markjgsmith

2023/06/30 #

• Building digital asset economies

  Dominic Frisby had an interesting episode on his Flying Frisby Podcast this week. It was an interview with Microstrategy CEO Michael Saylor. Saylor has made a name for himself initially by successfully running a publically traded software company for over 20 years, but more recently for his rather adventurous corporate investments in Bitcoin, something made possible by focussing on fair value rather than GAP accouting.

  Microstrategy holds some of the largest amounts of Bitcoin worldwide, and has quadrupled in value by taking Bitcoin onto it's balance sheet despite the volatility. Most organisations haven't been willing to do that due to gap accounting practices which don't differentiate between gains due to operation and those due to balance sheet changes, for instance appreciation in commodity assets. Gap accounting companies run the risk of big volatility swings in Bitcoin's value eclipsing otherwise normal operations, making a company's performance look much worse than it appears.

  Saylor is a Bitcoin promoter, given his holdings that much is clear, but the narratives he pushes are very well thought out, always attempting to be rooted in fundamentals. One of the narratives he has been putting forward recently, and which seems to be catching on, is that of building digital asset economies. He believes this is something that countries around the world are going to be getting into in the near term. And with the recent announcement of many large TradFi companies (Blackrock, Fidelity, Citadel, Schwab and Deutchebank) creating various flavours of Bitcoin ETFs, it does indeed seem to be trending in that direction. Multinational corporations need to park their money somewhere, and given the prospects of the US bond market, that place might very well be Bitcoin.

  In the interview he notes how it’s important to distinguish between the different digital asset classes. Regulation takes time to evolve and people are only starting to understand the difference between these. Namely the difference between:

  • Digital commodities
  • Digital securities
  • Digital currencies
  • Digital tokens

  I've transposed his definitions here, as I think they are useful, even if they haven't been fully endorsed by regulators yet, because they paint a very clear picture of the future financial environment that crypto and bitcoin proponents envisage.

  Digital commodities

  • Assets without an issuer
  • Like Bitcoin

  Digital security

  • Assets with an issuer, a controlling body creates the currency and they control it
  • Like Solana
  • Like Algorand

  Digital currency

  • Stable coins, they proport to be the US dollar as a bearer instrument and a digital asset, a blockchain representation of dollars
  • Like Circle
  • Like Tether

  Digital tokens

  • Like an NFT, piece of art 1 of 1
  • Like a Tom Brady coin, 10000 super fans coins that he issues, holders get to for example come to his house for a BBQ every month

  Saylor then outlines the growing importance of distinguishing these:

  A digital exchange is something that can trade all these things. To create a digital asset industry in your country's economy you have to have a universally accepted taxonomy of digital assets, a digital asset framework and political consensus that it’s ok to issue, trade and custody those assets.

  We don’t currently have the political consensus for many things. There is a big divide. Saylor gives several very specific examples that help to illustrate the sea change that decentralised finance is bringing to the table.

  • Moving 1 million dollars from A to B without KYC would be considered money laundering in TradFi
  • Trading a security 24/7
  • Self custody a million dollars of digital assets in stable coin from an android phone
  • Self custody a million dollars of Apple stock on an android phone in India
  • Trade with a counter party that you don’t know via smart contract
  • Buy a perpetual swap derivative on a Saturday night that represents a registered public company
  • A person living in India that has 10 million dollars of Apple stock, borrowing against that from any counter party in the world on a Saturday afternoon, getting the best bid

  All these things and much more are totally possible with DeFi. It's going to be interesting to watch how this evolves.

Today’s links:

• Decentralized Twitter rival Damus will be removed from the App Store over Bitcoin tipping - That does seem like a bit of an over-reaction. Surely they aren't the first iOS app to add a tipping feature on a social media app?appleinsider.com #

• Freefaces - Open source typefaces gallery - There are some pretty neat looking fonts in this collection. www.freefaces.gallery #

• Let's welcome Meta, not block them - The author Manton Reece, founder of micro.blog, writes about the movement to pre-emptively block Meta before it has even released it's product, mentioning Open APIs, data privacy, getting your own house in order, and monocultures. Micro.blog by the way supports posting to ActivityPub, Nostr and Bluesky. Pretty cool. www.manton.org #

• Hugging Face CEO tells US House open-source AI is ‘extremely aligned’ with American interests - I'm still amazed at how easy and cool it is that anyone in the world can watch these type of government hearings. It wasn't always like this. Isn't the world wide web awesome? Btw Hugging Face is like Github but for AI and large language models (LLMs). Open Source AI is a very interesting topic that we are only starting to grapple with. We are still in the new-ideas-permeating-everything phase which is quite confusing, but some academics trained in thinking at a macro level are starting to ask very detailed & specific questions which are highlighting the magnitude of the sea change that is ahead. venturebeat.com #

• Architecture of an early stage SAAS - This is a great writeup. It’s a somewhat more modern version of the nodejs architecure I used for my SaaS. The author also covers the frontend setup abd code organisation. It relies on Fly.io but you do get a lot of nice features out of the box. It is quite involved, but building and running a SaaS is complicated and difficult. www.feelback.dev #

• 🚀 New Post: Building digital asset economies - Distinguishing between the different digital asset classes that DeFi is making available, and looking at some specific examples that demonstrate the new reality that is fast approaching. We are entering a world where new digital asset economies are being forged. markjgsmith.com #

2023/06/29 #

• Revisiting Instapaper for offline reading

  Following on from Monday's piece about the issues reading web development content offline with Safari Reading Lists and Pocket, I wondered again about Instapaper. I had previously tried it but gave up after getting frustrated with the experience. There was a lot going on last time I tried it. I was having to jump around between apps, and was also in a bit of a rush. It's easy to get lost in a sea of features sometimes. Was the overall app experience really as bad as I remembered?

  I re-opened the app to have another look. The default screen on opening the app is the saved article list. The first thing I noticed was that it just sort of looks a bit classier than Pocket's listing page. The title font and the layout of each item results in a vaguely newspaperish vibe. The content just seems to fit the available space a bit better. There are both titles and descriptions and less white space around the thumbnails. The article titles all seemed to be correct too. When an item does get saved, it looks kind of nice. I decided to give Instapaper another try.

  Let’s look at the same screenshots:

  

  

  

  As with the Pocket, the article saved in Instapaper has most of the page styling removed. It looks much the same as Pocket’s version. The site’s style is lost but the page is still readable. I guess some folks like the consistency. Personally I prefer being able to see the original website styling.

  

  

  

  As with Pocket, the article saved in Instapaper is missing images. In fact it’s a bit worse because literally all the images are missing. It’s already game over as far as learning CSS. There’s not much point continuing to read the article. But let’s continue for the sake of comparison.

  

  

  

  Code snippets are looking better than in Pocket. At least the newlines haven’t been removed, but syntax highlighting has, and that has exposed the style hint meant only to inform the syntax highlighter what language is being used (CSS in this case). Readable but looks kind of ugly. An entire article full of these would be a real slog to get through. Syntax highlighting really makes a big difference when trying to comprehend code.

  

  

  

  The listing page for Instapaper looks pretty good, though notice it’s still “downloading”. That never completed by the way, nice though it is to have a progress indicator. I don’t have a screenshot of my Bitcoin article, but the title did get correctly saved. So it seems Instapaper has the better webpage parser, or at least it works better with my website.

  

  

  

  The article page for Instapaper is very similar to that of Pocket’s, though I think I prefer Pocket’s version, less wasted space. But as previously described, Pocket gets the title wrong, at least for my website.

  The overall Instapaper experience is better than I remember but the following issues are still there:

  • Images still missing much of the time
  • Code snippets are there and readable, but look weird, no syntax highlighting
  • You can’t copy and paste article titles on an article page, that’s really annoying
  • When images do get saved, there is no way to zoom, so anything with detail is unreadable

  This time I figured out how to ensure the Instapaper icon always appeared in the Share Sheet. That’s the box with the arrow emerging from it that you tap to share things throughout iOS. I think of it as the share button, no idea why they call it Share Sheet. Previously it was there sometimes, other times it wasn’t. I installed the bookmarklet too, very difficult installation procedure, and I found it didn’t always work. There was a 3rd way to post via email, which might be useful but it wanted access to my contacts and I’ve had bad experiences with that.

  The thing that struck me was how confusing setup was. That’s what got me last time. It never felt like I got to a safe place during setup. Even after completing setup it still felt like there were two many ways to post and when it didn’t work I wasn’t sure if it was me, or the setup, or iOS config. Eventually I gave up.

  I’ve been using Instapaper for a few days now, and for non-web development content it’s actually quite decent. The item download indicator is the biggest improvement over Pocket. The classier look is a big draw too. It’s nice when apps have the right vibe.

  Overall though for reading web development content while offline, none of these apps are adequate. If you are relying on them to learn web development you will be at a profound disadvantage compared to being online. That’s currently a certainty.

• Ideal offline reading app for web development content

  I've spent the early part of the week exploring various Read Later apps for reading web development content offline. First I compared Safari Reading Lists, Pocket and then I added a comparison of Instapaper. These are the most popular. What I found was that none of them were adequate. It's important because it's one of those things you don't realise you might need until you do, at which point if it doesn't work, well it's too late.

  The main reason the current apps all fail is their handling of images and code snippets. A huge part of web development content is diagrams, screenshots and of course code and configuration examples. Without these it's very difficult to understand what the author is attempting to communicate. Most of the current apps remove page styling in order to save on space and apparently to create a consistent reading experience.

  That's all well and good but by removing the styling they also remove the code syntax highlighting and occasionally the formatting, including newlines. Without newlines the entire code snippet is squashed onto 1 really long line. That makes it really really difficult, nay impossible, to comprehend web development articles with any kind of complexity.

  The code snippets problem is bad enough to make things really terrible, but on top of that, very often images from the saved articles don't get saved. Instead either empty rectangles where the images used to be, or indeed, nothing at all. You might not even know that something important is missing. Both these bugs together make reading web development content offline effectively impossible.

  I put together an opinionated list of features I think is necessary for a Read Later app in order for it to be useful for reading web development content.

  Basics

  • Setup should be easy, install the app and you are ready to start saving articles
  • It should be totally obvious how to save a page from Safari, no wondering if you have to use a bookmarklet or the "share sheet", whatever that is
  • No need to configure anything in iOS settings
  • Saving always at least adds a new item, or displays an error somewhere if there was an issue, a viewable activity stream for all modifying operations performed would be useful
  • All images in a page get saved no matter what
  • Code snippets and config file examples remain untouched, with syntax highlighting, newlines etc
  • App displays when it is downloading / synching items
  • App downloads and syncs in the background so you never have to check that it synched before you go offline, no surprises when you get offline, the articles you saved should be there ready for you to read
  • Clear docs with guidelines / best practices for website devs so they know how to structure their pages in order to have their pages save correctly
  • App doesn’t randomly refresh pages, once the page is downloaded / synched, you can switch between pages quickly and easily without the page going blank or loosing content
  • App doesn’t insist on restarting every time you bring it to the foreground
  • Title, description and website address are always saved and displayed correctly on saved articles listing screen
  • All text of an article can be copy pasted, including author, date/time, title, images and code snippets
  • Images can be zoomed
  • Listing page has a thumbnail next to item

  Nice to haves

  • The page looks identical to the page you saved, don’t remove styling
  • In the background save copies of pages linked to by the saved page, and be able to read these pages offline when you click on associated link in the saved page
  • Videos get saved for offline viewing
  • Widget to easily see the list of items being downloaded / synching
  • Automatically create an archive of the page using an archive service like archive.md or archive.org way back machine

  I think that a Read Later app with these features would be good enough for reading web development articles offline. Reading articles offline is not only extremely convenient in a busy world but it could even prove vital should you experience internet connectivity issues. Hopefully that's something that many of us have learnt, having now got through the many pandemic lock downs of the past couple of years. App builders please consider making your apps ready for these types of scenarios.

Today’s links:

• The top six rivals competing with OpenAI - Looks like quite a decent list of some of the big players in the wider AI space: OpenAI, Google DeepMind, Anthropic, Cohere, Stability A.I., EleutherAI, Hugging Face aisupremacy.substack.com #

• Six more companies competing with OpenAI - Adept AI, Inflection AI, Runway, Alph Alpha, AI21 Labs, BAAI - The magic film visual effects tools company sounds very cool, uses generative AI to add all sorts of effects from text prompts. There also appears to be a trend in training models in non-english languages, which of course makes sense. I'd also like to try out the AI gramerly tool. Which one of these companies is going to be the next google, microsoft, apple or meta? aisupremacy.substack.com #

• 🚀 New Post: Revisiting Instapaper for offline reading - A continuation of the exploration of Read Later apps that I did earlier in the week. This time I decided to try Instapaper again. The aim here is to have a way to read web development content offline. So far the results from using Safari Reading Lists and Pocket have been very variable. How about Instapaper, is it as bad as I remember? markjgsmith.com #

• The Mac Pro’s biggest problem is the MacBook - I started out reading this article thinking I wasn't that interested in the topic, but as I read, memories of jobs past, and various digital collaboration environment infras I helped setup, flooded my thoughts. Turns out I actually have quite a lot to say about high end mac hardware. Most of the VFX shops I worked for used Windows and Linux, because of upgradeability/modularity but also because it was easier to integrate them into the workflows. The folks running macs tended to be one offs, on their own island so to speak, with their own custom setup. From a sysadmin perspective, those systems were harder to maintain. At one place all the engineers actually got Mac Pros because then we could have any OS we wanted by running Parallels, which made sense given we were supporting artists running all 3 OSs. I loved my silver tower, it was freaking awesome! But probably way more power than I actually needed at the time. Oh how things change. Dear lord how I wish I had enough power to not have to wait 5 minutes switching feature branches or pushing to remote. Life is waiting for progress bar atm. www.theverge.com #

• 🚀 New Post: Ideal offline reading app for web development content - Reading web development content offline has some specific requirements over and above what is generally needed to read other types of content. I detail the ideal required basic and advanced feature set. markjgsmith.com #

2023/06/28 #

Today’s links:

• Amazon to start using bodegas and small businesses to deliver packages - I have seen something similar where I am in Vietnam. Some shops have a large metal cabinet into which staff place small packages customers give them. It appears to be for sending rather than receiving. Anyhow, such partnerships with small businesses could prove to be very valuable in a future where physical increasingly integrates with digital. Amazon famously has highstreet stores called Amazon Go, that automatically tally up the food you put in your trolley as you shop and charges you as you walk out. The new packages delivery program has a cool name too, Amazon Hub. Wouldn't it be interesting if Amazon opened a major crypto exchange and turned high street shop partners into crypto exchanges? That was an idea I put forward in my recent ultimately very pro-bitcoin article that in retrospect was sort of maskerading as a hit piece. It's strange how creativity works sometimes.www.axios.com #

• FTC sues Amazon over deceptive signup and cancelation process - I don't know that much about the flow to have an opinion. Ben Thompson has a piece that goes into a lot more details with screenshots. Seems like there might be too many steps. Reguardless though I'm happy that government insitutions are putting dark pattern usage on the agenda. I think it's something we really need to be aware of, especially with the rise of AI, it might get more difficult to detect. It's an area in which we probably want to avoid an arms race, that would likely be very bad for everyone. www.cnbc.com #

2023/06/27 #

Today’s links:

• Yesterday's issue loading images on my blog post about the state of offline reading for web development content appears to have been resolved, so I'm linking to the article again, it will make much more sense now with the images. I tried reading Chris' CSS article again last night, noticed that many of the images only load as you scroll down the page. I wonder if that might be what's preventing them from making it into the saved offline version. Anyway, good morning everyone, time for coffee :) markjgsmith.com #

• Leveraged Bitcoin Futures ETF to Start Trading Tuesday - Good news though a few things stand out, no idea how relevant. It's a futures based ETF rather that a spot ETF, which the Blackrock application was for. Rather than say it was explicitly approved, says only that it wasn't blocked. www.coindesk.com #

• IMF says banning crypto assets may not be effective long-term - The institution which is likely one of the most important financial institution in the world, appears to be softening their stance on crypto after being a long time critic. This made me smile a bit, comes just 3 days after my blog post about crypto not being enough. Ain't that always the way [rolls eyes]. www.theblock.co #

• JPMorgan Starts Euro Blockchain Payments for Corporates - Corporate use of crypto is a big theme currently, so it's noteworthy when one of the world's biggest banks introduces crypto rails, making it possible for large corporates to transfer large amounts dollars or euros outside of business hours. Crypto runs 24/7/365. It will be interesting to follow the popularity, currently it's used for $300 billion out of total possible $10 trillion of payments. So called "digital asset" projects appear to be the new black. Michael Saylor refers to this as the transition from physical to digital property. www.bloomberg.com #

• ‘ZK embodies integrity, privacy and magic’: Matter Labs - I've so far mostly concentrated on bitcoin when it comes to learning about crypto and DeFi. It's interesting to hear alt coin folks exploring internet fundamentals, being open to hard lessons learnt by previous generations of tech innovators, especially around soveraignty and gate-keepers. It reminds me a bit of the explosion of open source and creative commons licensing during web2.0. Diversity is great, but it gets kind of confusing. I hope things settle down to the point where we have a handful of well understood contenders, the equivalent of MIT, GPL, Apache etc. blockworks.co #

2023/06/26 #

• The state of offline reading for web development content

  I've been reading quite a few CSS related articles recently, several really interesting ones appeared in both my RSS reader and some of the newsletters that I subscribe to. I tend to skim through these when I'm online saving them to read later when I have more time available. I currently use a combination of Safari Reading Lists and the Pocket iOS app, but I have also tried Instapaper. Overall the experience has been really lousy :(

  When I find time to read the articles later I find 80% of the time one of the following has happened:

  • The article just didn’t get saved
  • The saved article is missing some or all the images
  • The saved article has squashed all the code snippets down to a single line
  • The saved article lists with an irrelevant title

  It’s slightly different in each app. For instance Reading Lists is pretty good with the titles, Pocket often gets it wrong. All the apps have the missing images issue, same with the squashed code problem. I usually also leave the page open in Safari but 9 times out of 10 when I bring the page back into focus Safari insists on trying to reload the page, which of course fails because I’m offline.

  I was trying to read an article by Chris Coyier earlier about Modern CSS which looks to have quite a few things that might be very useful for making my website look a bit more styled. But once again when I sat down later to read the saved article, I got blocked.

  I’ve taken some screenshots to illustrate the problem:

  

  

  It starts out pretty good with both Reading Lists and Pocket displaying something that is faithful to the original page. Though notice that Pocket has decided to save it in Article mode, which is essentially a stripped down version of the page, with most styling removed. That apparently saves storage space, but it also makes all pages you save look the same.

  That’s not a show stopper much of the time, but kind of ruins the vibe. It’s way nicer to see the actual page you were looking at. And in fact sometimes it is a show stopper, especially for web development articles, because authors tend to often reference their own site when describing various techniques they have used - “Check out the header for an example of that”.

  

  

  Things deteriorate quickly though. We get to one of the first major sections, one about Logical Properties & Layout in Reading Lists, and all the images have been disappeared. That's a massive show stopper when you are reading an article about CSS. It’s basically impossible to imagine what a web page looks like just from the textual description. You might get a rough idea, but basically it’s already game over at this point in Reading Lists.

  Switch over to Pocket and it’s a little better, but not by much. The section image has appeared in the right place, but the image of the website that Chris is commenting on has gone AWOL. So it’s basically game over in Pocket at this stage too. There’s some difference between the two apps, so maybe we can just switch between them? I doubt it, but I really would like to learn some CSS. Let’s continue.

  Update: As I write this article, looking back at the Reading Lists version, it appears as though some of the images have loaded, for example the website Chris was commenting on, but many are still missing in the rest of the page. It’s a perfect illustration of the frustrating offline experience, it somehow takes longer to load the images offline than online!

  

  

  Code snippets are looking good in Reading Lists (this time - it’s not always consistent). However in Pocket they’ve been squished down to 1 really long horizontally scrollable line with no syntax highlighting. That’s basically impossible to read and comprehend for anything with moderate complexity.

  

  

  How about the listing pages in these apps? Reading Lists displays Chris’ article well, but notice my recent bitcoin blog post just below it. It says “Not available for offline reading”. That happens on a huge number of sites, like above 50% of the sites I try to save. Again Chris’ site looks good in Pocket but my bitcoin blog post just has my website header as the title. That’s the same for any page on my site, which is obviously a show stopper.

  

  

  As for article pages, you can see above the 'Top of page' screenshots for Chris’ site. Both look good. But my bitcoin article on Reading Lists results in possibly the most annoying popup in offline reading. That's because it happens so darn often, and always results in a sinking feeling of disappointment. Oh no, not again :(

  I could continue with more examples, it’s essentially the same throughout the article. Between the apps there is no way to see screenshots or code snippets. You might get lucky switching between the apps, but probably not, and even if you miraculously did, it would be very difficult to get into flow of the article. It’s much the same for other web development content, such as javascript articles. Reading web development content offline currently is always frustrating and disappointing, if not impossible.

  I've been in touch with the Pocket support team. I brought up the following:

  • Since Webview mode fixes the code snippet squashing I asked if it was possible to set it as default. That's not possible and in any case it turns out the Webview mode requires an internet connection
  • If they had a guide on how to structure a webpage HTML for the content and title to be correctly captured by Pocket. No such guide exists, they said that Article view, which is the mode that removes styling, requires websites to be compatible with their parser. They shared a link to a doc that describes Article mode, but it does't have the necessary information

  I suspect that I need to modify my HTML in some way, perhaps adding meta tags or just better use of HTML5. I'm left having to figure it out myself using trial and error, not a great result.

  Offline needs to be one of the first things we as developers think about. It’s basically progressive enhancement. Not considering it is a recipe for shooting yourself in the foot some time in the future. I’m still looking for a way to have my entire website rendered offline on my iOS device. It happens so often that I need to check something while offline. I read recently about support for webapps on iOS being added, so perhaps I'll look into that at some stage.

Today’s links:

• AI Getting Started - A Javascript AI getting started stack for weekend projects, including image/text models, vector stores, auth, and deployment configs - Self hosted platform released by a16z that integrates with a bunch of online APIs so you can easily create apps that do image and text AI stuff. The setup looks quite straight forward, not much info on the operation side of things, like what it looks like to use once you have things setup github.com #

• 🚀 New Post: The state of offline reading for web development content - Reading web content offline is very convenient, and occasionally essential. I have found however that it’s a terrible experience for reading web development content. I describe what I have observed in the hope that others will notice the importance of this topic. It’s one of those things you don't realise that you need until you do, and if it does't work, well then it’s too late and you are now stuck. markjgsmith.com #

• 🚀 Issues with CDN for images? - I've been having very strange behaviour from the Netlify CDN all morning. Images that are definitely in the repo and deployed, are showing up in some browsers, but not others, then disapearing. Here's a support forum post I created about it. Appologies to all that are trying to read my latest blog post about problems reading articles offline, because the images aren't loading. Oh the irony, and frustration, a post about issues seeing images offline is not readable because images of the issue are not loading online. It's some sort of bizare support case inception fractal. answers.netlify.com #

2023/06/25 #

Today’s links:

• More on Preemptively Blocking Facebook’s Imminent ActivityPub Entry - More relatively wise words from John Gruber, along with some very worthy and relevant quotes from around the web concerning Meta being blocked before they have even started on ActivityPub. daringfireball.net #

• Arctic Monkeys at Glastonbury review – breaking rock’s rules at their own strange pace - I love reading about Glasto every year, I have such fond memories from the 3 or 4 times I attended around the turn of the millenium. It's a long time ago now. Back then my music tastes were squarely in alternative rock but they were starting to broaden into breakbeat, drum 'n bass and electronic. These days Glasto is almost unrecognizable based just on the lineups. There are some really main stream acts that play, but my sensibilities have relaxed as I get older. It would be awesome to see Elton for instance. Sometimes the juxtaposition of old and new and main stream and alternative is what makes the magic of Glasto. I think I might have seen the Artic Monkeys at a festival when they were a small indie band. My favorite festival sets off the top of my head would have to be Bjork, David Bowie and the unforgetable sunny Sunday afternoon Brian Wilson from the Beach Boys. Anyway all this to say, I miss music festivals and especially Glastonbury. It really is the closest experience I have had to being in a magical city. www.theguardian.com #

• How This All Happened - Author Morgan Housel writes a very readable piece, both visually and content-wise, about the history that happened to the US economy since the end of World War II. From the uncertainty in the GIs returning, to the boom that followed because of low interest rates and cheap credit, raising most out of poverty and creating a somewhat equal society, through a crash, then another boom where this time there was a big divergence between rich and the rest. Crucially though, the 'mostly equal' expectation persisted for a long time, causing many to take on too much debt. Then another worse crash, and now the expectation has finaly shifted to many folks generally thinking society is no longer working for them. There's lots more detail, and graphs and period magazine covers, and it's just basically a very informative piece that doesn't get too bogged down. IMO worth the read. collabfund.com #

2023/06/24 #

• Unfortunately bitcoin alone doesn‘t fix the fundamental problem

  Bitcoin and crypto in general are truely marvelous technologies. I've spent the past year or so really going deep to try to understand the fundamental dynamics. Along the way I've learnt a lot about the current financial system, as well as the history of money and trade, and of course thought about the possibilities that these new monies could open up in our future.

  The conclusion that I think I'm reluctantly coming towards is that bitcoin and crypto alone don't solve the fundamental problem, and in fact I think it's very likely they could make it a lot worse. As with a lot of things that are present in so many of the things we do as humans, it's a mixed bag. It could undoubtably make things better in many respects, but at the same time it could make things worse for a lot of people.

  The fundamental problem is that people exploit other people

  I think the fundamental problem, and I say this knowing it's rather vague, but you have to start somewhere. If I were to put it into words it would be something like - people exploit other people. Of course it's much more nuanced and it's a spectrum with well paid, fullfilling and consentual jobs and businesses on one end and really unpleasant things like slavery, entrapment, starvation, thirst and other types of physical and psychological mutilations on the other.

  How people treat each other varies enormously. I've travelled to many places around the world, seen and experienced many cultures, and it's abundantly clear to me that the notion that most people are good, is totally incorrect. In some situations, it's the complete opposite. Now whether it's malice or some sort of emergent group behaviour, or a combination, is perhaps something different entirely, however the fact remains that some people are treated horrendously by others.

  One way to think about it, and admitedly it's a somewhat crude way of approaching a very complex situation, but I think it's useful nonetheless because you can get a sort of macro sense of what might be happening, is to imagine that people either fix or break other people's lives. Again it's a spectrum, from small breakages to things that are more permanent or even terminal. People can be both fixers and breakers, at the same time even. Life is full of contradiction.

  We have to fix much faster than we break

  In this hypothetical world of fixers and breakers it's pretty clear that overall the fixing has to happen faster than the breaking, or else eventually everyone and everything will end up broken. You can look at the world at many scales from countries to individual people's lives, but it's sort of the same result. The reality is that we all live on the same planet shaped boat.

  Current fixing rates are woefully innadequate

  You just have to spend a little while speaking with homeless people, or indeed have a brush with homelessness yourself to realise that the fixing rates are woefully inadequate in most places. Supply outstrips demand 99 to 1, yet there's not enough. Yes but all homeless people are drug addicts you might say. Well there might be some truth to that statement, but the question is what circumstances led to that. There are very bad people out there who benefit enormously from thoroughly breaking other people's lives. We aren't talking about lone individuals either, we are talking groups that cement their places in society by ruthlessly oppressing others.

  There are also many people in need of help that are sober, and even for them it's an almost impossible existence. Guess what looks like addiction: hunger and thirst. In fact it's basically the same thing, humans are in a sense food and water addicts. Thinking all homless people are drug addicts is not only incorrect but it's basically a death sentence to sober homeless people. And one day you too could be a sober homeless person fighting for your right to survive because others broke your life.

  Crypto ups the game by an order of magnitude

  The thing with crypto is that it opens up enormous possibilities for everyone. That's great in an evenly ballanced world, but the reality is that people that break other people's lives whether through malice or misguidely thinking they are doing the opposite, are much more active. Much like the dynamics in open source software, there are just more people trying to find exploits (notice the similarity in language) than there are people writting and fixing the software.

  That's the current state of affairs, but we are entering a period of rapid change. Crypto technologies could enable great things to happen, lets say it's an order of magnitude change, well if the fixing rates remain the same, then things will necessarily get a lot worse for many. And we aren't just talking about homeless people, or people in developing countries. Think about how many people are stuck in jobs or relationships they want out of. This issue cuts across all parts of the world and through all parts of society. I don't see any evidence that fixing rates got any better in the previous big monetary system change. Moving from gold to fiat money made a lot of things possible, but I would argue that fixing rates didn't increase proportionately.

  AI will up the game even more

  Remember also that this is happening at the same time as the mass introduction of powerful AI technologies that can analize our collective behaviours in mindboggling detail. That's at least an additional order of magnitude in the technological power available. Oh and guess what, after that it's quantum computing. Yet another order of magnitude.

  AI financing is the subject of this week's newsletter, the part about cost curves and training is particulary relevant. Out a bit later today @ 12:45 +07:00:

  The Economics of Catching the AI Wave (Issue #120)

  In a different world

  In a different world, bitcoin and crypto could be truely amazing. Yesterday I had the realisation that if a small amount of people are controlling the supply of fiat money, the equivalent to controlling water spigots, literally starving and thirsting vaste swathes of people, then why not create new monies, new water supplies? Surely it's the moral thing to do.

  I imagined a world where there were a multitude of crypto currencies, for different loosely affiliated groups. Coins for scientists, coins for musicians, for people into a multitude of fringe and mainstream hobbies. Initially it sounds like it would be too complicated, but with a bit of imagination it needn't be. You would be able to effortlessly spend all of these from your crypto wallet because all shops would be crypto exchanges, and your wallet would be smart enough, because of AI, to always have enough for what you needed or wanted to do. With integrations with other AI software that would help you manage other aspects of your life, it's not unfathomable to think that it could 'just work'.

  And by the way, it really is a small amount of people controlling the spigots, cash money is being created at 7 times the rate than actual humans are being created. That's an indication to the current ridiculous innequality throughout the world. It's so mindboggling that it's difficult to fully comprehend. It doesn't have to be that way though. Things could be a lot better for everyone.

  And that's why I'm so conflicted on crypto. I return to the main point of the article. These amazing bitcoin and crypto technologies simply are not enough. IMO the cultural make up of our societies is in a state that could make crypto adoption very dangerous long term, where a class of folks, that remain somewhat in the shadows, have enormous power by ensuring breakages in peoples lives happen faster than they get fixed.

  That's the core dynamic that needs to be fixed. Maybe Fiat is the worste money, but until the breakages problem is handled, perhaps it's only the worste money except for all the others.

• The Economics of Catching the AI Wave (Issue #120)

  This week’s newsletter is out! (2020-06-24)

  In this week’s edition:

  We are heading into a future where AI is in every technology we use. You need to understand the fundamental cost curves, and maybe dangers, that are in play in solving the training conundrum.

  Issue details:

  • Title: The Economics of Catching the AI Wave
  • Issue: 120
  • Page: issue webpage

Today’s links:

• Thoughts from "Meet Safari for Spatial Computing" - Nothing earth shattering but gives a bit of an idea what to expect in visionOS as a web developer. Best to use resolution-agnostic units like rem. Hover states are ignored but visionOS Webkit will automatically add it's own version of :hover when a user looks at elements on a page. Generally speaking keep using progressive enhancement, well written accessible HTML and other web standards such as responsive images, media queries, and vector graphics and your website will still look good. blog.jim-nielsen.com #

• The New CSS - What I got from this article is that around about 2020 there was a significant sea change in CSS. With features such as :has(), container queries, custom properties, CSS grid, new color spaces, and auto scaling fluid typography, you can create designs that flexibly adapt to different contexts and different types of content. So much so that, if you usually only dabble enough to get the job done, it's probably worth doing another pass at learning CSS using the latest reading material because the way people are thinking about the landscape is fundamentally different now. There is a sense by many that all these new features are coalescing together into a 'Modern CSS'. matthiasott.com #

• Understand Units in CSS: A Comprehensive Guide - These all make sense, but yikes there are so many. I didn't know that you can use resolution units to activate different CSS across devices with differing resolutions. So for example high resolution devices could get higher resolution background images. Assuming only the appropriate image is downloaded, then that's pretty cool. www.hongkiat.com #

• 🚀 New Post: Unfortunately bitcoin alone doesn‘t fix the fundamental problem markjgsmith.com #

• 🚀 Latest Newsletter: The Economics of Catching the AI Wave (Issue #120) markjgsmith.com #

2023/06/23 #

Today’s links:

• How "Exit Traps" Can Make Your Bash Scripts Way More Robust And Reliable - Like a finally in a javascript try/catch but for your entire bash script. Several very useful examples that I wouldn't have thought of. redsymbol.net #

• Sitemaps XML format - The way to tell search crawlers about the structure of your site. Yet another thing I'll need to implement. You literally have to have an entry for every single page, but you can build index sitemaps that point to other (nested) sitemaps. Without sitemaps I'm guessing your site isn't going to showup very often in search results. And you need a robots.txt file too. www.sitemaps.org #

• AI Is a Lot of Work - Quite a long article, somewhat ironically a bit of a slog to read, but it's well written and certainly eye opening. The AI Annotation Industrial Complex that is highlighted is the iceberg behind the recent AI models like ChatGPT. There are elements that really remind me of the VFX industry, such as project based work that moves around the globe quickly based on demand, and workers clubbing together to help each other out. But it also feels like there could be some exploitative practices going on. The numbers are bordering on mindboggling, with current estimates in the millions, but projected to move into potentially billions of people. That's right, some predict double digit percentages of the entire human population of earth employed training robots, though it would likely be somewhat temporary until base models are trained. The article ends on a bit of a strange snake eating it's tail thing which left me worried about how structuraly sound this house of cards actually might be.www.theverge.com #

• Islamic State: Woman jailed in Germany for keeping Yazidi woman as slave - Reminder that people's views on slavery still vary enormously around the world. Something to keep in mind both at home and abroad. www.bbc.com #

2023/06/22 #

• The return of number go up

  From the linkblog 2023/06/22:

  Bitcoin Breaks $28K for the First Time Since May - A flurry of activity in the crypto space has raised the price above $28k, up from 22k in just a few days. It appears to be in response to the world's largest asset management firm Blackrock's announcement of the first Bitcoin ETF, which would open up the possibility of investing to many that previously would not have been able to. Previous to this, it had been a slow and steady decline in price, which appeared to be following the general sentiment that the economy is heading into a deflationary period. This comes at the same time as US regulators have been charging big crypto exchanges for handling unregistered securities, and prominent TradFi figure John Stark has been straightup advising folks to 'get out of crypto' because of the storm brewing and major podcasters in the space are predicting big regulatory headwinds. Nevertheless at least for now, it's the return of number go up.

  I thought the graphs were pretty interesting given all the background. Here they are:

  

  

  

  

  

Today’s links:

• AI meets the other AI - With talk recently of the House Oversight Committee having a hearing about UFOs, and with the present moment being so dominated by AI, it's dawning on people that not only could we use AI to detect UFOs, but that learning about AI will make us much better at preparing for a future encounter with alien intelligences. www.politico.com #

• Not That Kind of ‘Open’ - John Gruber gets to the core if why those trying to block Facebook's new ActivityPub powered Twitter app are essentially participating in something akin to pre-crime at best. And also however you feel about Facebook/Meta, there are a lot of people that use the service, so really all you are doing is blocking the open web from growing. It seems that might be exactly what some people would prefer, open for we but not for thee. daringfireball.net #

• Style your RSS feed - Describes how to style your RSS feed using XSLT and CSS. I'm not sure exactly where I fall on styling your RSS feed. It's true that feeds look a bit intimidating to new users, especially for non-programmers, but is a styled feed really necessary since most blogs & podcasts have an HTML page that lists all the posts/episodes anyway. For example my blog and podcast, and the seperate page that lists all the feeds. At least when you see the weird looking XML you know you've found the feed to copy/paste. Aren't you just confusing things even more by styling the feed? Also when you actually want to see the XML of a feed, something that I find myself needing to do quite a lot, there isn't an easy way to remove the styling. darekkay.com #

• Sitemaps XML format - The way to tell search crawlers about the structure of your site. Yet another thing I'll need to implement. You literally have to have an entry for every single page, but you can build index sitemaps that point to other (nested) sitemaps. Without sitemaps I'm guessing your site isn't going to showup very often in search results. And you need a robots.txt file too. www.sitemaps.org #

• Bitcoin Breaks $28K for the First Time Since May - A flurry of activity in the crypto space has raised the price above $28k, up from 22k in just a few days. It appears to be in response to the world's largest asset management firm Blackrock's announcement of the first Bitcoin ETF, which would open up the possibility of investing to many that previously would not have been able to. Previous to this, it had been a slow and steady decline in price, which appeared to be following the general sentiment that the economy is heading into a deflationary period. This comes at the same time as US regulators have been charging big crypto exchanges for handling unregistered securities, and prominent TradFi figure John Stark has been straightup advising folks to 'get out of crypto' because of the storm brewing and major podcasters in the space are predicting big regulatory headwinds. Nevertheless at least for now, it's the return of number go up. news.yahoo.com #

• Binance starts whipping up Bitcoin Lightning Network nodes - In other crypto news Binance, one of the biggest crypto exchanges, has massively expanded their ability to handle the layer 2 Lightening protocol, setting up their own instances. They plan to use the Lightning network for users to receive and export funds. If I am understanding correctly, this should significantly speed up transactions, taking the load off of the base layer. www.theblock.co #

• 🚀 New Post: The return of number go up markjgsmith.com #

2023/06/21 #

Today’s links:

• AI: First New UI Paradigm in 60 Years - Makes the case that we have entered the third major UI paradigm in computing. First it was batch processing (punch cards), then command based interaction (cli's and later guis), and finally it's intent based outcome specification (tell the computer what you want rather than how to do it). Doesn't exactly roll off the tongue. We need something a bit like WYSIWYG but for AI interfaces. www.nngroup.com #

• Eight days later: Kbin, Lemmy, the landed gentry, and the rise of the "threadiverse" - Current state of the open source decentralised Reddit alternative space, content moderation and defederation as a temporary protective measure. I think the idea of using the term 'landed gentry' to describe certain online groups is very clever and in many cases quite appropriate. privacy.thenexus.today #

• Phishing frenzy: School kids are stealing millions of dollars of NFTs — to buy Roblox skins - Is this article for real? Kids making millions of dollars and then spending it in Roblox? Surely not. www.theblock.co #

2023/06/20 #

Today’s links:

• AI shakes up way we work in three key industries - Looks at Legal Professional Services, Filmmaking and Coding. In filmmaking I thought it was kind of interesting that actors are negotiating deals that include their AI double do they can be recording an advert using their AI at the same time as shooting a film and get paid for both. And in coding aparently AIs are better at making frontend code. I'd like to see more data on that. Though many folks across industries project potentially more work, it seems people are really worried that wages will be erroded over time. www.ft.com #

• Celebrities Use AI to Take Control of Their Own Images - Another article that looks at the possibilities and challenges of AI body double technology, in particular for acting, fashion and sports. I wonder if 3D photobooths will start poping up in public spaces so normies can get in on this AI body double craze. www.wsj.com #

2023/06/19 #

Today’s links:

• The Staggering Frontiers of CSS - The author runs through some of his favorite new features including View Transitions for making multi-page apps more slick, Anchor Positioning, Dynamic Viewport units and Subgrid. He finishes up by expressing a sentiment I've been feeling recently about the web platform in general, I call it webdev vertigo, that things have gotten too complex to fully understand by one person. "It feels like we’ve entered this new phase in the life of CSS where just keeping up with what’s shipping each year is too much for any one person to keep track of, let alone take advantage of". I just hope it all doesn't turn to grey gew. buttondown.email #

• Update from Lemmy after the Reddit blackout - One of the big winners from the Reddit civil war has been Lemmy. It's a decentralised open source Reddit alternative. Their monthly active user count has balooned from 1000 to 27000, and is now one of the largest Fediverse projects. It's an interesting project, lots of neat sounding features. Seems incredibly futuristic compared to my humble linkblog. It makes me wonder if there are any easy ways I could integrate my setup into the Fediverse.join-lemmy.org #

2023/06/18 #

Today’s links:

• GitHub Actions Trigger Via Webhooks - I've been looking for options to automate site rebuilds on creating new content. There is reusable workflows, which is quite neat because the called workflow runs and literally appears as part of the caller workflow. Another optiom is to use webhooks, which, if I'm understanding the docs correctly, results in the called workflow running outside of the caller workflow, so you get essentially 2 workflows running, in 2 seperate repos. It's a somewhat subtle difference but operationally it's a very different way of doing things. I'm currently exploring using a combination of the two techniques to effectively share workflows across many repositories, which should make setting up new projects much less error prone and far easier to maintain. mainawycliffe.dev #

• Unify and modernize your web architecture with Netlify Connect - There are a couple of things that tend to be true which you learn over many years of being in tech and being a developer. Eventually every enterprise sized company releases a product called 'Connect'. The other thing is that one-api-to-rule-them-all solutions are usually not the safest thing for the end user long term, because though it might be convenient, switching becomes very difficult, and so eventually over time the platform treats the user worse and worse because the incentive structure promotes that behaviour. I love Netlify, they have been a great platform, so maybe they will find a way to avoid this dynamic. The only way I can think that might aleviate it would be making their fancy API open source, so people aren't locked-in to their solution. But building a healthy ecosystem has it's own challenges, case in point the on-going total destruction of the Reddit multiverse situation. We don't seem to have figured out the recipe for creating successful, healthy and sustainbable communities. Seems like that might be a multi-generational endeavor. www.netlify.com #

• Netlify presents: Edge Functions general availability - The idea of edge functions is awesome. Practically speaking though it means handing over full control of the DNS of your entire domain, and currently at least, building against a propriety API. Having more open standards in this area implemented by multiple companies is a real urgency. www.netlify.com #

• JavaScript’s Backtick Strings are Likely the Wrong Tool for Your Job - Short and useful post that highlights some of the pitfalls to think about when using string literalls, and makes the case for using URLSearchParams when constructing urls. spin.atomicobject.com #

• Why you should pin your GitHub Actions by commit-hash - Basically you can avoid supply chain attacks by specifying the full commit hash of every community built action you use, for example the 'checkout' action, instead of specifying the module version. Yes that avoids the attack in the immediate term but how do you maintain your repo now? You are now tied down to the ground like Guliver. They mention using dependabot which supports version numbers in comments, but how does dependabot know the version it's promoting hasn't been compromised? And now your workflow files are full of illegible commit hashes. Supply chain tragedy. blog.rafaelgss.dev #

2023/06/17 #

• Don‘t Missout on Magic (Issue #119)

  This week’s newsletter is out! (2020-06-17)

  In this week’s edition:

  The sometimes haphazard nature shipping, the subtleties and nuance of expressive changes and the changing landscape of digital tools for creators.

  Issue details:

  • Title: Don‘t Missout on Magic
  • Issue: 119
  • Page: issue webpage

Today’s links:

• Sequence diagrams, the only good thing UML brought to software development - For all of you out there getting into the latest perfect rust idioms or what have you, spare a thought for those of us that had to trawl through entire UML books, diligently studying them to pass masters degree exams, only to find it was basically all a total waste of time. Apart from sequence diagrams that is. I have to agree with the author of this blog post, sequence diagrams are indeed very useful. They are the computer science equivalent to electrical engineering circuit diagrams. If you are going to learn 1 type of computer science diagraming, learn sequence diagrams. They make it possible to visually and accurately communicate very complex flows of information. There are times during implementation where groking the big picture is vital because hings can get confusing quickly. You will need a map with just the right anoubt if detail of what's actually going on. That map is often a sequence diagram. www.mermaidchart.com #

• My 24 year old HP Jornada can do things your modern iPhone still can't do! - The author of this article is very very nerdy but I totally agree with his main argument. You should be able to program a device from the device itself. That's not the case with most modern mobile devices, but it doesn't have to be that way. That's what I was alluding to when I imagined the HTML5 Phone. That's what computing used to be all about. I think with a bit of thought we could have both nice looking devices, and user programable devices. raymii.org #

• Everything Must Be Paid for Twice - Makes the interesting observation that people often forget about the cost that will be incurred after making a purchase, corresponding to the time and effort required to fully see the utility of the purchase. Quite a timely piece for me. Doubling seems to be trending at the minute. www.raptitude.com #

• Architecting a high throughout digital effects facility - Following on from yesterday's Fragments of Capital FX blog post, having re-read all the various fragments I found, I think this article has the best description of the facility we built, from a technical infrastructure perspective. If you are technically inclined and spend time thinking about tools and environments for creators, this is definitely worth reading. markets.businessinsider.com #

• 🚀 Latest Newsletter: Don’t Missout on Magic (Issue #119) markjgsmith.com #

2023/06/16 #

• Fragments of Capital FX

  If you have been reading the blog or newsletter for a while, you'll know that I'm a big fan of podcasting. I've been listening to podcasts since the medium started in the early 2000s. I'm always reading about the developments in the industry with a curious eye. A recent pivot by Spotify that I read about really reminded me of my time working in the VFX industry, there are a lot of similarities in how that unfolded.

  With a hint of nostalgia in my sails, I went looking for Capital FX, my old company, online and was suprised to find there weren't that many traces remaining. This post is a compilation of some of the fragments I found during my search.

  It's not that suprising that there wasn't much remaining. It was a different time back then. The internet wasn't so ubiquitus. Sure we had the web, but mostly that just meant being able to send email, looking up company addresses and perhaps reading the news on the BBC website every now and then. Social media wasn't a thing yet. Facebook was only just starting to get popular at about that time. The idea of having an internet presence was totally foreign to basically everyone. Why would you want to tell everyone about the cash cow you had built? To most normal people that actually seemed like a rather stupid idea. You would occasionally put out press releases, but many of the publications that published those have since gone offline.

  First a bit of background info. Capital FX was a feature film post production house specialising in special effects, digital intermediate, and later, digital cinema. I worked there from around 2005, building out the infrastructure as the company grew rapidly in size, taking on increasingly large high profile Hollywood blockbuster projects. The guys that started the company were, in a lot of ways, film production pioneers. They provided various services to film makers, most important and lucrative of which was foreign language versioning. They started small but quickly became central in London's cottage feature film industry.

  Capital FX was acquired by Deluxe in 2006, and I continued to work there until 2008. It was a very interesting time to be there, because the entire industry was transforming to fully digital pipelines, from capture to theatre, but we were also dealing with movies shot on old school film. We worked on hundreds of movies, some of which are listed in IMDB. The aquisition was particulary interesting because we suddenly had access to other Deluxe properties including world famous digital intermediate EFilm. We worked closely together to build precission transatlantic production workflows that were used on films such as Children of Men, and The Golden Compass. Big name directors liked working with us because we had the most advanced equipment in the industry, and we knew how to get large complex jobs done on time at very high quality.

  So what remains of Capital FX? Well the website is long gone but I was able to recover the company logo from the internet archive.

  

  This is perhaps my favorite fragment of all. Look at that beautiful cubist and somewhat brutalist combination of small and big letters. And only 1 color! Imagine the confidence that this little photoshop fill tool special gem inspired as it zipped by on reels being watched on a Steenbeck, or indeed on expensive 4K digital projectors in blacked out colour grading theatres. All the goodie bags we gave to clients had this printed on the side of them. Wonderful. Such memories.

  Before moving on to the rest of the fragments in our VFX archeological digg, let's take a short detour to learn about How London became the VFX capital of the world. I think it's important to set the scene a little. Ok now that you are feeling the vibe, lets look at the rest of the pieces that I was able to recover. I have also included links to internet archived versions of these pages, denoted by [ia].

  Company info listings

  DSG DIGITAL LONDON LIMITED (Companies House Listing - The Company) [ia] - You can see it was previously called Capital FX before the acquisition.

  DSG DIGITAL LONDON LIMITED (Companies House Listing - People) [ia] - Some of these folks were the big cheeses so to speak, also features names of folks I used to hear about all the time but didn't actually meet because they weren't actively envolved anymore.

  With Capital FX [ia] (IMDb - Sorted by Popularity Ascending) - This one gives you a sense for all the awesome movies we worked on. There were a lot more that aren't listed here. Turns out it's quite difficult to get in the credits, especially when the company that acquires you, in our case that was Deluxe, literally has it's logo at the bottom of about 1/2 of all films. The other 1/2 by the way have rival company Technicolor's logo at the end of the credits.

  Capital FX, London (IMDbPro - Client & Contact Info) [ia]

  Capital FX (London) (BFI - British Film Institute) [ia]

  Capital FX Ltd - CFX - London (4rfv) [ia]

  Staff press releases

  Shelton, Clarke join London's Capital FX (Screen News) [ia] - We had many people join us as the company grew, it was a very cosmopolitan environment, folks from all over the world.

  Tech press releases

  Capital FX expands Digital Intermediate Services (TV Technology) [ia] - We built a for hire digital intermediate facility that directors would hire out during their production. We developed workflows where footage was being shot in LA and graded the next day in London, all sent digitally over the internet. The color science to get it all looking the same in all locations was quite complicated.

  SAN Speeds Capital FX’s DI process (Infostor) [ia] - So much storage, everytime we setup new petabtyes of storage it would get gobbled up almost immediately.

  Capital FX Speeds 2K and 4K Workflow on Major Hollywood Movies Using SGI InfiniteStorage Technology (Markets Insider) [ia] - We had a lot of SGI equipment, very fast very large storage for editing at high resolution. This article has probably the best description of the digital pipeline we built. Such a lot of equipment all pieced together in a few short years all while using the infrastructure at near capacity. Not only did we upgrade the plane's engine while flying the plane, so to speak, but no one had ever really built such an engine before. Using this kind of technology at such scale in the creative industries just hadn't been done before.

  Lustre Digital Color Grading System Shines Worldwide (Animation World Network) [ia] - All the color grading software and hardware cost enormous amounts of money.

  UK's CAPITAL FX BUYS THIRD FURY FILM RECORDER (Post Magazine) [ia] - We also had many Arrilaser recorders, and of course an Arriscanner which could scan between 4-8 frames per second.

  Doremi DCI mastering and playback purchased by Deluxe Digital Cinema (UK Broadcast News - 16/05/2006) [ia] - All the d-cinema stuff was super high security, everything encrypted, very expensive pieces of kit, often required custom workflow integrations.

  Case studies

  Case study: Archiving the blockbusters at Deluxe [ia] - The tape machines were awesome, basically a giant robot jukebox. So many gigabytes of storage.

  That's all the fragments I found, there's likely more out there. I remember looking about 4-5 years ago and there was a lot more. So many things on the internet eventually disapear, that's why I wanted to capture a bit of a snapshot now of Capital FX, the pioneers of digital intermediate, visual effects and digital cinema. It was tremendously fun working there!

Today’s links:

• Spotify pivots from being HBO for podcasts to YouTube for podcasts - Previous acquisitions Gimlet Media and Parcast are losing their branding, joining the umbrella Spotify Studios. It has also been organising high profile events on the french riviera in Cannes, where companies can court advertising agencies. It's incredible how similar this sounds to what I lived through when a VFX house I worked for called Capital FX got bought out by Deluxe. It was simultaneously exciting, wonderfull and a very difficult process, where the old guard got slowly transitioned out while the new upstarts were brought in. A sort of multi-year game of musical chairs, that you only realise you were playing in retrospect, executed like a perfect no down time infra migration, where you never expect to be phased out, until you too are phased out because, while you were extremly-busy-doing-important-things, without realising it, you turned into the old-guard-next-generation. www.bloomberg.com #

• Meta AI researchers unveil I-JEPA, a computer vision model that learns more like humans do - I'm not by any means very well versed in all things AI, but this sounds like a big step forward. These new models create an internal model of the world and compare abstract representations of the data instead of the actual data. That sounds more similar to how humans learn to me. Apparently it results in models that are less likely to get caught up in irrelevant detail and minutea. siliconangle.com #

• 🚀 New Post: Fragments of Capital FX markjgsmith.com #

2023/06/15 #

Today’s links:

• The culture war for your face - Some push back against the future that could be created by the Vision Pro. It's an important conversation to be having as we move forward. Honestly even though I'm optimistic and enthusiastic about the future and the possibilities of spatial computing, I worry a lot about distopian scenarios. Historically speaking humans haven't exactly treated each other very well over the years, slavery is a big part of our past. It's also possible that we could innadvertantly create something horrible. Or, and this is perhaps the most likely and scary, some bizare combination of the two. Having said that, we must move forward, discovering and avoiding bad patterns along the way by listening and having empathy to/for each other. www.politico.com #

• Scientists Discover Phosphorus On Enceladus - Pretty big news in astrobiology, phosphorous is a fundamental component of DNA and energy-carrying molecules. It now looks like Enceladus has many of the elements and compounds that were necessary for life here on earth. tlpnetwork.com #

2023/06/14 #

Today’s links:

• Naked JSX: Use JSX without React - NakedJSX is a command-line tool for generating HTML files from JSX - I'm still trying to wrap my head around this, it seems so unbelievably simple that I'm thinking I must be missing something. It's a replacement for whatever templating language you use. For me that's EJS. So instead of EJX template files, I guess I would have javascript files that build the page using components created in JSX. It also has some clientside features. Very interesting indeed. nakedjsx.org #

• The US Is Openly Stockpiling Dirt on All Its Citizens - So governments are slowly turning into tech giants? Maybe eventually it will be so obvious they'll have to give us all free storage. Of course the other thing that might happen is that the tech giants turn into the government. Or I suppose a bit of both. This modern world we live in is very very strange. www.wired.com #

• A Developer's View of Vision Pro - Thoughts from a developer that was there from the start for the iPhone, iPad and Apple Watch. He was at the launch event and was left speechless by the quality of the experience. High on his list of reasons to get in early is to build up his "platform intuition". Nice article though a bit light on actual ideas about what sort of things will be possible. david-smith.org #

• Basic authentication with custom HTTP headers - Some perhaps good news in my side quest to organise my RSS feeds. If I'm parsing this netlify doc correctly, looks like I can add basic auth to static pages by setting an HTTP header, which solves one of the issues I listed. I also might have found a solution for an RSS reader that supports subscribing to dynamic OPML files. Ideally though I need a client that runs on iOS with good offline features. docs.netlify.com #

• Awesome HTML5 - So many incredible features in this web platform we have built. It will be interesting to see how the Vision Pro and visionOS changes the focus. It's a good time to get a broad view of what's currently possible because there will be a lot of possibilities for integrating the two platforms. github.com #

• 🎙Apple's Web Apps, Meta Quest and Vision Pro, and Missing Sticky Headers Ep#569 (Shop Talk Show Podcast) - Lots of interesting webdev bits and pieces but the highlight is clearly Chris and Dave getting into some proper web developer Vision Pro dev talk. We are at the very beginning of the roller coaster. That's when folks are slowly realising how awesome and cool this might turn out to be. I love this bit of the journey. shoptalkshow.com #

2023/06/13 #

Today’s links:

• WhatsApp Channels turns on broadcasts for creators and organizations - It's such a popular app that this is likely a big deal. Feels like every app eventually turns into a Twitter clone. www.theverge.com #

• How to Use The HTML5 Sectioning Elements - The past few days, I've been trying to ensure that pages from the website look good on social media sites, as well as in read later apps. Sometimes you have to re-read the basics just to make sure you understand important details. blog.teamtreehouse.com #

2023/06/12 #

Today’s links:

• Archive.today is a time capsule for web pages! - Similar to the Internet Archive save feature, this service enables you to take a snapshot of any url at a particular point in time. That's useful because sites go offline, but they also change. Looks to use something called the Momento protocol. I wish there were more services like this, and that the technology was more pervasive and robust. It doesn't feel like there is an accepted standard. How do I create a url backup on my own hardware for instance? archive.today #

• Kong Yiji: The memes that lay bare China's youth disillusionment - I think what this article describes isn't particular to China. When you zoom out, there are many many people worldwide that spend literally multiple decades of their lives in isolation. The unwritten promise from society has been that it will lead to a better life. But that promise is being broken in a big way. Folks that didn't choose that route find themselves with money, and now with AI, it turns out they didn't need to study anyway. The discontent from the skinny educated is bound to become something that will need to be reparationed. www.bbc.com #

• Tools for thought: the first 300,000 years - A super high level summary of some of the big technological innovations that humans have made which have resulted in our modern world. It's funny to me because I've been preparing something similar in my spare time recently. I was focusing on science whereas this focusses on actual inventions. Two sides of the same coin in a way. Anyhow worth reading if you are at all interested in history and culture and perhaps the future. subconscious.substack.com #

2023/06/11 #

Today’s links:

• Coding a deep learning model using TensorFlow.js (2020) - Following on from yesterday's tutorial for getting a pre-trained model up and running, here's the next tutorial in the series. It shows how to create and train a model from scratch. I haven't managed to get completely through it yet due too lack of internet connectivity and Safari reading lists not working for this site for some reason. The impression I get is that it's considerably more complex an endeavour. Hoping to circle back to this at some stage, or perhaps find a more recent example. developer.ibm.com #

• Why AI Will Save the World - Mark Andreessen makes the techno-optimist case for AI. I'm on board. Honestly though the past few days I've been feeling a bit of tech vertigo. Even just the web stack is mind boggling, and that's just a tiny part of it all. I need a hoverboard because the escalator is going so fast there will be a point in the not too distant future where there will be no way to keep up with any of it in a meaningfull way. a16z.com #

• US urged to reveal UFO evidence after claim that it has intact alien vehicles - I keep an eye out for UAP articles as a sort of curiosity. This one actually looks sort of credible because it's from a well reputed intelligence official who has gone before Congress. www.theguardian.com #

• Someone Should Write a Screenplay - Hugh Macleod piece about how technology and the internet is affecting the creative trades. Touches on the current Hollywood writer's strike, AI and the parallels in the wider economy. Gets to the core of the issue. www.gapingvoid.com #

2023/06/10 #

• Apple’s Vision Pro, Crypto vs SEC (Issue #118)

  This week’s newsletter is out! (2023-06-10)

  In this week’s edition:

  Apple announces the most complicated and futuristic device ever made, the SEC tries to blowup the entire crypto industry, and I figured out and fixed the newsletter issue numbering thing

  Issue details:

  • Title: Apple’s Vision Pro, Crypto vs SEC
  • Issue: 118
  • Page: issue webpage

Today’s links:

• How First Principles Thinking Fails - A short exploration into situations where first principles thinking doesn't lead to the right result, something I've been wondering about myself for a while. The danger when it happens is that it's often completely unnexpected, and so runs the risk of being catastrophic. It's a good idea to keep an open mind to other ways of approaching a problem. commoncog.com #

• 🎙We Studied Thousands of Heads (Accidental Tech Podcast) - The guys review the entire WWDC conference, which is pretty standard. Vision Pro coverage starts ~2 hours into the episode. Quite technical discussion from the experienced Mac user / developer angle, covering topics including field of view, social stigma, latency, not being presented as a gaming device at all, capturing 3D video, no killer app yet, currently lots of 2D screens in 3D, new area of the AR/VR market focussing on work applications, spacial audio, potentially cool dev environment setups, not really priced for consumers atm, power cable disconnect shuts everything off, and possible apps in shopping, interior design, meditation and fitness spaces. atp.fm #

• Redditor creates working anime QR codes using Stable Diffusion - People are inventing some quite innovative ways of using Stable Diffusiom AIs that take existing QR codes and moddify them into art while keeping the encoded QR data intact. arstechnica.com #

• An introduction to AI in Node.js - Reading through this tutorial, it's not actually that difficult to get a pre-trained model up and running. I find the way that these AI systems are architected to be a bit strange, perhaps unintuitive, or maybe it's just the way the interfaces are described. I'm not able to parse them as quickly as say web APIs. Anyway the cool thing is it's possible to do it all from Node.js using Tensorflow.js. developer.ibm.com #

• Here’s what Mark Zuckerberg thinks about Apple’s Vision Pro - It will be super interesting to see how this all unfolds. It's actually good that there is diversity in approach right from the start. It's an ecosystem, lots of room for everyone. Hopefully we will remember some of the lessons learnt from the past, like interop and open protocols. Healthy competition. I think there's a path where all the companies and people in the space are helping each other, building the future together. That's when the tech industry is at it's best. www.theverge.com #

• News from WWDC23: WebKit Features in Safari 17 beta - Based on WWDC reviews on various podcasts, it's clear this year's WWDC was kind of mega. Of course there's the Vision Pro, that's what everyone's talking about, but there was a ton of other hardware and software announcements. Reading through the Webkit release notes I'm blown away by how many features have been released across HTML/CSS/JS, it's truely mind boggling. I'm also getting a sensation that I think can best be described as a sort of general vertigo, about how unfathomly complex the entire web platform is. Just wanted to mention that now, because the complexity is about to increase significantly. The 3D stuff has my mind racing about how Vision Pro is going to change webdev. Awesome but also a bit scary. webkit.org #

• Web Apps on macOS Sonoma 14 Beta - MacOS joins iPad/iOS in supporting web apps, this is a detailed writeup, complete with lots of screenshots. Works with any site, no need for a manifest file, and appear as a regular app in the Dock. I'm imagining all sorts of cool things I could add to my website that would enable it work offline as a web app. Wouldn't it be awesome if the whole site worked offline? blog.tomayac.com #

• 🚀 Latest Newsletter: Apple’s Vision Pro, Crypto vs SEC markjgsmith.com #

2023/06/09 #

Today’s links:

• Royal Navy Says Quantum Navigation Test a Success - Seems like this could be quite important because it eliminates the need for GPS and hence satellites which can be jammed. Theoretically should work for subs too. Happy to learn the technology was developed at Imperial College, that's where I did my Materials Science undergrad. thequantuminsider.com #

• Automattic launches an AI writing assistant for WordPress - What happens to content on the web when the most popular open source publishing platform adds generative AI features? Many are concerned about spam but the people that really want to make great things are still out there, so it's possible that quality might increase, with ever more intricate and detailed stories that were previously unnimaginable. techcrunch.com #

• Apple Vision review from Stratechery - Ben found the experience to be "extraordinary", his piece covers the real-time OS, it’s mirrorless and mixed reality nature, the eye tracking powered user interface, the opportunities in the productivity space, replacing the iPad as the ultimate consumption device, and Meta being totally out-classed on hardware but perhaps not on human connectedness stratechery.com #

• 🎙Apple Vision (Another Podcast Podcast) - The two experts in all things brainstorming return to review the latest WWDC announcement, covering amoung other things, why the Vision Pro is the dream device from 5 years in the future but now, hardware description, spatial computing, AR vs VR, Apple’s insistence to not compromise on quality, the Apple vs Meta approach, whether mass adoption will occur, how demoing for spatial computing is different, and the you-look-through-it metaphor podcasts.google.com #

• 🚀 New Recommendation: Chris Reichhelm, founder & CEO at Deep Tech Leaders and Deep Tech Jobs markjgsmith.com #

2023/06/08 #

Today’s links:

• Culture Design Strategist - Given my lifelong fascination with the notion of culture, this job advert from Hugh MacCleod turned my head. He's always light years ahead of everyone, and occupies a different realm in terms of how he sees the world. The idea that you can engineer culture had occurred to me, but I had no idea it was turning into an entire discipline. I once had a fun lunch with Hugh at one of his meetups in London, we all went to an art gallery afterwards as I recall. I think I'd like to be a web developer culture design strategist. www.gapingvoid.com #

• Github Actions - trigger another action after one action is completed - I'm trying to streamline some of my publishing workflows, looks like reusable workflows is what I'll need to implement so that the site gets automatically re-built when I close a PR on the content repo. Currently it's a two step manual process. stackoverflow.com #

• GGML - AI at the edge- Implementing one of these AI models still feels a little mysterious to me. This project looks quite interesting, works on commodity hardware. It would be cool to setup a minimal example of a setup where you could train a very basic AI classifier and deploy it to the cloud. I guess most of these LLMs are more for generating text, but would work well in concert with a classifier. ggml.ai #

2023/06/07 #

Today’s links:

• Why did Usenet fail? - It's articles like this that make blogging such a wonderful medium. Well written with nice pacing whithin a story arc that flows, some useful links and good comments. Really like the site design too. And totally relatable content, I too tried usenet a bit back in the day and eventually moved to centralised websites for basically the same reasons. shkspr.mobi #

• Vision Pro: Apple's new augmented reality headset unveiled - They sure do look like snowboard goggles, and quite the price tag at $3499. I'm curious how they will do zoom-like conferences, could be very cool, but won't it be strange to look at several other people also wearing goggles? Or will they GPT generate fake faces? www.bbc.com #

• Introducing Apple Vision Pro: Apple’s first spatial computer - The official press release from Apple, worth reading to get a sense of how they envison this new world of spatial computing. There's a new operating system called visionOS, with a new interface, lots of AR features, and they appear to have seriously thought on data privacy and security. Overall it's very intriguing and compelling. It's their first new hardware product line in a decade, so quite a big moment. www.apple.com #

• The Open Graph protocol - I'm trying to add the right meta tags to the site <head> so that pages appear nicely on various social media platforms. There are a lot of possibilities but it's not so obvious how to engineer a way to add more than just the basics. The only two ways I can think of are hard code into the site templates or by using markdown frontmatter, or a combinatiom of both. ogp.me #

2023/06/06 #

Today’s links:

• The Culture Map: How to Navigate Foreign Cultures in Business - Gives a concise breakdown of some of the major cultural dimensions to consider when interacting across cultures. Written for business but much of it is useful in other scenarios too. ahalbert.com #

• Iceland: the emerging tech-ecosystem of the Nordics - I've never been to Iceland, sounds pretty cool, I hadn't considered timezone-wise how convenient it is when working with US east coast and western EU. erikdestefanis.substack.com #

2023/06/05 #

Today’s links:

• Photography's new bag if tricks (1984) - Interesting in the age of AI enhanced and generated photography to see how people viewed regular photo editing technologies when they were introduced www.nytimes.com #

• JavaScript Macros in Bun - Cool feature that sort of turns your code files into render templates. You import functions specifying an attribute, and use those functions in your code. They are executed during bundling, with returned values added into the final outputted code. The functions can be async like running a fetch. bun.sh #

• Beijing's comedy crackdown is hitting its music scene - This obviously sucks if it's true, music is so important for society. The other thing that sucks though is that all evidence presented are contained in other BBC published articles. How about linking to some external sites to back up claims? www.bbc.com #

• Oil production to drop after Opec+ nations meet - It's sort of weird when you think about it. If your main export is oil, it's totally fine for you to print / burn, but if your main export is money itself, and everything that makes the money possible, then print / burn is bad www.bbc.com #

2023/06/04 #

Today’s links:

• OpenAI is pursuing a new way to fight A.I. 'hallucinations' - Their technique called 'process supervision' rewards based on each individual correct step in a multi-step reasoned problem rather than only the final outcome. Interesting approach but sounds a bit like mind control. Who decides on the steps in the first place? Isn't the whole point of AI that it would be able to generate steps that humans would never be able to envisage? www.cnbc.com #

• Telly dual-screen TV first look: it’s free and may be the future - This new product seems comically silly and dumb at first, but the more I read about it, the more I'm cuious about what's going to happen when it arrives in stores. Pretty sure it's going to ruffle a few feathers at the very least. There's a chance it's a stroke of pure genius, which doesn't happen very often. Also as a brit clearly I like the name, it literally is 'the telly'. www.theverge.com #

• Sanda Dia: Belgium reckons with verdict over black student's hazing death - I really dislike these hazing rituals where foreigners are systematically abused and harrassed. I didn't growup in my home countries and I've always felt like a bit of a foreigner everywhere I've been, including at 'home'. IMO it's an outdated part of our collective cultures. Great to see the students standing together against these horrid incidents. www.bbc.com #

2023/06/03 #

• Quite A Lot Of History (Issue #117)

  This week’s newsletter is out! (2023-06-03)

  In this week’s edition:

  This week there’s quite a lot of history and culture in the podcasts, all very interesting, and often relevant to the tech conversations du jour. The links have some good tech and javascript items.

  Issue details:

  • Title: Quite A Lot Of History
  • Issue: 117
  • Page: issue webpage

Today’s links:

• Japan Goes All In: Copyright Doesn’t Apply To AI Training - Very topical with the writers strike in Hollywood, and also really important for all, there's a new arms race beginning, we need to ensure we don't inadvertently destroy entire culture ecosystems technomancers.ai #

• Zeal is an offline documentation browser for software developers - This looks cool, it's similar to devdocs.io, but better because the data isn't stored in ephemeral browser storage. Wish there was an iOS app. zealdocs.org #

• Steak, coffee and cheese locked up as shoplifting rises - Not a very good omen www.bbc.com #

• WebRTC support being added to FFmpeg - Simple HTTP-based protocol that will allow WebRTC-based ingestion of content into streaming services and/or CDNs, which would make it much easier to for example setup YouTube, Telegram or Facebook livestreams news.ycombinator.com #

• The new Motorola Razr Plus looks kind of amazing. It's foldable with a pretty big outside screen that you can use with most apps, and it opens from clam shell to a massive screen. Great tech specs, and has the vibe of the old Razr flip phone but modern. www.theverge.com #

• Elon Musk: Twitter boss reclaims title of world's richest person - At $192 billion net worth, it's mind boggling to me that Elon could buy 128 million Macbook Airs. What is this world we are living in? www.bbc.com #

• 🚀 Latest Newsletter: Quite A Lot Of History markjgsmith.com #

2023/06/02 #

• Organising my RSS feeds

  One of the items that's been on my todo list for a couple of weeks now is to organise my RSS feeds. I've been tidying them up best I can in short bursts here and there, but I'm struggling, feels like I'm walking up an escalator that's going the other way. It never feels like I'm making much progress.

  Feedly has been my app of choice for many years now. The interface is pretty good, there are some nice features. I have the iOS app, and it works quite well, but the one thing it doesn't do well, or at all, is offline. When I'm offline, the app opens but it's just an empty screen, no way to do anything. I find the best time to catchup on RSS feeds is when offline so that's not very practical.

  I had a bit of a look for some feed readers in the App Store and found one called Newsify. It's quite good, even offers to login using my feedly account so I was up an running quite quickly, no need to export and re-import feeds. I configured it so the articles appear in a compact way, and the reading experience is relatively good.

  All the articles get re-formatted so they all look the same no matter which website you are reading. That's sort of annoying. I know it probably saves a bunch of bandwidth, but I really like to read from the website in question. Striping out all the styling kind of ruins the vibe.

  Couldn't there be an option to fetch the actual published web page and store it similar to how Safari Reading Lists work? Surely in this day and age that would be possible. Looks like I'm able to re-organise my feeds in Newsify, though I haven't tried that yet.

  I think though what I'd like to be able todo is to manage my feeds in the same way I manage my blog, linkblog, podcast and newsletter data. I just want to add a feeds folder in my data repo and have a markdown file for each feed. Then I could build one or several OPML subscription list files which I would host on my website. I'd like to be able to import that into whichever feed reader I was using. That way I can easily try out new readers. Maybe even try using the feeds in some app I could build myself sometime in the future.

  The biggest show stoppers:

  1. Feed readers only seem to support OPML files for bulk import, not for dynamically getting the list of feeds
  2. The OPML file would need to be public, or if basic auth could be supported by the reader app, how do I add authentication to the static files I host on Netlify?

  Maybe these are solvable issues, but figured I'd write a blog post about it. If you know any solutions that could help, please email me, I'll write up the setup if I manage to get it working :)

  Btw, here's an example OPML file that lists the feeds on this website.

Today’s links:

• These tiny, medical robots could one day travel through your body www.colorado.edu #

• Advanced Fastify: Hooks, Middleware, and Decorators - It's been a while since I've looked at the fastify project, really liking the plugin and hooks systems, not to mention express compatibility blog.appsignal.com #

• Node.js built-ins on Deno Deploy - This is awesome. The deno team are on a phenominally good run of consistent and impressive releases. I really want to try out their platform for some apps. deno.com #

• Vanilla With Candy Sprinkles - A heck of an impressive demo app, so many options and global too. Would be cool to try out the fly.io platform too. I especially like how easy it is to spin up databases. fly.io #

• tj/node-migrate - Abstract migration framework for node - Database migrations can be tricky, but if you have the right tools, it's much easier. This library is database agnostic, which is awesome, and has a very express middleware vibe to it, also awesome. Of course it does, it's from tj. github.com #

• Leaders have signed the US-Taiwan Initiative on 21st Century Trade in Washington which aims to strengthen economic ties. China has denounced the move, and has previously described such negotitations as 'extremely dangerous'. www.bbc.com #

• 🚀 New Post: Organising my RSS feeds markjgsmith.com #

2023/06/01 #

• When will AI start to affect macro?

  The world of AI is moving at such a huge pace at the minute and it seems to be accelerating. ChatGPT started it all off with it's chat based interface, and we were all blown away with what it was able to do.

  Shortly after that people started to create multi-step versions of these types of applications that can use external tools to retrieve and manipulate data. Some of the first use cases for these AutoGPTs are code assistants but they can be used for any task. These AutoGPTs are incredibly powerful, orders of magnitude more powerful than their predecessors.

  There are already comprehensive new javascript libraries with which you can build much more than just chat apps. Entire advertising campaigns that previously would have taken months to construct can now be created in minutes. And there's a lot lot more.

  And the money is flowing in that direction too. There have been some reports that important venture funds are repositioning away from crypto to so called frontier tech. In fact blockchains and AI are a very good match, with the ability to autonomously spin up wallets and perform very complex chains of activity.

  There's also the recent annoucement that Elon Musk's company Neuralink is going to start clinical trials of human computer interactions via brain implants. It might be a good time to seriously consider our right to mental privacy in the age of brain sensing tech, because when you combine some of these technologies, say with something like edge computing, the possibilities for creating unpleasant adversarial environments are very real. Nice income streams you have there knowledge worker, it would be ashame if you were suddenly unable to function. The future is here, and it's going to be crazy weird.

  With all that in mind, it's also important to step back and ask how this will affect the wider world. Something that worries me is escalation. We know things in the economy already escalate. To a large extent it's always been this way, you literally can see it in the data. But with AI the occurrences of dynamics of escalation are going to increase significantly. These aren't always bad per say, but I believe they can be, and it could get a lot worse so to speak.

  We should start collecting key metrics so we know how things used to be before AI, what the world looked like, what it felt like to live in. We need to be able to see if and how things are being affected at a macro level. Wouldn't that be prudent?

  We've also got quantum computing on the horizon, IBM plans to have a 100000 qubit machine operational with 10 years. What will happen when we add that to the mix? Surely it's going to be another step shift.

  Big leaps are also being made in nanotechnology, where tiny robots can be sent into the body to deliver medicine in an incredibly targetted way. All very amazing, but the fact is they could also very well be used to deliver harm to people too. We will need to be able to detect things at that level too. Vulnerable populations like the homeless are already targetted with foreign material placed in their food. It could get a lot worse, and when it does, it could happen at scale and very quickly.

  Let’s boldly go into the future and explore the possibilities of these new technologies but let’s avoid a multi story car park collapse of humanities’ cultures. Let’s make sure some of the worste outcomes are avoided with a bit of structure in important areas. Remember ultimately we are all in the same earth shaped boat.

  What type of data would we need to collect to get some visibility without destroying the very culture and creativity we are trying to protect in the first place?

Today’s links:

• CSS Blend Modes - At first these don't look that impressive but as you go through the examples you eventually realise the examples look pretty darn cool, and you realise they could have called this feature CSS Andy Warhol, or CSS Cartoon Strips garden.bradwoods.io #

• LangChainJS the popular AutoGPT library is announcing support for multiple JS environments including browsers, Cloudflare Workers, Vercel/Next.js, Deno, Supabase Edge Functions, alongside existing support for Node.js ESM and CJS blog.langchain.dev #

• Going Beyond Chatbots: How to Make GPT-4 Output Structured Data Using LangChain blog.langchain.dev #

• Libuv devs have added io_uring support for several asynchronous file operations, including read, write, fsync, fdatasync, stat, fstat and lstat. Works on newer linux kernel versions and gives an 8x performance boost. Falls back to the thread pool for older kernel versions. I'm wondering if that means my ssg is going to run that much quicker. Sounds kind of amazing. github.com #

• WPP one of the world’s largest advertising groups, is teaming up with Nvidia to use AI to create in minutes campaigns that would have previously taken weeks. The system will be able to generate fully photorealistic images using 3D software which can then be used in AI generated video or 2D, eliminating the need for filming or green screens. 10000 variations of a car advert can be generated in a couple of minutes, varying all elements. www.ft.com #

• 🚀 New Post: When will AI start to affect macro? markjgsmith.com #