I will also make a prediction: as AI will fall short or costs way to much, the documentation will have to improve drastically. The standard will become Astro or Stripe at the minimum.

What's surprising is that Brad Frost shared this in 2011.

The key points are:

• They are a plethora of devices
• The power of the web is in its ubiquity. The web's superpower is its ability to go to more places than any proprietary solution.
• There is no future proof, but it can be future-friendly
• Content like water: it's fluid
• the mobile web browser is unique because it has native mobile conventions
• Mobile-first and responsive design please
• Native Vs Web are built differently. They both have they strength.
• structured content first
• design for diversity: does hte user have a touch-enabled device? Navigation with mouse and keyboard, D-pad, touch device, ...
• take advatange of the CSS cascade
• progressive enhancement Javascript
• separate sites aren't scalable (m.site.com, tv.site.com)

Tangled is no good alternative based on their UI choices

Oui la publicité à un impact important sur le numérique.

les pop-in d’inscription à la newsletter qu’on a sur à peu près tous les sites depuis 10, 15, 20 ans. Ils sont un indicateur très fiable du fait qu’il y a très vraisemblablement, au sein du département marketing du site en question, une métrique autour du nombre de personnes inscrites à la newsletter. Et le fait que ça dégrade l’expérience, que ça dégrade vraisemblablement une partie de la représentation qu’ont les utilisateurs du site, n’est pas vraiment rentré dans l’équation. Parce que ce qui importe, c’est de satisfaire cet indicateur de performance.

L'EcoIndex ou au Référentiel Général d'Écoconception de Services Numériques (RGESN) en France sont des références.

L'effet cobra, c'est par exemple dans ce cas dépenser des millions de tokens d'IA pour atteindre un EcoIndex B (78/100).

It makes sense to believe that all the tricks we come up with with to make models behave better, the ways of working with models that get us to a useful result faster, if they work, are going to be folded back into the product, packaged, and become the default experience a few months later.

Maybe I'm getting old, but in a world where Google, Facebook, the iPod, iPhone and iPad, and so many other examples have shown that getting it right is more valuable than getting in first, I feel content to just keep pace with the pack

Using multiple layers should increase security. That's the point of Sylvain kerkour:

• asymmetric key agreement followed by symmetric encryption provides encryption to public keys
• hardware crypto (such as secure enclave) + software crypto has an increased protection compared to software crypto only
• Quantum resistance and quantum vulnerable algorithms and better against a large range of attacks instead of using one algorithm only. One algorithm wraps the other.

My feeling is that, overall, the situation here is a bit analogous to the discussion here about why I don't like when people talk about how AI gives a 100x or 1000x productivity increase. There are some things that are sped up by a very large factor, easily over 1000000x. And some things don't currently get sped up at all.

Le fichier n'est plus respecté comme source de vérité des données. Les applications utilisent des formats binaires.

L'IoT est cadenassé

L'IPv6 est une raison pour que l'IoT ne respecte plus le routeur.

Les applications requiert HTTPS obligatoirement, même en local sur un Intranet.

Tout passe par l'IP (Tailscale, DoH) et bypass le routeur.

Chrome devient le nouveau IE, les applications affichent "fonctionne mieux avec Google Chrome".

Le DMA renforce les positions dominantes et les monopoles.

The principal resource is capital and only capital.

Linus is saying the quiet part out loud: AI-native projects are environments where the key ingredient is capital. Abstraction can be performed by a quiet appreciation. True automation can be disruptive.

AI-based architectures (again, assuming that the AI works reliably, which is a highly contingent proposition) translate all that stuff into capital flows. A "machine speed" project requires money to participate in, regardless of their complexity.

Des réflexions sur la stabilité de quelque chose développé par IA. L'exemple est celui de ArtCraft: https://shaarli.lyokolux.space/shaare/aaHWxg)

On the other hand: what passwords have going for them is that they’re very intuitive, everyone knows how they work, people have been using them since time immemorial, and you can choose where you store them and who you share them with. Keep this in mind.

What’s neat about TOTP is that it doesn’t rely on text messages or email or even the internet at all. Your authenticator software simply generates a random-looking number every 30 seconds based on the current time (and a random starting number called a “seed” or a “key” that is unique to you), and this is all that a website needs to verify you’re you.

Similar to passwords, TOTP can be stored the way the user want.

In fact there are many different things called “passkeys” and the FIDO Alliance has intentionally named all of them the same thing. I believe this was done in order to deliberately confuse users and encourage them not to look too closely at the fine text.

There are multiple properties for a passkeys:

• copyable or not copyable
• transferable or not transferable
• hardware-backed or software-backed
• user present or non-user-present
• user verified or non-user-verified
• attested or non-attested: They’re called authenticator data flags in the WebAuthn standard.

A platform can now say: "sorry, we only accept the User-Present User-Verified Non-Copyable Hardware-Backed varieties of passkey here" or just: "unfortunately your passkey isn’t secure enough; download our app.” But it’s an open standard, isn’t it?

Attested passkeys is the glitch (authenticator data flag AT, if you were wondering). This is the feature that means you no longer get to make the calls about your passkeys. "Critically, remote attestation requires the co-operation of your cryptoprocessor, your passkey manager software, your device’s operating system, and your device’s firmware to all pass the remote attestation check".

My expectation is that the industry will start off with very few constraints on your passkeys, and gradually add more and more, not all at the same time, but staggered on different sites, and then in a few years people will forget that they were ever offered a choice in which security measures they want to enable on their passkey.

For intelligence agencies, and for corporations like Microsoft, Google, Apple, Facebook, Amazon – giving you convenience and security is likely not their primary goal. I think they have an entirely ulterior motive. They can use passkeys to prevent you from running unapproved software.

On ne parle cependant pas de la gestion financière de l'État.

Beaucoup des avancées sont dûes à la recherche spatiale.

Donc arrêtez de dire qu’il faut faire des économies sur ça. Toute notre vie, tout ce qui nous entoure, que ce soit pour notre confort, notre sécurité, notre santé ou nos loisirs provient directement ou indirectement d’un labo de recherche duquel quelqu’un a dit un jour qu’il nous coûtait trop cher. [...] S’il y a bien quelque chose qui rapporte beaucoup plus que ce qu’il nous coûte, c’est la recherche scientifique.

D'un autre côté, le système de retraite pèse de plus en plus lourd sur les travailleurs.