pradn

> When I was taken to the Tate Modern as a child I’d point at Mark Rothko pieces and say to my mother “I could do that”, and she would say “yes, but you didn’t.”

Actually, no you couldn't. The subtlety of the choice of colors, their shading, and their soft shaping, and the program of their creation over many years - you couldn't do that. They're lovely and sublime, and wonderful and an abyss. If you want to throw all that away and reduce it two boxes of paint, go ahead - but you'll be wasting a lifetime's engagement, of the joy of seeing with your intellect wide open.

> The value got extracted, but compensation isn't flowing back. That bothers me, and it deserves a broader policy conversation.

It bothers me, too. But, look at the history of the internet. There's no reason to expect we'll be able to fix this problem.

1. Search engines drove traffic to news/content sites, which monetized via ads. Humans barely tolerate these ad filled websites. And yet, local news went into steep decline, and the big national players got an ever-larger share of attention. The large, national sites were able to keep a subscriber-based paywall model. These were largely legacy media sites (ie: NYT).

2. News sites lost the local classifieds market, as the cost of advertising online went to zero (ie: Craigslist). This dynamic was a form of creative destruction - a better solution ate the business of an older solution.

3. Blog monetization was always tough, beyond ads. Unless you were a big blog, you couldn't make a living. What about getting a small amount of money per view from random visitors? The internet never developed a micro-payment or subscription model for the set of open sites - the blogosphere, etc. The best we got were closed platforms like Substack and Medium, which could control access via paywalls.

All this led to the internet being largely funded through the "attention economy": ads mostly, paywalls & subscriptions some.

The attention economy can't sustain itself when there are fewer eyeballs:

1. Tailwind docs have to be added just once to the training set for the AI to be proficient in that framework forever. So one HTTP request, more or less, to get the docs and docs are no longer required.

2. Tailwind does change, so an AI will want to access the docs for the version its working with. This will require access at inference time. This is more analogous to visiting a site.

All this measurement is useful only if you change your behavior in response. How often is this the case?

A common pattern I see is data-plane nodes receiving versioned metadata updates from the control-plane. As long as the version is higher than the node's previous one, it's correct to use. So, the metadata is a sort of monotonic counter with a bag of data attached to it. This pattern produces a monotonic counter, which I assume is a naive sort of CRDT - though the data itself doesn't benefit from CRDT merge semantics. In this world, as long as a node gets an update, its going to incorporate it into its state. In the article's terms, the system has Strong Convergence.

I'm trying to figure out under what practical circumstances updates would result in Eventual Convergence, not Strong Convergence. Wouldn't a node incorporate an update as soon as you receive it? What's causes the "eventual" behavior even after a node gets an update?

It seems to me the trouble is actually getting the update, not the data model as such. Yes, I realize partial orders are possible, making it impossible to merge certain sequences of updates for certain models. CRDTs solve that, as they're designed to do. (Though I hear that, for some CRDTs, merges might result in bad "human" results even if the merge operation follows all the CRDT rules.)

We can work toward closing security gaps with new technology, yes. It is necessary for large-scale adoption of LLM tech.

I didn't see this, thank you! They have a follow-up as well:

https://machinelearning.apple.com/research/ferret-ui-2

I have used it actually! It's been years so the fact that it's an accessibility tree just like in a browser didn't come to mind immediately. Both Mac and Windows have such representations for native apps. The actual functionality apps and the accessibility clients support is something like a two-way negotiation. A lot of stuff that should be supported in apps, in theory, is not, just because no client supports it, etc.

It's not a smart move to avoid integrating the most important capability advance in computing in the past decade - LLMs. They do support small cases, like summarizing text. But there's scope to do more.

They're stuck on the privacy angle, because what it means is you can't call remote services. You'll always have access to more resources at a data-center than a phone. So, while the frontier of what's possible with purely-local models will keep advancing, it'll never exceed what's possible with remote models.

People care about extra privacy when the delta in capability is minimal. But people won't allow a massive discrepancy, like the difference between a 8B model and a 700B model.

Apple needs on-device AI to do chores for me with the apps I have installed. Apple has everything it needs:

* Apps are already logged in, so no extra friction to grant access.

* Apps mostly use Apple-developed UI frameworks, so Apple could turn them into AI-readable representations, instead of raw pixels. In the same way a browser can give the AI the accessibility DOM, Apple could give AIs an easier representation to read and manipulate.

* iPhones already have specialized hardware for AI acceleration.

I want to be able to tell my phone to a) summarize my finances across all the apps I have b) give me a list of new articles of a certain topic from my magazine/news apps c) combine internet search with on-device files to generate personal reports.

All this is possible, but Apple doesn't care to do this. The path not taken is invisible, and no one will criticize them for squandering this opportunity. That's a more subtle drawback with only having two phone operating systems.

They have a super fast and slick file storage app. Some of the features that are natural additions to that feature set work quite well, like document scanning. But so much of what Dropbox does seems like they can't stay put and be happy with their core offering. Of course, they have to do this to increase revenue, for fear of becoming a mere commodity. It's tough.

Thanks for writing a balanced article - much easier to take your arguments seriously! And a sign of expertise.

How can anyone actually defend these silly, self-defeating tariff maneuvers?

There's already this hilarious bot. It's able to use people's outfits to woo them, or insult them. It's pretty good!

https://www.instagram.com/rizzbot_official/

Random sampling over time is substantially as effective as having someone enforce the law 100% of the time. It's something like how randomized algorithms can be faster than their purely-deterministic counterparts, or how sampling a population is quite effective at finding population statistics.

All this points to "personality" being a big -- and sticky -- selling point for consumer-facing chat bots. People really did like the chatty, emoji-filled persona of the previous ChatGPT models. So OpenAI was ~forced to adjust GPT-5 to be closer to that style.

It raises a funny "innovator's dilemma" that might happen. Where an incumbent has to serve chatty consumers, and therefore gets little technical/professional training data. And a more sober workplace chatbot provider is able to advance past the incumbent because they have better training data. Or maybe in a more subtle way, chatbot personas give you access to varying market segments, and varying data flywheels.

I'm sure we're already there with musicians as well. If you take Taylor Swift's annual revenue and multiply it by 5-10, you'll easily cross a billion. But she does require tons of staff for the stadium shows that provide the bulk of the revenue, etc. Not literally one person, but it's a company with a bus factor of one.

Just complaining that the world is bad is a good way to waste your energy and end up being a cynic.

So why is not all information organized in structured, open formats? Because there's not enough of an incentive to label/structure your documents/data that way. That's if you even want to open your data to the public - paywalls fund business models.

There have been some smaller successes with semantic web, however. While a recipe site might not want to make it easy for everyone to scrape their recipes, people do want Twitter to generate a useful link preview from their sites' metadata. They do that with special tags Twitter recognizes, and other sites can use as well.

The good news is that LLMs can generate structured data from unstructured documents. It's not perfect, but has two advantages: it's cheaper than humans doing it manually, and you don't have to ask the author to do anything. The structuring can happen on the read side, not the write side - that's powerful. This means we could generate large corpuses of open data from previously-inaccessible opaque documents.

This massive conversion of unstructured to structured data has already been happening in private, with efforts like Google's internal Knowledge Graph. That project has probably seen billions in cumultative investment over the years.

What we need is open data orgs like Wikipedia pick up this mantle. They already have Wikidata, whose facts you can query with a graph querying language. The flag example in the article could be decomposed into motifs by an LLM and added to the flag's entry. And then you could use SPARQL to do the structured query. (And that structured query can be generated from LLMs, too!)

LLMs and structured data are friends.

Finally, someone from the product side got a word in. Keep it simple!

A silly question: do any of these models generate pixels and also vector overlays? I don't see why we need to solve the text problem pixel-for-pixel if we can just generate higher-level descriptions of the text (text, font, font size, etc). Ofc, it won't work in all situations, but it will result in high fidelity for common business cases (flyers, websites, brochures, etc).

You can't avoid sending the user photos in a news article, for example. So the best you can do is start fetching/rendering the page 200ms early.

I'm with you - there are few elevators, yes. Still somewhat clunky for me.

Cyberpunk is certainly an achievement for technical graphics - fog, light, reflections etc. These are all enhanced with cutting-edge hardware.

What I'm trying to say is that we're quite far along in one visual axis, but for the whole experience to be amazing, we have a lot of work to do in other axes. Clunky character animations and body shapes take away from the amazing fidelity we get with light/fog, etc.

A large section of the gaming public sees photo-realistic games as serious, and prefers them for high-budget games. It's a rat race for devs though - its just incredibly expensive to create high quality models, textures, maps.

I've been playing Cyberpunk 2077, and while the graphics are great, it's clear they could do more in the visual realm. It doesn't use current gen hardware to the maximum, in every way, because they also targeted last-gen consoles. I'm thinking in particular of the PS5s incredibly fast IO engine with specialized decompression hardware. In a game like Rachet and Clank: A Rift Apart, that hardware is used to jump you through multiple worlds incredibly quickly, loading a miraculous amount of assets. In Cyberpunk, you still have to wait around in elevators, which seem like diegetic loading screens.

And also the general clunkiness of the animations, the way there's only like two or three body shapes that everyone conforms to - these things would go farther in creating a living/breathing world, in the visual realm.

In other realms, the way you can't talk to everyone or go into every building is a bit of a bummer.

Recently, some of it seems to be just to highlight raytracing hardware. Cyberpunk uses a lot of metal reflective surfaces to give a futuristic/tech vibe. But that's one sort of futurism. There'll be plenty of use of natural stone, wood, and tile far far into the future.

> Btw, we are releasing GPT-5 soon, and we’re excited for you to try it. But just to be clear: the IMO gold LLM is an experimental research model. We don’t plan to release anything with this level of math capability for several months.

GPT-5 finally on the horizon!

I can't imagine voluntarily giving access to my data and also being "scared". Maybe a tad concerned, but not "scared".

They say its easier to change your company to fit SAP than to mold SAP to fit your company.

I hear about the "Django philosophy" in this thread. What exactly is it?

I’ve bought a bunch of games just to support the creators, to vote with my dollar and say: I want more of this. Dwarf Fortress comes to mind as an example. I’ve got to sit down and learn to play it. Will probably take a week!

I couldn't immediately find information about how the metadata is stored? Is it one shadow file per RAW file, as I've seen in other OSS RAW editors?

I'm not sure what the perfect solution is, but it is hard to sync a ton of shadow files to cloud storage, versus one big catalog file.

Is the metadata in an open format, so I can take the edits to other programs?

I am glad there's alternatives to having to shell out for Light Room every month. I only need to edit RAW files after holidays!