24 February 2024 | 11:36 am

At the risk of [insert It’s Always Sunny post room conspiracy theory meme] reading too much into things, it’s quite interesting digging into Arc’s staff and VC investors. There’s quite a Thrive Capital presence in its management team.

Link: Who makes money when AI reads the internet for us?
24 February 2024 | 9:11 am

This is good on how Arc’s push to present the web through AI hasn’t been thought through and is, of course, a profoundly bad idea.

“The best thing about the internet is that somebody super passionate about something makes a website about the thing that they love,” tech entrepreneur and blogging pioneer Anil Dash told Engadget. “This new feature from Arc intermediates that and diminishes that.” In a post on Threads shortly after Arc released the app, Dash criticized modern search engines and AI chatbots that sucked up the internet’s content and aimed to stop people from visiting websites, calling them “deeply destructive.”

You probably read Ryan Broderick’s similar sentiments.

Why even bother making new websites if no one’s going to see them? At least with the Web3 hype cycle, there were vague platitudes about ownership and financial freedom for content creators. To even entertain the idea of building AI-powered search engines means, in some sense, that you are comfortable with eventually being the reason those creators no longer exist. It is an undeniably apocalyptic project, but not just for the web as we know it, but also your own product.

Don’t say I didn’t warn you about Arc. CEO techbro gonna techbro, even when said techbro comes with impeccable liberal credentials.

Josh Miller has been on an interesting journey. From designing tech solutions to the problem of expensive nappies for the Obama administration to CEO of a VC-funded, IA-touting browser company via Thrive Capital. Thrive Capital was founded and is run by Joshua Kushner, brother of Trump’s son-in-law, Jared Kushner. It is truly a small world. No good can ever come from tech solutionists, no matter how shiny whatever product they’re pushing seems.

CSS and trying to map the whole world
19 February 2024 | 7:04 am

This is very good on how Tailwind marketed itself in order to become the de facto way of doing CSS for a lot of developers. Tailwind is, if you’re not aware, a “utility first” approach to CSS which rejects “semantic” class names such as introduction, entry-title or even card in favour of more literal transpositions of CSS properties and values. These might include classes like db, flex and mb4, which would map to these CSS declarations:

.db {
	display: block;

.flex {
	display: flex;

.mb4 {
	margin-bottom: 1.5rem;

I bought into this way of building web pages for several years because it is instinctive and quick. I never got into Tailwind – its vibe was always offputting and it seemed complicated – but I did use the Tachyons framework a lot – for a few years on the Suffolk Libraries site, and it still powers leonpaternoster.com. Its author also wrote a far more nuanced, standout introductory blog post, which you should read.

In addition to Tailwind’s complexity and its attempt to lock-in users, I see two problems with using a utility first approach.

Firstly, CSS got a lot more expressive, allowing authors to do some powerful things in a few lines of CSS which would cascade through the document. When Tachyons was first released, flex box, grid, custom properties, light-dark and clamp didn’t even exist. Often you just can’t do these things when your approach is to remove the cascade altogether and add styles as near to the bottom of the inheritance tree as possible. Or you have to start adding a load of complexity, which goes against one of the main benefits of utility first – the instinctiveness of adding a descriptively named class while knowing exactly what it will affect.

The other problem is that it’s impossible to represent all possible CSS property value pairs, especially as they become more complex. Tachyons was openly opinionated in its choice of colours and type and spacing scales, and I think it got these choices right. It was pretty easy to remember 90% of its classes without even referring to the (excellent) documentation, while producing a wide range of layouts, components and pages. However, I’d argue it encouraged a certain minimal look: it got to the stage where I could sense when a site had used Tachyons.

You can now define a consistent set of colours and sizes on a per project basis using custom properties, so you get the best of all worlds:

:root {
	--red: #AF0000; /* Whatever shade of red I want */
	--base-space: 1.5rem;

article header p {
	color: var(--red);
	margin-block-end: var(--base-space);

The problem with a hardline utility first approach is that if you also try to offer the benefits of “semantic”, cascading CSS while mapping as many declarations as possible, you end up creating a new kind of complexity in service to a principle. Tachyons works because even now I know that typing <header class="flex justify-between items-center mid-gray">…</header> will create an element that behaves like a standard website header with a logo and a nav bar. That’s easy but limiting, acceptable on a site where I don’t need, say, light-dark mode and I just want to layout a few sections on a page. It seems right that Tachyons stopped developing new features at a set of flex box classes.

More News from this Feed See Full Web Site