andyferris

The OS's do provide webviews pretty uniformly, and the browser engines they ship with behave more similarly than in the days of IE.

I get some of this... but the argument around versions like 3.10 being footguns in YAML apply here too - in both YAML and SOML both you want to quote that as a string. (I suppose in SOML the user might be used to quoting strings, but in both cases you'd still want your application-level parser to reject floats for that field for safety's sake).

I think many here would say the hyperscalers enjoy that the masses can't afford, e.g., ample local RAM (especially GPU RAM) because then they are encouraged to "rent" it from them. And that could lead to a feedback cycle where no-one could independely run models, due to (artificially created) economic conditions. And going further, if your personal economic prosperity is dependent on running capable LLMs, then while you may not be "enslaved" per se but you are certainly forced to pay these large companies for the racket they created and benefit from.

(No one is saying the RAM is "enslaved")

EDIT: upon re-reading, maybe we agree

I am a bit confused on your position - are you saying HOA's are uniquely awful, or saying that we should all have equal access to local RAM (and Sam Altman is like the chairman of the biggest & least efficient HOA ever)?

I'm from Australia and always assumed the word "shithouse" was mostly used here, and at first glance assumed this post was an amusing joke by an Aussie (though - being somewhat cutesy with a Trump-ism?) rather than a Norwegian government website.

Though I might be wrong - I doubt this word comes up in tech blogs and academic conferences very often?

It can go to extremes. Like 100x by going from a dynamic to static language then 100x more adjusting the data structures/algorithm…

Once I got very confused when exactly this kind of 10k x speed up I demonstrated for a problem introduced to me as a business bottleneck then got ignored (I wasn’t on this team, though, but it’s still hard to understand the mentality and incentives there).

Apple was correct in the beginning here - the only way to make a service satisfactory to users is to have streaming aggregators (hosting all content) fed by different producers/studios. The "Spotify model", if you would.

The modern "Netfix model" has seemed to resulted in large number of new TV shows (relative to number of new movies) which always seem to get killed before their plotlines are resolved. And when I have something in mind I want to watch I often don't even know where to find it.

I'd prefer to go get DVDs from Blockbuster at this point - it'd be more convenient!

I suppose something like a skill would help those starting with the language. If nothing else, a well written skill can be read by a human - or you could load the skill and generate a program you are familiar with.

As someone working on decision AI models well before Jev, yes, I can’t see the concept disappearing or becoming less impactful/important.

Sure, extend was always the key part of “embrace, extend, and extinguish”.

The browser API has become so big because the implementers are quite tenacious, with multiple parties doing “extend” in parallel…

I definitely read that the same way Apple means it with ApplePay - they are improving your privacy with respect to the merchant you are purchasing from, by them not learning anything unnecessary about you.

Any government-approved payment system is obviously going to be auditable under existing legal frameworks.

I’d say a business requiring ever fewer humans would, due to competition and similar dynamics, become a kind of commodity business producing ever reducing margins.

Yep, the ruling seems to cover exactly that plus stopping Apple themselves treating in-app purchases and subscriptions (from actual App store apps) differently to how they currently approach physical goods (which are already allowed by Apple to go via other payment channels without any fee, or a developer might choose to go via ApplePay with possibly slightly less effort).

I agree, the end effect should be them charging say "5% of something" and the developer choosing not to bother thinking about external payments. They seem to avoiding this outcome on purpose though. (I dare say Apple might prefer to chase advertising revenue from the app store rather than accept small fees, but they'd risk trashing their own garden).

This is a good point.

I think the legal side of it is, once a precedent like this stands, the cost of going after others reduces significantly. One could argue Nintendo created a closed market for Nintendo apps and abuses their monopoly power in a completely analogous way, and ask the court for completely analogous recourse.

A _better_ solution though would be legislation around freedoms and rights for general purpose computing on any general purpose computing device, rights to repair owned hardware, rights to "own" software paid for, etc. Then you wouldn't need to rely on any monopoly-abuse arguments. The powers that be would have to give up their control (e.g. governments like the notorization thing because they can ask Apple to stop something they consider harmful from being distributed easily, but the requirement to seek notorization clashes with what I perceive as my right to run any program I want on any computer I own) and to be fair having some way to protect the (non-techie) public from malicious software is a reasonable goal (where we could argue over different solutions).

Seems like sideloaded apps will require notorisation from Apple and some form of automated security/privacy scanning (plus a system for authorising alt stores to do the same).

Payment fees for App Store apps seem to be affected in section 9(c) - I have trouble parsing it but it seem to eliminate any fee difference between purchasing physical goods through an app (currently 0% fee) and digital goods (currently 30% fee). There's a carve out for "reasonable costs" for hosting and distributing the digital goods - this could be sensible in certain scenarios (like Netflix asking Apple to a free CDN for them).

We'll see how much malicious compliance the above produces. Otherwise I'm feeling like a happy Aussie today :)

Does anyone have any links to benchmarks or stories of attempting this in practice? (Even what their internal benchmark that lead them to say "it's fast enough to compete with general-purpose databases like PostgreSQL on quite a few workloads" was?)

Yes, my favourite programming project is a tetris clone my family and one or two friends have played.

The community you share with doesn't need to be online - your croissants aren't going out for delivery but I'm sure your friends and family appreciate them (and you). It can be the same with your software products, if you choose.

I'd ask - is your hobby programming the tool, or distributing a tool you built for others to enjoy and working with that community?

I'd say the post deals _strictly_ with the former. LLM code generation absolutely changes community dynamics, what other projects are published, etc.

I thought the best metaphor Rich Hickey had was the fast food analogy - drive-thru food is not a "tool for cooking". And the externalities may be similar - fast food is convenient and fast, but may be less healthy, more expensive, and you deprive your ability to learn to cook better or faster in future. LLM use might be thought of as "fast thinking" or "fast coding" in analogy.

Yes - in multiple companies. It works perfectly fine in production, and is especially useful if mathematical optimisation is part of your app somewhere or you need to write any custom, performant algorithms. It's pretty flexible - you can use it for dynamic code like python scripts, you can write optimised data pipelines, it can function as pretty decent server (the async stuff is working well these days). Not sure on adoption, but it defintely works.

While I see your point, these people do benefit greatly from great teachers! (They’d do well in either case but do even better in the right environment).

Haha love it. Long live the fax machine!

I think free movement of people doesn’t imply Schengen Zone style borders - I thought this was mostly an economic lever that people can move and work in another country by default, not a human right where a country cannot exclude chosen people (known criminals, etc) and track all crossings. I think of it as another country giving me a visa is opt-out for that country, not opt-in.

The CE rules are relatively sane and followed commonly right? At least I see a lot of CE-stamped goods in Australia… so I always assumed that was the case.

The package system typically uses a git URL and tree hash to identify a package version. It supports private registries.

Is it really tied to github or is that just the path documented in tutorials?

Nitpick: monkeys have enough brains for "oh, I shouldn't do that again".

In raw numbers of humans... "the early majority" surely would be those that use ChatGPT or Gemini (aka Google) and pay between $0 and $20 a month?

I would be surprised if the specialist that knows that various Chinese models exist and/or that a user might choose a harness and model separately are a "majority" even of the early variety... in terms of revenue, humans, tokens, or any metric.

(Happy to be proven wrong)

I've spent my spare time in the last _days_ rewriting a pretty small document made in collaboration with Fable. It was reluctant to simplify the proposed design. And the language is so dense - it's almost poetic in nature and concision, but I wanted a clear discourse about a complex topic with people whose native language is not English.

It's an interesting model/tool. Powerful but still chock full of trade-offs. I hope the next model's language is more like e.g. OpenAI models in terms of language use. (oh, and the code comments, yikes).

I was curious to see they are advertising OLTP-like transactional processing speed. It would be super convenient to have one DB for OLTP and OLAP purposes!

Has anybody here tried using it that way?

(though I don't see any way to deal with write skew and other transactional guarantees - no SERIALIZABLE optimisitic concurrency, no SELECT FOR UPDATE pessimistic concurrency, etc)

I'm assuming most decent people just help out their friend the next day, right? The lesson appears to be prioritisation, not being a jerk.

We are very fortunate that my wife got to work at the local school. She now walks to work and I work from home about 4 days out of 5 workdays.

The benefits are huge in terms of time available and costs, and not needing childcare before or after school.

It’s hard to solve couples having two jobs in different places, but having at least one person get to work basically instantly and for zero cost is a massive boon to society.

The issue I saw is that the government department in charge of education is not the one in charge of housing and infrastructure, and there are no incentives to hire locally or make it easier to move locally. Instead you get teachers in suburb A travelling to suburb B and vice-versa, and no way that they can choose to swap roles. We lucked out.