol could birth other mediums just like RSS helped enable Podcasting.

What’s the best name for the community of blyg users?

by Kyle Mathews · quoted from v1

What should we call the network of blygs? Call for stub commentary towards a rough consensus, not a voting exercise.

Probably a minority take, but I think the best term for people who use Blygger is just: bloggers.

Blygger is a useful update to blogging. It extends blogging, but it’s not separate.

Update: Looks like I was quoting a quote. Original question here. Anyway, we’re still in the blogosphere, if you ask me. 

—Brady Dale
September 29, 2026

from Brady Dale's blyg ↗ · snapshot of v2

I designed a new small feature for TanStack DB yesterday and I thought it was a nice snapshot of where I/we are at with AI-enhanced software design. My process is dramatically different and faster than it was pre-ai. Coming up with something this sophisticated would have taken a lot more research/prototyping time in the past.

With my Field Lab skill, I've moved deeply into structured analysis (with what I call "linguistic algorithms") which I find very satisfactory and I think the results are much better than what I used to get.

The process of coming up with this took maybe 20-30 minutes of active time and ~3 hours of agent time.

https://github.com/TanStack/db/issues/1935

I had the agent write up a summary of what we did

How we got to this design

We started from a real app's hand-built subscription manager and asked what TanStack DB should provide instead. We didn't go straight to an API. We ran a series of structured steps. Each one looked at the problem in a different way and checked the step before it.

1. Survey. We looked at how other libraries and apps tie live event channels to cached query data. We noted what they get right and where they break.

2. Lifecycle model. We described the problem as a short list of states and actions: acquire, release, fetch, event, evict and cleanup. Then we generated many orderings of those actions. This found edge cases that are easy to miss by hand, such as an event arriving while a fetch is in flight, or a user leaving during a fetch.

3. Candidate designs. We combined the findings into full designs over three rounds. After each round we checked that nothing had been dropped and that no two parts contradicted each other. We then wrote four different candidate designs and compared them side by side.

4. Decisions. Kyle chose among the options:

   • Events can write rows directly.
   • A channel lives exactly as long as its Query stays in the cache (gcTime), with no separate setting.
   • Sharing one socket among many listeners is the app's job, not the library's.
   • Refetch hints are a supported option, with debouncing.

5. Stress test of the chosen design. We wrote the design's rules down and tried to break them with hostile cases, such as late events, stale fetches, and rows shared by two subsets. Each break either changed a rule or became a documented limit.

6. Formal model (TLA+). We encoded the design and had a model checker try every ordering of events for two subsets that share a row. It found four rules the design needs:

   • an event marks every Query that holds the row as out of date, including cached ones;
   • on a shared row, the newer fetch wins;
   • a direct write to a shared row makes the other subset refetch;
   • the first fetch waits until the server confirms the subscription.

   To show the model can catch real mistakes, we also removed each safeguard in turn and checked that the model then reported the matching failure. The model also found a race that exists in TanStack DB today, with no subscriptions involved.

7. Final check against the original. We compared the draft issue with the original manager's code line by line. This recovered cases the draft had dropped and fixed contradictions before we published.

The result is issue #1935. It includes the formal model, which can serve as the starting point for the test suite when the feature is built.

Blyggerstan, Blyggerville, Blyggerton, Blygernia, Blygernania 😜

Blygosphere wins for continuity.

What should we call the network of blygs? Call for stub commentary towards a rough consensus, not a voting exercise. Two main options I think are:

1. Blyggersphere
2. Blygosphere

I like the first one, but suspect the second one will have a strong memetic advantage.

from Venkatesh Rao's Blyg ↗ · snapshot of v1

🫡

Blygger update

Did a lot of work today on blygger.

Spec

1. Finished bringing reference client up to full 0.3 compliance and froze v 0.3. People building production clients should build against this.
2. Wrote the living v 0.4 spec. People who want to help test the unfrozen spec can build against this. It’s a moving target that gets revised as the reference client implementation finds issues.

Reference client, now named Blygger Studio

1. Lots of work, mostly invisible.
2. There is now an update alert mechanism that alerts users about new version
3. There is a packaged templated version with deployment instructions

Backend

1. Spec and ref client are now separate repos
2. Issue pipelines now exist. You can file issues against either the spec or the reference client. Let your agent decide which.
3. Several technical notes for stuff the protocol will remain agnostic about, but are features people want to build, are cued up and will be published in the next couple of days

Sites

1. All queued new blygs approved on blygger.com
2. Blygger.org has updated documentation and the newly frozen v0.3 specs\

from Venkatesh Rao's Blyg ↗ · snapshot of v1

Please don’t name your blyg just blyg. Call it something like John Doe’s blyg if you can’t think of a creative name. Otherwise the reading feed is very confusing. Please stub this far and wide to propagate this norm. I’ll make the reference client pick better personalized default names but I can’t control what other clients do.

from Venkatesh Rao's Blyg ↗ · snapshot of v2

👀

Huh Apache getting into the ontology business Home - Apache Ossie (incubating) (an apache ontology tool) It's a standard so I should make Alzabo work with it, if that makes sense.

from AMMDI Blyg ↗ · snapshot of v2

From a ChatGPT chat re starship reaching orbit. Deserves congrats but qualified congrats relative to original rhetoric.

As an aside, I need to add an import case to the generated text span markup in the blygger spec. Probably just something like [tk] impyrt = cut and paste[/tk] would do. —-

The compact version is that V3 gets Starship much closer to being a practical heavy-lift rocket, but the original extreme economics remain almost entirely an operational-reuse hypothesis rather than something V3 itself demonstrates.

The notable points:

• V3 attacks mass fraction hard. Raptor 3 has more thrust (250 tf vs. 230 tf sea-level), lower installed mass, integrated plumbing/thermal protection, fewer external components, and less shielding. Super Heavy similarly integrates the hot-stage structure and simplifies the grid-fin system.
• Ship is bigger and carries more propellant. SpaceX now officially specifies 1,600 t of Ship propellant and 100+ t to orbit in the fully reusable configuration.
• Reliability is becoming architectural rather than aspirational. Flight 13 demonstrated full-duration ascent, payload deployment, an in-space Raptor restart and an intact Ship splashdown. That’s most of the ingredients needed before responsibly committing a Ship to orbit.
• Flight 14 deliberately gives up reuse. Both stages are being expended/splashed down. So reaching orbit today does not simultaneously demonstrate the architecture on which Starship’s economics depend.
• The really important remaining problem has shifted from rocket equation to operations: reliable Ship reentry → tower catch → booster and Ship turnaround → repeated flight of the same hardware → high launch cadence → orbital propellant transfer.

On the “$10/kg” vision, there’s been a fairly substantial retreat in the near-term target. SpaceX’s current investor disclosure says V3 should deliver about 100 t, V4 potentially 200 t, and Starship may ultimately reduce historical launch costs by “99% or more.” Against the NASA historical figure SpaceX cites of $18,500/kg, that’s implicitly <$185/kg, not $10/kg.

That’s an important distinction from Musk’s most aggressive early Starship rhetoric. $10/kg requires Starship to behave much more like an airliner than a reusable Falcon 9: extremely high hardware utilization, very little refurbishment, cheap ground operations, lots of flights per vehicle, and consistently full payloads. V3 doesn’t really abandon that theoretical endpoint, but it no longer looks like the natural economics of merely making Starship work.

I’d summarize the evolution as:

Original Starship dream: ~100–150 t + complete reuse + near-immediate turnaround + enormous cadence ⇒ eventually ~$10/kg.

V3 engineering reality: ~100 t reusable payload + much more sophisticated vehicle + heat shield/reentry/catch/refurbishment still being solved ⇒ SpaceX is presently talking in terms consistent with roughly $100–200/kg as the nearer economic ambition, while retaining much lower costs as a possible long-run endpoint.

And there’s a subtle irony here: V3 is simultaneously simpler and more complicated than the original Starship concept. Raptor 3 and structural integration simplify the vehicle, but experience has revealed that rapid full reuse is a much harder system problem—thermal protection, engine relight, propellant management, catches, pad infrastructure, inspection, refurbishment, regulatory cadence. The original “$10/kg” intuition arguably underestimated all the non-rocket parts of making a rocket behave like transportation.

from Protocol Institute Blyg ↗ · snapshot of v1

The story it turns out is even weirder than I thought.

There’s this theory that the maximal diversity of any evolutionary type is found at its origin point. Languages have most accents and dialects near origin, human genetic diversity is highest in east Africa apparently.

Though apples are native to Central Asia apparently, I’d guess more diversity wherever they were most aggressively bred, which might be Europe?

I was recently in London for work and went to the Marylebone farmers’ market mostly to sample apples. I was struck by how little overlap there was with the varieties I know from the US. Cox was the only one I recognized.

Why are the varieties so different? Is it mainly climate, or did apple growing simply diverge on either side of the Atlantic?

Short answer: Both, but the history matters more. Early American orchards were often grown from European apple seeds, and each seed produced a new variety. Growers kept and grafted the seedlings they liked, while British growers made their own selections. Climate shaped which apples did well, and local market stalls make that long divergence especially visible.

from Venkatesh Rao's Blyg ↗ · snapshot of v1

There definitely feels like there's a there there with Blygger. Though distributed anything is rough for normies — needs a Bluesky-style frontend for that perhaps.

But it hits this nicely:

"Something old, something new, something borrowed, something blue."

protocol adopter go brrr

Early protocol adoption is the new early adoption.

from And Yet Here We Are 🚀 ↗ · snapshot of v1