Sitrep

It's been too long (again) since my last "here's some interesting things worming their way around my head" update. I had a couple of weeks off for the Easter holiday when updating my blog was one of the many things that started out on my list of things-to-do that ended up on my list of things-I-didn't-do…

I mentioned last time that 'sharpening my Python, Pandas and Matplotlib skills' had been keeping me busy - since then I've been looking at Django (a framework for building websites - similar to Ruby on Rails) with the idea of being able to build some sort of front-end to go along with the data-driven/analytical back-end stuff, and then moved on to some deep learning tools (Keras and TensorFlow and some natural language processing libraries), which have been... challenging, but fascinating.

As my last couple of posts probably indicated, the world of artificial intelligence, machine learning, deep learning and large language models have been taking a lot of my attention (like a lot of people in tech/media at the moment, I suspect.) But with a bunch of not-quite-finished posts about how deep learning models actually work, whether AI can create 'art' and other questions about what they can/can't do - and a few other related topics - this is the place for... other stuff.

The Things That Made Me Want To Write More Than A Pithy Sentence/Short Paragraph

Not that they are necessarily more interesting - just that they caught my attention a bit more...

Tech

As mentioned, my attention to the world of tech has been almost entirely diverted by my AI/ML/whatever obsession that I'm not going to allow myself to be distracted by here... other than to say that Elon Musk's TruthGPT idea is possibly his most stupid idea I've heard yet - and thats a pretty high bar that he's set himself after the whole Twitter thing. Also, thats without account for the fact that he's also signed the open letter asking everybody to stop building powerful AI systems.

That said - WWDC is only a month-and-a-bit away, and as someone who is a) firmly ensconced in the Apple ecosystem (I recently added an iPad to my Macbook/iPhone/Watch/Homepod suite of Apple gadgets) and b) interested in development tools, it looks like it should be an interesting one this year.

Maybe Apple are going to talk about what they have been working on in terms of AR (VR?) - they make the best mobile chips, they have a significant mobile software platform, and a whole bunch of interesting tech thats already out there (eg. spatial audio in all their wireless headphones, 3D-cameras on lots of iPhones/iPads). Basically, the headset might be the next big thing as far as hardware products go, but what makes an Apple headset intrinsically more interesting than a Meta/Samsung/Microsoft/Sony headset isn't the hardware device but the software platform.
Siri is getting old (its almost teenage), and since its launch the world has come to expect something more impressive from chatting to a computer than its getting from the current generation of 'smart assistants'. Again - in the background, Apple has been building in AI-related tech (iPhones, iPads and Macs have had dedicated 'machine learning cores' for a good few years now), putting Apple in a position to do something that Google/Amazon/Microsoft/OpenAI can't. Siri might have been slowly improving since its original launch, but I'm not sure if anyone has really noticed (If you read the Wikipedia page for Siri you'd think that the only improvement since its launch other than being available on new devices were the addition of some new voices) - it feels like now would be a good time for Apple to make a choice; either a significant "Siri 2.0" update if Apple wants to dedicate the resources to making Siri a competitive and distinctive offering, or to announce something like a Google/OpenAI partnership if they just want their 'smart assistant' to stay relevant.
I think its safe to say that I'm never going to own a Lamborghini (or a 1959 Les Paul), but that doesn't mean I'm not interested in them... I like the analogy of the Mac Pro being like a high-end car; its ridiculously expensive, but what you learn by designing and building things at the bleeding edge of technology helps you know what will filter through to the mainstream at some point in the future. I'm not sure if its still true; what Apple learnt from selling iPods and Macbooks helped them build iPhones and iPads, but now its what they have learnt from building iPhones and iPads that is driving innovation in Macbooks and iMacs. All that said - I'd still really like to see what Apple would come up with when they try to build the best desktop computer in the world. (Particularly how they approach the question of how much GPU power is 'enough'.) I want to believe...

(Speaking of which - I'm loving the 3rd (and probably final) season of Ted Lasso on Apple TV.)

Guitars

I had a 'proud dad' moment a couple of weeks ago; my son saved up for and bought his first electric guitar of his own (replacing what was my first electric guitar back in... 1993?) Which means not only is his guitar collection already as big as mine is now (2 electrics and an acoustic), but at about the same age that I was when I first started playing. He's also getting to be about as good as I am now (please, somebody make it stop...), which has incentivised me to spend a bit more time thinking about practicing my technique rather than just noodling away at trying to play whatever the latest song stuck in my head is.

It sounds like boring homework, but it actually means that, for the first time in quite a while, it feels like I'm getting noticably (to myself...), technically better - which means when I pick up my guitar, I'm having more fun with it. So thats nice...

It also means that, a bit like how upgrading my family's iPhones last year prompted me to buy myself a new iPad, its got me thinking more than I should about whether I deserve a new guitar for myself… Maybe a more modern Strat as an upgrade for my 30-year old model (that hasn't been particularly well looked after for a fair chunk of those decades and has a few dodgy knobs and odd rattles...)- or maybe something different with a hollow body...

Or maybe there are better ways I should be spending my money. (Like guitar effects pedals...)

Cars

Well, 'a car' - my old Nissan Note (bought used shortly after my son was born) has served me well, but a quick 'top up shop' trip to the supermarket about a week ago ended with me getting into the car, turning the key and... nothing happening.

After a failed attempt to jump start it (by the way Ford, when you're designing the layout of all the bits under the bonnet, please spare a thought for the people who are trying to jump start a car when you decide to tuck the battery in a position that makes it completely impossible to get a jump lead to one of the terminals), 3 hours of waiting around for the breakdown recovery service and a bit of awkwardly standing around and agreeing while the recovery man pointed at things around the engine and saying critical things about whoever put something somewhere, it turned out to be a blown fuse.

Once upon a time, I had a car (a Vauxhall Nova) and some sort of idea what I was looking at when I opened the bonnet - I could even do some basic maintenance myself. Now, I can say that I can replace a fusible link. It could have been easier - in fact, the most challenging part was getting the old one off, because apparently the designers of this car seemed to assume that any work being done would be done by someone with at least 3 hands (at least one of them the size of a small child's) and the strength of... well, of someone with more strength than me.

Or maybe just a wrench with a longer handle…

Anyway - the thing that caught my attention was how, before towing my car home, the engineer/mechanic/very nice man texted me a link to a YouTube video that explained what you need to do when you're driving a car thats being towed. I remember getting towed for the first time about 20-odd years ago and being absolutely terrified of doing something wrong because I hadn't properly understood what the mechanic told me but being too intimidated to say that I wasn't entirely confident that I knew what I was doing - I can't help but think that watching a very short and very simple YouTube video would have made me a lot more comfortable that I knew what I needed to know. (The issue of having enough phone battery to actually watch a video after waiting alone in a car with no power for a few hours is one drawback though...)

Still - my car starts now, which was pleasing. Almost as pleasing as discovering I wasn't bleeding once I got all the grease off my - quite large - hands. But again - its got me thinking about whether I deserve to be driving something more reliable (and newer, more economical to drive/ecologically friendly/technologically up to date etc.) I can't help thinking that a guitar would make me happier - and a lot cheaper. (Thats whats's called 'framing' in the world of behavioural economics - a £1,000 guitar might sound expensive, but compared to a car that costs ten times as much its actually quite cheap…)

(If I start posting things about electric cars, now you'll be able to guess why.)

Games

I've not really got stuck into a 'new' game for quite a while; although I've been playing through Ghost of Tsushima recently, I was left with the feeling that I was 'playing it wrong' (spending too much time on side-quests at the expense of the main story - there are a lot of side-quests that get quite repetitive, so I think I missed out on some of the fun of a great storyline.) Before that, "Evil Genius" quickly turned into a bit of a time-sink (that is, the kind of game where the amount of time I spend playing it doesn't quite seem to align with the amount of fun I'm having playing it- but for reasons I can't quite explain, I keep on playing…)

But I think its looking like a good summer for games; I think the last Zelda game (Breath of the Wild) is probably my favourite game of all time, so the sequel next month is going to be an immediate must-buy. I also liked the Star Wars Fallen Order game, so thats another sequel that I'm looking forward to playing - although it probably won't be an immediate launch purchase (because if I'm going to drop £60 on a game with two to choose between, I'm not going to start with the one that probably won't still be £60 in a few months time...). Also, Spider-Man 2 is scheduled for later this year; the first game was excellent on the PS4 and even better with the graphical overhaul it got on the PS5 (as was the Miles Morales game - although a considerably smaller game than the original, it had some interesting and fun twists of its own.)

Also, I resisted the urge to buy the last - and final - FIFA game (although the inclusion of Ted Lasso and several players from the fictional Richmond F.C. team tempted me, it didn't seem like enough to justify the price tag), so I'm very interested to see what EA's new 'non-FIFA licenced' game will look like. Apparently it isn't going to be a return to the International Superstar Soccer days, when there was the 'official' football game and the 'unofficial' one where the players names were all mis-spelt versions of actual players and played for weird clubs like "London F.C."; they still have likeness rights for lots of clubs and players. Not quite sure what I'm hoping for; I always liked the 'management' side of the game (buying better players, training up the ones that you liked, building your own formations etc.) as much as the actual game, but I felt that the FIFA franchise never really had as much depth as the 'other game'.

What I'm really hoping is that they don't expect people to buy a new full-priced version of the same game every year - having played FIFA '22 and seen how the game kind of 'shifts' once the new game comes out and the player base gradually migrates across (basically, online matchmaking gets worse, the market for buying new players dries up, the quality of players you can buy improves throughout the season so at the end you end up against some ridiculously overpowered players that you'll never be able to buy for your own team because the market dries up...)

Some bookmarked links

Not many, partly because things I thought were interesting a couple of months ago - as usual - don't seem quite as interesting now, and partly because I've consciously left out any that are about AI, Deep Learning or Large Language Models, as I'm pretty sure everybody with even the remotest level of interest has heard enough about all that stuff over the last couple of months...

OK - just one' Stephen Wolfram explains ChatGPT - technically LLMs, but I've not seen a better explanation that doesn't just gloss over some of the important technical bits just because its hard to understand while still keeping it (reasonably) understandable.
Designing for colorblindness - The Verge - I'm colourblind (a similar 'flavour' of colourblind to Andy Baio, it seems - given that when I move the sliders on the pictures in this article, I don't really see any difference), so I'm always interested when people try to explain colourblindness. There's a word ('qualia') that describes a subjective, conscious experience that you can't describe without referring to the experience itself - a commonly used example is how to describe 'redness' without just referring to something red. I've known for years that the reds that I see (or maybe the greens, or maybe just the in-between shades) aren't the same as the red that someone else sees. But language simply doesn't give us a way to describe our subjective experiences; there's no way to know that two people using the same word are actually describing the same experience. Which means that there are some pretty significant limitations about exactly how much information/knowledge about human experience can be derived from a language model - no matter how large it gets... (OK - nothing more about LLMs from here on. I promise.)

Pro tip: when you meet a colorblind person, don’t repeatedly point to things and ask what color they are.
Privacy in the Metaverse might be Impossible - this has some really interesting implications; I know that a lot of work has been done on how to 'fingerprint' people based on how they use a computer (eg. tracking patterns in how people type, where they point their mouse, what you can glean from the size of a browser window etc.) Well, this study has found that if someone is wearing a VR headset with a controller in each hand (ie. how someone would normally use a VR headset), then even ignoring all the other data that the headset can collect - the microphones, the depth-sensitive cameras that can build a 3D picture of the room you're in, whatever it is that means when I turn my Quest 2 headset off and on again that it still knows the position of my desk/sofa etc. - the motion of just those three points in space (headset + 2 controllers) is enough to identify the user from 100 seconds of data (from playing Beatsaber) with over 94% accuracy. In fairness - Beatsaber is about as physically intense a game as I've seen; maybe a couple of hours of practice might be enough to shift your movements from what has been profiled, but still - this could have some.... interesting consequences.
Social media is doomed to die - The Verge - technically, I guess this is really why all social media platforms are doomed; there's no reason why a new platform won't pop up and go through the whole lifecycle all over again (Mastodon, BlueSky, Substack Notes - are you reading?) Essentially, the article blames advertising - which isn't unfair; a platform starts up, it burns through investment, it builds a nice platform with a nice community, it builds an advertising product to keep the cash flowing... then it has to grow the advertising product to keep investors happy and it spoils what it already had. The irony here is that (I think) the way to make it work is something beside an advertising product - like, say, a subscription model that gives users additional benefits/features. Except the way to do that - obviously - is while you're building a platform; not buying an establised platform and then trying to turn it into something different...
Whats fundamentally wrong with Elon's Blue Tick strategy, from the perspective of someone who knows what they are talking about.
A good post on Linkedin (not sure if I've ever said those words in that combination before...) about data literacy and data science.
The Block: Taylor Swift did due diligence on FTX, says lawyer who finally served Shaq, Tom Brady - Taylor Swift is smart. (Or just better advised than most - I like to believe that the artist who could have used her status to lobby for better Apple Music/Spotify/etc. royalty rates for herself and chose to publicly lobby for all artists has a better idea of what she's doing than most though.)
Uniqlo’s Parent Company Bets Big on Tiny RFID Chips - WSJ - via Ben Evans' newsletter; Uniqlo has this cool thing where you get to the till, drop the clothes you want to buy into a little boxy thing by the till and it identifies all of them through RFID tags in the price tags. The weird thing is that this sounds like a new thing - they have already had them in the Westfield store for quite a while. (I've used them there a couple of times.)
The End of Computer Magazines in America | Technologizer by Harry McCracken - An end of an era; America has no more magazines about computers. As someone who used to read them (and deal with their embarassing covers - I have a particularly strong memory of a copy of PC Format with a glamarous cover model licking a graphics card) from the early days of ZX Spectrum magazines through to early 21st century PC magazines, this makes me feel old… On the other hand, I've been using Macs for over a decade and have never bought a Mac magazine, so this feels more overdue than anything else. (Surely the only people buying computer magazines today are people who don't really understand computers?)
RMS Carpathia - This was in James Whatley's Five Things On Friday (which neither has five thing, nor does it often come out on a Friday - but is still very good); a story about the only boat that managed to pick up any survivors after the Titanic sank.
Obsidian - a note-taking app - since getting a new iPad, Notability has become an essential day-to-day app, and I was thinking that maybe I don't really want/need a 'big bag of text files' notes app - the more I use Notability, the less 'important' nvAlt/Day One become. But the way text notes in Obsidian can link to one another - or even be embedded in one another - takes me back to what I liked about Voodoopad when it was still in active development. I can see this becoming my new favourite place to put my words - at least, while the wait for nvUltra continues...
Bo Burnham's Inside - I can't remember if I posted anything about this (I have a feeling that I started writing something that got too long too quick), but something made me listen to the soundtrack again the other day and reminded me how excellent it was.
UK Honey isn't necessarily honey - I found it weird when I heard John Siracusa talking on a podcast about how what is sold as 'olive oil' in the US isn't necessarily 'olive oil' - or even contains the real stuff - but assumed it was just one of those US things. Turns out, the picture in the UK isn't necessarily any better...
John Higgs' KLF book is getting updated and re-released. I loved this book - and all of his other books that I went on devour - and I think the promise of an added commentary is likely to be enough to get me to buy it again.
(Video) Which game do you say "Check" when you're playing? Loved this...

Thats all I've got for you for now - well done for making it to the bottom of the page; I sincerely, honestly hope it was worth it. (And if you had to skim the stuff about artificial intelligence, I don't really mind- there will be plenty more of that for you to skim though before too long.)

B'bye!

Some Random Nerd