markjgsmith

My techno-optimism - It's an interesting read, but bloody hell it's a slog to get through it all. You have to have galaxy brain turned on for like an hour and if the world doesn't co-operate and you have to move location several times, as I did, battle through bouts of tiredness, and the boring parts, don't expect to retain much. The ending is quite up lifting which is nice, but on the whole I'm more worried about our tech future than before embarking on the ride. None of this feels like freedom to me. Rather it feels like never ending war, with impossible odds. It was already bloody difficult before AI and crypto. I wish I could be more optimistic, but that's how I feel about it today. vitalik.eth.limo #

An attempted taxonomy of web components - Zach Leatherman summarizes his usage of web components, with lots of examples. This space still feels quite unsettled though recently I've noticed many have taken interest and are experimenting. Perhaps web developers are getting into their web components groove. zachleat.com #

Shein files for U.S. IPO, as fast-fashion giant looks to expand its global reach - I hear Shein mentionned so much this has got to be quite a big deal. Perhaps they'll launch a companion company called Himout. Marketing is a lot more direct in asia. I'm suprised at how much westerners appear to like their approach. Maybe they just don't see it? cnbc.com #

🚀 Latest Newsletter: Passing the Baton of Chaos (Issue #142) markjgsmith.com #

What’s New in CSS Land (JS Party Podcast) - The latest episode has Una Kravets on from the CSS Podcast to talk about the latest CSS features. So many great new features. If I were starting a new project I'd be exploring sass style nesting, cascade layers, popover and select list components, anchor positioning, scroll driven animations, view transitions, the :has selector and responsive design with container queries. There's also some general advice about separating style from logic. Great episode. changelog.com #

Git branches: intuition & reality - There's definitely confusion in this area, but for me it isn't around the concept of a branch. I'm happy with my mental model of a branch. However I've always been confused by the concept of a rebase, while I'm happy with the idea of a merge. Though the article was looking at it from a different angle than I would have liked, the diagrams midway through that compares a simple rebase with a simple merge make a lot of sense. The rebase basically squeezes in all the latest commits from main, shoving them below the common ancestor, thus re-creating the base that the feature branch is built on. Presumably this results in a branch that has a more full history, whereas a branch updated via merge has loads of code appearing all in one go at the merge point. The article doesn't go into that, which is a bit of a shame. I'm going to continue favoring merges, for now I'm still not convinced that rebases are worth the trouble. jvns.ca #

HTML Web Components - Chris Ferdinandi comments on Jeremy Keith's HTML Web Components piece from earlier in the week. He's got a great example of how they can be used to enhance existing HTML, massively reducing the amount of code needed in a recent Nasa project he's been working on. Pretty cool. gomakethings.com #

Welcome to the Next Era of Loom - I haven't used this product but the feature descriptions sound awesone. It's an AI that watches your videos and automatically does tedious things you normally have to do yourself like creating a title and description, but it can also summarize discussion points, aggregate tasks and even create chapters for easy navigation. It's really impressive, I could imagine that these sorts of tools will make working with video really attractive for teams. I'd love to have some of these features in my static site generator. loom.com #

Sam Altman to return as CEO of OpenAI - It would be interesting to compare the complexity of these big tech events over time. Similar to how in code you can measure cyclic complexity, it seems the blow ups are ever more complex and over shorter and shorter timescales. Eventually...I'm not sure, do all our heads explode? Or perhaps a massively distributed never ending race condition, like a star-trek-phazers-on-stun-appocalypse? theverge.com #

Everything is a file - "Everything is a file" describes one of the defining features of Unix, and its derivatives--that a wide range of input/output resources such as documents, directories, hard-drives, modems, keyboards, printers and even some inter-process and network communications are simple streams of bytes exposed through the filesystem name space. en.wikipedia.org #

Getting Started with Mistral-7b-Instruct-v0.1 - I've been hearing Mistral mentioned a lot on various podcasts recently. It's totally open source LLM, all the pieces run on your laptop using WASM. Has a chat client to interact with it and an API server so you can access it over HTTP. Pretty cool. secondstate.io #

The new Picadilly Line trains have air-conditioning - Anyone that has spent time living in London will have been on the Picadilly Line trains as they pass right through the centre of the city. They are some of the oldest lines on the network and are the deepest underground. They're also very compact, which makes them unique stylistically, but horrid in the summer rush hour. It's not uncommon for people to temporarily mentally turn into literal sardines. As you can imagine then, the news that these new trains will have aircon, is pretty big news. I used to commute in central London on the Picadilly Line everyday, first from West London and then a few years later, from East London. Turns out most trains throughout europe are tested in Germany, and they just had a staff and press day outing and wrote it up. The new design has a modern art deco vibe going on. Mind the gap mates. ianvisits.co.uk #

unjs/h3: Minimal h(ttp) framework built for high performance and portability ⚡️ - I thought this library looked pretty cool. Fully supports promises aswell as Express middleware. github.com #

Bootstrapping Self Awareness In GPT-4: Towards Implementing Recursive Self Inquiry - Even without much knowledge of AI or computing it's easy to understand the methodology used here. Basically self awareness by repeatedly feeding it's output back in as input. Interesting though self awareness alone is not general intelligence. thewaltersfile.substack.com #

Microsoft hires ex-OpenAI leaders Altman and Brockman to lead new AI group - Quite a topsy turvy start to the week in tech, my week is similarly stop / start for some odd reason. techcrunch.com #

Police and thieves - Classic reggae track from Junior Murvin. m.youtube.com #

Tether Reveals Bitcoin Mining Expansion With $500 Million Investment - They will be opening bitcoin mining farms in Uruguay, Paraguay, and El Salvador. I guess since we are at the bottom of the market, it's probably the time when mining operations start to ramp up, with expectations of a bull market. There was recently the first city hall to experiment with bitcoin mining, with the mayor of Fort Worth in the US installing nodes, aiming to become a big commodities hub. Oil, beef, natural gas and now bitcoin. If more institutions start this type of experimentation, the next phase of bitcoin adoption could be very interesting. cryptobriefing.com #

67 Weird Debugging Tricks Your Browser Doesn't Want You to Know - Not sure I agree with the title, there might very well be forces out there that don't want you to be able to debug, but I don't see why the browser would necesarily be that entity, though it could be. In any case, there are some really useful debugging tips and tricks in this article. alan.norbauer.com #

New Post: All-in Dirty Sanchez - Why don’t we have a podcast of similar caliber to the All-in Podcast coming out of the UK? If we want to compete in tech, and maybe politics, we are going to need one of those. Otherwise even the french will have one. markjgsmith.com #

🚀 Latest Newsletter: Surrounded by Bears (Issue #141) markjgsmith.com #

Some kind of purge happening over at OpenAI - The company that arguably kicked off the recent revolution in AI tech appears to be going through some kind of purge. Sam Altman looks to have been pushed out, and a bunch of others including founders are quiting. It's always weird when this happens, but it's very common for this type of thing to happen in companies that are growing quickly. I've had it happen at two previous companies I've worked for. It will be super interesting to hear their side of the story. theverge.com #

How to use NPM packages outside of Node - Some useful techniques that sort of remind me of symlinking on Linux filesystems, but you use package.json with local file:path style dependency paths that point to shims which might re-implement some parts of the package, or completely stub out some bits that aren't necessary. Quite technical but good to understand how this works, could be very handy in serverless / edge function environnents. neon.tech #

YouTube Shorts to introduce generative AI tools - Users will be able to create music for their videos using the style of various famous artists. It will be interesting to see how many artists get onboard with this trend. It's Napster all over again, but maybe this time artists will be able to ride the wave too. siliconangle.com #

What I learned building an audio plugins system for the web - I built a plugin system for my static site generator. In some ways it's quite simplistic, but it's very flexible. In the future I imagine I'll need to modernize it in some ways, but for now it's good enough. That's why I'm always interested to read about plugin system implementations. It's amazing how different they tend to be depending on the runtime environment. This audio plugin system has almost no similarities to that of my SSG, but it's interesting to understand the constrainsts and requirements, if only to more fully appreciate the variety of patterns out there. Also once you know several implementations, similar to with spoken languages, you gain new perspectives on your environment. blog.benwiley.org #

X rival Bluesky hits 2M users, says federation coming ‘early next year’ - This is good news. I think social media is a place where having a handful of different protocols will be beneficial. Human interaction is so varied in the real world, it's unreasonable to expect there to be one protocol. It would be great if Bluesky also supported RSS, but maybe they want to force people to always be using their service rather than people flowing their content from their personal site through their service. I think that's shortsighted. If I could more easily flow content from my site through their service, I'd have more time to actually hop on their service, interact and find more followers than Satan, who is still my only follower. He doesn't even answer my questions. techcrunch.com #

Microsoft is finally making custom chips -- and they’re all about AI - There are two chips. The Azure Maia 100 competes with Nvidia's very popular H100 GPU, primarily used for AI. They have been developing them with input from OpenAI, optimising everything down to the silicon. The other called Cobalt 100 is an ARM based CPU aimed at data centre cloud workloads. It has the ability to "control performance and power consumption per core and on every single virtual machine". They are 40% faster than current chips in their data centres. theverge.com #

git rebase: what can go wrong? - I used to use rebase -i a lot more than I do now. I tended to only use it on local feature branches before pushing to remote. It's very useful for combining lots of small commits into bigger commits that make more logical sense. I think the UI makes quite a lot of difference. In vim it's very straight forward most of the time, not so much in my current editor. Anyway the article gets very technical but worth a quick read to get a better sense for the pitfalls, and for which use cases it's safe. jvns.ca #

How to Build a Server-Side React App Using Vite and Express - It's essentially a minimal example that demonstrates data fetching and server side rendering (SSR) with a browser client that hydrates the app with a copy of the fetched data that is stored in a