Smart in theory

Now the embargo on reviews of the Humane Ai Pin has come to an end, Humane might be wishing it hadn’t. Let Cherlynn Low at Engadget stand as an example of what’s out there:

[Low is describing the way you find yourself needing to enter a number in order to validate your identity multiple times a day, on a device that lacks a keypad so has to project numbers onto the palm of your hand. Sounds complicated, but I’m sure Humane’s designers thought users would get used to it over time.]

This gesture is smart in theory but it’s very sensitive. There’s a very small range of usable space since there is only so far your hand can go, so the distance between each digit is fairly small. One wrong move and you’ll accidentally select something you didn’t want and have to go all the way out to delete it.

“Smart in theory” That damning phrase might serve as Humane’s epitaph.

Given how frequently the response to reading about the Ai Pin seems to be some variation of “what does this do that my smartphone can’t do better and faster?”, three thoughts arise:

1. Displacing the smartphone is going to take quite some doing; and,
2. Paying Humane US$699 for the Ai Pin itself, plus a monthly bill for the supporting cell service/gateway to off-device network services, is a sign that some people have more money than sense;¹ and,
3. Are we 100% certain that the founders of Humane weren’t sent on a kamikaze mission by Apple to remind us all of how useful smartphones are?

Of course, almost everyone asking about using a smartphone to accomplish similar goals has not yet had hands-on experience of using an Ai Pin. Perhaps using the device in real life will transform opinions. That’s not the impression I get watching Joanna Stern’s 90 second video review on Twitter/X, but it’s early days yet.

[Via Daring Fireball]

View Source

Gary Ings’ article in HTML Review issue 3, a view source web, takes me back:

On my personal websites view source meant being able to adapt and remix ideas. Like drawing a map, elements and pages acted as landmarks in the browser to be navigated between. As a self-initiated learner, being able to view source brought to mind the experience of a slow walk through someone else’s map.

It’s also very nicely presented. Look at it on a decent web browser and be impressed.¹

View Source on modern web browsers tends to reveal a whole heap of custom CSS, so it can take some digging to get down to the HTML building blocks of the article you’re reading. I’m glad I taught myself HTML in those simpler times.

Nowadays 99% of what I post here I write in Markdown² because that’s the core of what I’m trying to communicate here: the words and hyperlinks with the occasional embedded image or video. Nobody will learn very much by using View Source on this content, I’m afraid.

[Via Pixel Envy]

Mind the Orwellian Surveillance Apparatus ¹

The biggest surprise to me reading this piece about Transport For London’s experiment at Willesden Green tube station is that all this extra technology could piggyback on the existing, slightly old and outdated CCTV cameras. ²

[This] was not just about spotting fare evaders. The trial wasn’t a couple of special cameras monitoring the ticket gate-line in the station. It was AI being applied to every camera in the building. And it was about using the cameras to spot dozens of different things that might happen inside the station.

For example, if a passenger falls over on the platform, the AI will spot them on the ground. This will then trigger a notification on the iPads used by station staff, so that they can then run over and help them back up. Or if the AI spots someone standing close to the platform edge, looking like they are planning to jump, it will alert staff to intervene before it is too late.

In total, the system could apparently identify up to 77 different ‘use cases’ — though only eleven were used during trial. This ranges from significant incidents, like fare evasion, crime and anti-social behaviour, all the way down to more trivial matters, like spilled drinks or even discarded newspapers.

So, the system identified 77 different use cases but they only decided to use 11 of them. That graphic would look a lot scarier if the left pane listed all 77 potential use cases.

Given how much the alerts the system generates rely upon station staff reacting to them in order to fix the issues being identified, it’d be nice to imagine that the quantity of incidents revealed might argue for increasing staffing levels.

Why do I have an uneasy feeling that it might not go that way?

[Via LinkMachineGo]

WilliamsF1Parts2024.xlsx

It’s not so much that Excel is a bad place to store your parts catalogue as it is that it’s a less than ideal starting point if you go no further than dumping your data in Excel:

It is not an exaggeration to say that up to and including at least the initial work on the 2024 Williams, its car builds were handled using Microsoft Excel, with a list of around 20,000 individual components and parts.

Unsurprisingly, ex-Mercedes man Vowles - someone used to class-leading operations and systems — had a damning verdict for that: “The Excel list was a joke. Impossible to navigate and impossible to update.”

Managing a car build is not just about listing all the components needed. There wasn’t data on the cost of components, how long they took to build, how many were in the system to be built.

[…] “When you start tracking now hundreds of 1000s of components through your organisation moving around, an Excel spreadsheet is useless.”

“You need to know where each one of those independent components are, how long it will take before it’s complete, how long it will take before it goes to inspection. If there’s been any problems with inspections, whether it has to go back again.”

“And once you start putting that level of complexity in which is where modern Formula 1 is, the Excel spreadsheet falls over, and humans fall over. And that’s exactly where we are.”

I have a vision of some poor intern at the back the garage¹ working their way through an in-tray, which is struggling to contain a huge pile of printouts² listing changes that needed to be entered into the spreadsheet YESTERDAY. Not a fun or glamorous job, even if you get to wear the pit crew’s uniform for Williams Racing.

Fair to say that Williams Racing Team Principal James Vowles is talking about this so that in a couple of years time he’ll be able to reassure us that WilliamsF1Parts2024.xlsx is a thing of the past.³ Hopefully by then Williams will be a bit further up the leaderboard (if only on the principle that other teams might be having a worse time that year.)

[Via The Overspill]

Spreadsheet errors

I’m always up for someone reminding us of the impact VisiCalc had on the world.¹ Let Tim Harford pick up the story:

Despite its stuttering beginning, VisiCalc quickly became a phenomenon. Watching those two strangers walk out of his presentation in 1979, Bob Frankston could hardly have dared to hope that, three years later, Apple II computers were being sold as “VisiCalc accessories” — the $2,000 entry fee to get access to the spreadsheet, a $100 miracle. Unsurprisingly, it was the accountants who caught on first and drove demand.

Bricklin recalled in a 1989 interview with Byte magazine, “if you showed it to a person who had to do financial work with real spreadsheets, he’d start shaking and say, ‘I spent all week doing that.’ Then he’d shove his charge cards in your face.” […]

Tim Harford, being Tim Harford, has to remind us of the downside of all this reliance on technology:

[…] Looking at the way spreadsheets are used today certainly suggests a warning. They are endlessly misused by people who are not accountants and are not using the careful error-checking protocols built into accountancy for centuries. Famous economists using Excel simply failed to select the right cells for analysis. An investment bank used the wrong formula in a risk calculation, accidentally doubling the level of allowable risk-taking. Biologists have been typing the names of genes, only to have Excel autocorrect those names into dates. ²

When a tool is ubiquitous, and convenient, we kludge our way through without really understanding what the tool is doing or why. And that, as a parallel for generative AI, is alarmingly on the nose.

I acknowledge that electronic spreadsheets might have been invented to retroactively justify the aphorism, “To err is human. To really $!&@ up — takes a computer,” but I don’t think it’s fair to spreadsheets to equate them with Large Language Models hallucinating all over the place.

Shōgun

I’m old enough to have watched the Richard Chamberlain/Toshiro Mifune version of Shōgun back in the day, but I don’t remember it well enough to fairly evaluate the quality of the 1980 version this long afterwards. I remember it was a huge hit at the time, part of a resurgence in the limited series/miniseries format that followed in the wake of Roots a few years earlier.

Having at least three lead characters, two of them Japanese, sharing the main storyline seems likely to work better than focussing as heavily on our English central character’s experience of Japan as the earlier adaptation did.

The first four episodes of the 2024 take on the story have me gripped, not to mention glad that we’re getting episodes weekly rather than being invited to devour the whole story in one lump. I’m glad that I now have a whole week to think about how the fourth episode ended with things taking a distinct turn for the worse¹ for Lord Toranaga’s attempt to quietly outmanoeuvre his fellow regents.

Lots of reviews of Shōgun have made comparisons with Game Of Thrones and that’s not unreasonable because it’s referring to early seasons of that show, when Game Of Thrones was still seen as ground-breaking work, perhaps the best fantasy show on television.²

Fair to say that I’ll be watching all ten episodes of Shōgun with great interest despite my having a general idea of how the story turns out. Shōgun looks likely to repay as much attention as you’re willing to give it.

The facts on the ground

An Icelandic take on how the European Union sees industrial monopolies:

What should surprise you is how accommodating and outright gentle the EU has been with Apple’s shenanigans over the years, whether it was about exploiting loopholes to avoid phone plug standardisation, or their violations of EU antitrust regulation that pre-date the new Digital Markets Act.

That’s because the EU is manifestly pro-business, but it feels forced to act because the single market is the EU and the EU is the single market. To Apple, the App Store is a side line. To the EU, the single market is the foundation of its existence.

Any time you see two entities of similar size fight, bet on the one that thinks it’s fighting for its life.

Remember, Steve Jobs initially presented the iPhone as having a modern web browser called Safari, “[So…] you can write amazing Web 2.0 and Ajax apps that look exactly and behave exactly like apps on the iPhone.”. Jobs allegedly had to be pressed by insiders to open up the platform with an SDK that allowed third parties to get into the App development game on the iPhone. Or was all that Jobs, knowing the SDK wasn’t ready for release yet, spinning a tale to cover Apple until the App Store was ready for the world?

In the years following 2008’s opening of the App Store income from the store was a pleasant bonus for Apple, arising from the way the smartphone market exploded in those first few years with Apple’s products representing the premium-priced end of that market (both because of Apple’s pricing and because Apple customers supposedly exhibited a greater willingness to pay for third-party apps.) Apple got lucky, to the point where nowadays the App Store is arguably much less of a “side line” for the company than it started out being.¹

[Via Links for the Aboard Newsletter]

The Douglas Adams Age

Matt Webb reckons that Tech has graduated from the Star Trek era to the Douglas Adams age:

See, AI is the most Douglas Adams of all technologies.

Large language models, GPT-2, GPT-3, ChatGPT and all the rest, are relatively simple under the hood. There’s not much complexity to the code, so I’m told. But there is a monstrous quantity of data and training.

OpenAI didn’t invent the transformer architecture LLM. But they were the first to do the engineering to make it really, really big, and see what happened. That it would work out was unexpected!

[…]

It is absurdly improbably that you can hoover up the internet, shred it, then talk to the mulch pile and it talks back.

Truly this is the age of Douglas Adams technology.

Bring on the Babel Fish. ¹

A funny old game

Is Dan Ashworth a marquee signing that Dan Ashworth would recommend?

If Ashworth does manage to squirm out of his enforced gardening leave and take on his third long-term vision in just over five years, then he will be reunited with his close friend, the marginally gainful Sir Dave Brailsford. [Emphasis added]

A bit off-brand for my weblog, I know, but the number of long-term visions that come to naught because, in part, of what eleven players do on the pitch¹ over 90-plus minutes ought to be humbling.

The Big Picture

The Big Picture: How We Got Into This Mess, And How We Get Out of It from Robert Reich.¹

[Via Memex 1.1]

How 1984 almost didn’t happen

Just over 40 years ago, Apple decided to run a TV advert during the Superbowl. Just the one time, mind.

GOLDBERG I had them do a theater test. We get back the results, and it’s the worst business commercial that they’ve ever tested, in terms of persuasiveness.

SCULLEY The board said, “We don’t think you should run it. Try to sell the time.”

GOLDBERG And it was Jay Chiat who told us to drag our feet, basically, when we were told to sell off the time on the Super Bowl.

HAYDEN At long last, it came down that we would run the “1984” commercial once.

Hard to imagine a world where Apple didn’t run this ad. Whether or not it was a persuasive advert for a very expensive piece of technology that could just barely accomplish what it claimed to do,¹ it certainly reminded anyone paying attention of the contrast with IBM’s vision of the future of computing.

[Via Pixel Envy]

Standing room only

Interesting to see an Apple Vision Pro in action in something resembling a real-world environment.

Seems to me that hemming yourself into your living and working space with pinned windows showing various apps would get old very quickly. But then, I spent most of today working from home¹ hammering out replies to emails and I could imagine myself doing that surrounded by various spreadsheets and windows from my email client pinned in front of the walls of the room I was working in.

Would it seem more or less oppressive if those windows were hovering everywhere the eye could see rather than all fitting on one display² in front of my keyboard? I can imagine the immediate effect being liberating, as I spent time rearranging windows into the optimal pattern, but by the end of the day I suspect the novelty would have worn off.

No question, this video looks as futuristic as hell. Whether we want to work like that is a different matter.

[Via Scripting News]