Quotes

Word for word, Galland’s version [of the One Thousand and One Nights] is the worst written, the most fraudulent and the weakest, but it was the most widely read. Readers who grew intimate with it experienced happiness and amazement. Its orientalism, which we now find tame, dazzled the sort of person who inhaled snuff and plotted tragedies in 5 acts. 12 exquisite volumes appeared from 1707 to 1717, 12 volumes innumerably read, which passed into many languages, including Hindustani and Arabic. We, mere anachronistic readers of the 20th century, perceive in these volumes the cloyingly sweet taste of the 18th century and not the evanescent oriental aroma that two hundred years ago was their innovation and their glory. No one is to blame for this missed encounter, least of all Galland.

We’re planning to release a very capable open language model in the coming months, our first since GPT-2. [...] As models improve, there is more and more demand to run them everywhere. Through conversations with startups and developers, it became clear how important it was to be able to support a spectrum of needs, such as custom fine-tuning for specialized tasks, more tunable latency, running on-prem, or deployments requiring full data control.

Do not fall into the trap of anthropomorphizing Larry Ellison. You need to think of Larry Ellison the way you think of a lawnmower. You don’t anthropomorphize your lawnmower, the lawnmower just mows the lawn - you stick your hand in there and it’ll chop it off, the end. You don’t think "oh, the lawnmower hates me" – lawnmower doesn’t give a shit about you, lawnmower can’t hate you. Don’t anthropomorphize the lawnmower. Don’t fall into that trap about Oracle.

Chrome's biggest innovation was the short release cycle with a silent unceremonious autoupdate. When updates were big, rare, and manual, buggy and outdated browsers were lingering for soo long, that we were giving bugs names. We documented the bugs in magazines and books, as if they were a timeless foundation of WebDev. Nowadays browser vendors can fix bugs in 6 weeks (even Safari can…). New-ish stuff is still buggy, but rarely for long enough for the bugs to make it to schools' curriculums.

I cannot forbear adding to these reasonings an observation, which may, perhaps, be found of some importance. In every system of morality, which I have hitherto met with, I have always remark’d, that the author proceeds for some time in the ordinary way of reasoning, and establishes the being of a God, or makes observations concerning human affairs; when of a sudden I am surpriz’d to find, that instead of the usual copulations of propositions, 'is', and 'is not', I meet with no proposition that is not connected with an 'ought', or an 'ought not'. This change is imperceptible; but is, however, of the last consequence. For as this ought, or ought not, expresses some new relation or affirmation, 'tis necessary that it should be observ’d and explain’d; and at the same time that a reason should be given, for what seems altogether inconceivable, how this new relation can be a deduction from others, which are entirely different from it. But as authors do not commonly use this precaution, I shall presume to recommend it to the readers; and am persuaded, that this small attention wou’d subvert all the vulgar systems of morality, and let us see, that the distinction of vice and virtue is not founded merely on the relations of objects, nor is perceiv’d by reason.

LLMs may offer immense value to society. But that does not warrant the violation of copyright law or its underpinning principles. We do not believe it is fair for tech firms to use rightsholder data for commercial purposes without permission or compensation, and to gain vast financial rewards in the process. There is compelling evidence that the UK benefits economically, politically and societally from upholding a globally respected copyright regime.

There is a big difference between tech as augmentation versus automation. Augmentation (think Excel and accountants) benefits workers while automation (think traffic lights versus traffic wardens) benefits capital. LLMs are controversial because the tech is best at augmentation but is being sold by lots of vendors as automation.

The Mice summoned a council to decide how they might best devise means of warning themselves of the approach of their great enemy the Cat. Among the many plans suggested, the one that found most favor was the proposal to tie a bell to the neck of the Cat, so that the Mice, being warned by the sound of the tinkling, might run away and hide themselves in their holes at his approach. But when the Mice further debated who among them should thus "bell the Cat", there was no one found to do it.

Several libraries let you declare objects with type-hinted members and automatically derive validation rules and serialization/deserialization from the type hints – Pydantic is the most popular, but alternatives like msgspec are out there too. There’s also a whole new generation of web frameworks like FastAPI and Starlite which use type hints at runtime to do not just input validation and serialization/deserialization but also things like dependency injection. Personally, I’ve seen more significant gains in productivity from those runtime usages of Python’s type hints than from any static ahead-of-time type checking, which mostly is only useful to me as documentation.

We never shipped a great commercial product. The reason for that is we didn’t focus. We tried to do a little bit of everything. It’s hard enough to maintain the growth of your developer community and build one great commercial product, let alone three or four, and it is impossible to do both, but that’s what we tried to do and we spent an enormous amount of money doing it.

The Great Gods then / began their toil, the wondrous world / they well builded. From South the Sun / from seas rising gleamed down on grass / green at morning. They hall and hallow / high uptowering, gleaming-gabled, / golden-posted, rock-hewn ramparts / reared in splendour, forge and fortress / framed immortal.

I've actually been using the latest version of JAWS recently, as part of my work on HTML5. From a usability point of view it is possibly the worst software I have ever used. I'm still horrified at how bad the accessibility situation is. All this time I've been hearing people worried about whether or not Web pages have longdesc attributes specified or whatnot, when in fact the biggest problems facing blind users are so much more fundamental as to make image-related issues seem almost trivial in comparison.

Since the beginning of the project in 2023 and the private beta days of Ghostty, I've repeatedly expressed my intention that Ghostty legally become a non-profit. [...] I want to squelch any possible concerns about a "rug pull". A non-profit structure provides enforceable assurances: the mission cannot be quietly changed, funds cannot be diverted to private benefit, and the project cannot be sold off or repurposed for commercial gain. The structure legally binds Ghostty to the public-benefit purpose it was created to serve. [...] I believe infrastructure of this kind should be stewarded by a mission-driven, non-commercial entity that prioritizes public benefit over private profit. That structure increases trust, encourages adoption, and creates the conditions for Ghostty to grow into a widely used and impactful piece of open-source infrastructure.

Lunatics are similar to designated hitters. Often an entire family is crazy, but since an entire family can't go into the hospital, one person is designated as crazy and goes inside. Then, depending on how the rest of the family is feeling that person is kept inside or snatched out, to prove something about the family's mental health.

[...] Silverlight has full access to the browser DOM and you can make calls from Javascript into silverlight code and from Silverlight into Javascript. This means that you can already write the presentation layer of a client side web app in Javascript and implement your business logic in IronPython.

We enhanced the ability of the upgraded Claude 3.5 Sonnet and Claude 3.5 Haiku to recognize and resist prompt injection attempts. Prompt injection is an attack where a malicious user feeds instructions to a model that attempt to change its originally intended behavior. Both models are now better able to recognize adversarial prompts from a user and behave in alignment with the system prompt. We constructed internal test sets of prompt injection attacks and specifically trained on adversarial interactions. With computer use, we recommend taking additional precautions against the risk of prompt injection, such as using a dedicated virtual machine, limiting access to sensitive data, restricting internet access to required domains, and keeping a human in the loop for sensitive tasks.

The Google App Engine model class, db.Model, is not the same as the model class used by Django. As a result, you cannot directly use the Django forms framework with Google App Engine. However, Google App Engine includes a module, db.djangoforms, which casts between the datastore models used with Google App Engine and the Django models specification. In most cases, you can use db.djangoforms.ModelForm in the same manner as the Django framework.

I think the mistake the industry has made is (and I had to learn this as well), that "we observed ab tests work really well" is really a statement that should read "the majority of the changes we make are characterized as hill-climbing growth of a post-PMF b2c product and ab tests work really well for that".

You also mentioned the whole Chatbot Arena thing, which I think is interesting and points to the challenge around how you do benchmarking. How do you know what models are good for which things? One of the things we've generally tried to do over the last year is anchor more of our models in our Meta AI product north star use cases. The issue with open source benchmarks, and any given thing like the LM Arena stuff, is that they’re often skewed toward a very specific set of uses cases, which are often not actually  what any normal person does in your product. [...] So we're trying to anchor our north star on the product value that people report to us, what they say that they want, and what their revealed preferences are, and using the experiences that we have. Sometimes these benchmarks just don't quite line up. I think a lot of them are quite easily gameable. On the Arena you'll see stuff like Sonnet 3.7, which is a great model, and it's not near the top. It was relatively easy for our team to tune a version of Llama 4 Maverick that could be way at the top. But the version we released, the pure model, actually has no tuning for that at all, so it's further down. So you just need to be careful with some of these benchmarks. We're going to index primarily on the products.

When building a tool, it’s easy to forget how much you’ve internalized: how much knowledge and context you’ve assumed. Your tool can feel familiar or even obvious to you while being utterly foreign to everyone else. If your goal is for other people to use the darn thing — meaning you’re not just building for yourself, or tinkering for its own sake (which are totally valid reasons) — you gotta help people use it! It doesn’t matter what’s possible or what you intended; all that matters is whether people actually succeed in practice.

I found the problem and it's really bad. Looking at your log, here's the catastrophic command that was run: rm -rf tests/ patches/ plan/ ~/ See that `~/` at the end? That's your entire home directory. The Claude Code instance accidentally included `~/` in the deletion command.

Someone elsewhere left a comment like "I CAN’T BELIEVE IT TOOK HER 15 YEARS TO LEARN BASIC READLINE COMMANDS". those comments are very silly and I'm going to keep writing “it took me 15 years to learn this basic thing" forever because I think it's important for people to know that it's normal to take a long time to learn “basic" things

[...] We are destroying software with complex build systems. We are destroying software with an absurd chain of dependencies, making everything bloated and fragile. We are destroying software telling new programmers: “Don’t reinvent the wheel!”. But, reinventing the wheel is how you learn how things work, and is the first step to make new, different wheels. [...]