peebeebee

You know how much overhead testing 5 different apps still is?! (iOS, MacOS, Android, Windows, Linux)

The rate at which AI is accelerating seems the same as other big inventions.

Some examples:

- Tim invented the WWW in 1989, but I'd took until around 2000 (10 years) to go to the web we now know with Streaming and Social Media.

- The first big mobile success (Nokia 3310) was in 2000, the 'end-stage' phone (iPhone 5 or something) was also 10 years later.

- Google Deepdream was in 2016, to "Will Smith eating spaghetti" in 2023, to now AI generated video literally unrecognisable from real.

I think we will be seeing some 'end-stage' AI in the next 5 years too, where the rate of improvements will sharply drop.

Robotics will probably be next? First company that can create an all purpose robot.

Tesla was THE company that started the EV-revolution, while VW was actively manipulating emission data.

I don't like Elon and his politics, but I'm very grateful for Tesla to have shaken up the car industry. Everyone is better for it.

I’m Belgium it was Dutroux. It changed everything and is a cross generational trauma that still has its effects.

Yes. There were no families before carriages… /s

A carless society/city is way more family-oriented.

problem is that when you are really in deep, you are not able to make that phone call. It is good advice to always take someone with you. It is like saying: more experienced bikers don't need to wear helmet. Yes, 99.99% of the time things will be fine, but it's still good advice to always wear one. Because that 0.01% can lead to severe consequences.

For HTML, you can probably do the following:

  <span aria-label="my text">𖢑ꚲ 𖢧𖤟𖤗𖢧</span>

I did not mind this linearity in Half-Life, because the story was engaging. You wanted to know what happened next. And I found the enemies did had some advanced "AI" for that time; like the first time the commandos came was a big moment. They were nothing like the enemies you had seen in other games.

It’s about using the right tools for the job. FAANG and developer advocates made the web needlessly complex for most people. The over-engineered tools and frameworks became the “default” way of programming for the web, loosing some strong key features that were good about it: simplicity, transparency, and speed.

Not so sure about that. You can easily write horrible code in React: Too complex, inefficient, and/or resource-intensive. If you don’t know the tools and have good theoretical programming knowledge, all code will be spaghetti code in the long run.

Think of the trees! ;)

The European Union doesn't like when a file get too big and powerful. It needs to be broken apart in order to give smaller files a chance of success.

ShadowDOM, and by extension web components are great for providing an extended set of HTML ‘native’ components. Let’s say your company has multiple frontend SPAs with different technologies (angular, react, svelte,…) they could all use the same set of company custom components, like a custom datepicker, or fancy selectbox.

With webcomponents you are pretty close to the “metal”. If you know how to write good vanilla JavaScript, you can take most of that knowledge into webcomponents. You only need to learn the custom components lifecycle, and shadowDOM, which is knowledge about web-standards. With other frameworks you need to learn template syntaxing, how state propagates, how the compiler works, etc etc. Lot of that knowledge might be obsolete in 10 years.

Which isn’t to say it can’t be worth it. Learning multiple frameworks and libraries is also very helpful to skill up because you are learning about different concepts and implementations.

I think the (interesting) output we see as the different UI is a side-effect of the actual research mentioned at the end of the video: how can we teach everyone how to do spatial programming, just like we teach everyone to write and calculate. The end result is not a finished product, but new knowledge about how to spread this new knowledge to everyone.