0.0 Context Setting

It is Tuesday 7 July in Portland, Oregon, where we are totally not living through what could be described as William Gibson’s Jackpot.

I mean, the minority senate leader might be in a persistent vegetative state and there’s like a non-zero chance there’s a bunch of techbros going around lobbying for him to become a McConneLLM and thus be uploaded for all eternity, and they’re probably fighting with another band of techbros who see him as the ideal candidate for a brain-computer interface so he can keep legislating from the not-grave.

0.1 Events: How People Work, Live!

Matt Jukes will be joining me on Friday 10 July at 9am PT / 12pm ET / 5pm London for the inaugural How People Work, Live!

How People Work, Live! is based on my workshop, How People Work, itself based on my 25-odd years worth of experience of making software with other people for other people to use: in other words, all the messy stuff of making products and services in a world of humans. There’ll be a bit of business therapy, some talk about how most organizations don’t have good or useful strategy, definitely lots of opinions about writing/communication and generative AI.

If you’ve been to one of my Hallway Track events, How People Work, Live! is different in these ways:

• it’s recorded -- I’ll be sticking them up online at some point
• it’s a lot looser
• more people can come

But it’s the same in exactly these ways:

• it’s free!
• it’s about 90 minutes long
• it’s casual
• always interesting people
• audience participation
• talking about interesting things
• and you will probably learn something (that might even be useful, too)

Friday, 10 July 2026 on Zoom at 9am PT / 12pm ET / 5pm London and I even checked so you can come to it and still watch the world cup game. You should go and register now.

1.0 Some Things That Caught My Attention

1.1 Well, What Would You Do?

Distracted driving is bad. I’m not even going to look up a source for it. Let’s just stipulate it, okay?

Technology that makes driving/cars safer is also good! It can be good for the driver and passengers (who presumably do not, in general, want to be injured or die), and it can also be good for people who are not in cars at all and just trying to live their lives without being hit by an average of around 1.8 tons traveling at least thirty miles an hour (bonus points for mixing units there).

We got a new car last year; I think I’ve written about it before. It is a computer on wheels plugged into a 98 kilowatt hour battery. It’s pretty cool! It is super fun to drive, and when we have an electrical blackout (which has not happened since we got the car), then we can plug things into it because it also does vehicle to load. It also has a stupendous amount of safety features that our 2014 Subaru Outback (we live in the Pacific Northwest?) does not have, like blindspot sensors (great), and 360 cameras that produce a synthesized overhead view of the car that you can also spin around because did I mention it’s a computer on wheels plugged into a big battery?

Many of these things -- like a rear/backup camera -- are now required in new cars by national transport safety agencies. Backup cameras are also great, it’s like: sure I learned how to reverse park/parallel park, but having a camera help me do it? I would love the help! Those blindspot sensors? Also great!

So anyway, new cars in the EU are required to have a(n advanced) system that will produce a driver distraction warning if the driver is, uh, distracted. It’s commonly implemented by sticking an infrared camera somewhere in front of the camera (sometimes in the steering wheel column, I guess) that’s aimed at the driver’s face and telling the magic sand that can do maths to tell the speaker to make a noise if the maths says the driver’s face looks like it is doing the dangerous thing.

(A reminder: this newsletter is mostly reckons and me thinking out loud.)

Look, the maths that you make the magic sand do to tell whether the driver is doing the bad dangerous thing or not isn’t, like, hard. I mean, it’s hard to do accurately and consistently, but that’s because the world we live in is super complicated and you can’t ever do anything perfectly anyway. That’s how you get situations where my wife calls me and is all “hey, is the car broken, it keeps telling me that I’m not paying attention to the road” and I’m all “oh the driver attention system is probably not working” and then we have a think about it and then we’re like “oh! it’s probably the polarized sunglasses!”, but not just the polarized sunglasses, probably a unique confluence of the sunglasses and the orientation of the car and the fact that it’s around sunset?

You may be wondering why I’m even writing about this in the first place, and I will tell you why (the title of this section is a clue): it’s because being alive right now is super complicated and we have to make far-reaching decisions about technology in an age of implicit capitalism!

One of the places where I saw this issue of advanced driver distraction warning systems discussed was over on the orange place¹ where there’s a predictable mix of outrage over heavy-handed state intrusion into peoples’ god-given right to pilot vehicles at speed as manually as possible, and also at the fact that these systems keep going off on one at false positives.

So what would you do? How would you balance these priorities? Let me just list some bullet points to get them out of the way:

• Yes, the user experience in cars is terrible, we don’t have to rehash the whole physical controls vs screens things
• There exist entire fields of study around cockpit design and information overload and human computer interaction, at least one of which I’ve temporarily forgotten the name of (oh, it’s human factor design) and what good and less-good ways are of managing pilot/driver attention
• Let’s not even talk about the differing standards as to when one is allowed to pilot the car in the first place
• Or the fact that you just have to be assessed once
• I’m not interested either in any debate around “well cars are just dangerous anyway, we shouldn’t have them, we should have bikes and public transit”
• Or any debates around autonomous vehicles and how they are/can be safer than human drivers (and also for passengers) but have you considered the cost of what it took to get there and the (unrecognized) labor those systems are built on?

I mean yes, all of that. But some polity somewhere has decided -- through its democratic process, which involves people voting to choose representatives and those representatives setting policy to be then enacted and delivered by a (hopefully competent) civil service through rulemaking, regulation, and enforcement -- that in general, it’d be aces if we had less distracted driving and if there’s a technical means to accomplish that then go for it.

Rachel Coldicutt wrote a while back (and I’m very sorry that I can’t find the reference right now?) about I think the difficulty in having a political class that is well-informed about technology so that at the very least it doesn’t stick its foot in the mouth when making comments to journalists and at best doesn’t go around making shitty legislation that hinders more than helps, and that part of this is unfortunately (I think? I’m really sorry if I’m misremembering!) down to the fact that technology is really fucking complicated and that the people who have the time to advise are the ones with the money, who are invariably industry lobbyists.

(For political class let’s also include the civil service and not just elected representatives and their staff)

What would you do? Sure technically you can have a camera that watches a face and looks for patterns that may indicate inattentiveness, but again, those are merely patterns or the signs of inattentiveness. They are not inattentiveness itself.

(Inattentiveness here isn’t just scrolling on your phone, which some might see as a conscious intentional act that fair enough you could prosecute, but also driving-whilst-drowsy which in my head has a different bar in which one should have a responsibility to not drive when they are at risk of becoming drowsy. Anyway!)

But that camera will get it wrong sometimes. So now you need to consider the plethora of driving safety assistance systems as a whole. Presumably you don’t want to legislate for that otherwise you end up with unhelpful legislation or material alongside regulations that says something like “the system shouldn’t overload a driver with unnecessary warnings to the extent that the warnings themselves are distractions” and one might say well duh and then there will be at least one countering story of “well you wouldn’t believe what people end up doing if you don’t specifically tell them not to do the stupid thing”.

So what would you do? You’re a product manager at a car company and you’re in charge of maybe just this advanced distracted driver warning system. Do you get to say to everyone else on the advanced driver systems “hey hold on a bit, looks like we’re going to overload the driver here?”. Someone gets to decide the false positive threshold, and that consideration is ultimately going to come down to “how much money can we spend on this software” which is part of “how much money are we willing to spend on this system”. And before you know it, you’re off wondering about other policy interventions like “well what if we mandated that driver warning systems -- which are clearly beneficial to society as a whole! -- were made open source or free-to-use like the way Volvo did with seatbelts” and that’s a completely different kettle of fish but again, it’s kind of one if you’re thinking in the big picture about “well how do you want society to work in the 2020s?” (and the answer there is clearly: well better than it does now thank you very much, which is not to say that the people working very hard are not working very hard; you are doing a good job thank you).

I mean I think on surface thoughts that it would be great to socialize and spread the cost of developing and maintaining advanced driver safety systems. But you know that would not be a free market approach that would let firms compete with each other to provide the best driver warning systems?

OK, but then there’s another axis. Those cameras are cameras. Do you have regulations about the data that’s gathered by the cameras? Do the cameras keep the data? Do you mandate how long the data must be kept, because you never know when it would be handy to see who was in a car at a particular time. Or do you mandate that the camera data is essentially just a realtime system that doesn’t remember anything, a sort of amnesiac magic sand. But then what if insurance companies start lobbying and say: well hang on a second, we could offer lower insurance rates if we were able to take into account how often a driver is driving in a (purportedly) distracted state. But then if you do that, what do you do about people who maintain that the sensor was wrong, that they weren’t distracted, but now they’ve been dropped by all the insurers and can’t get competitive rates? And all this is because you wanted to reduce the rate of distracted driving?

This is a very long way of pointing out the tension between legibility and, say, safety. And that you can drive a truck through any definition of safety, because while you can talk about safety in terms of quantitive risk and put numbers on it, safety is also an inherently human emotion that understandably is finely tuned to trigger your nervous system, and your nervous system’s ability to deal with perceived safety isn’t absolute and can change over time (as it should! I mean, seems like that’s been a fairly successful evolutionary adaptation so far!)

And I didn’t even get to pointing out legibility to whom or what. I mean I guess I did in terms of your behavior in the hands of insurers who will (rationally, I guess) want to get their hands on as much data as possible in order to, like, have a business? And there’s legibility in terms of to the state who, like, societies have generally entrusted to do things like “make sure not too many people get killed by drivers”. And I didn’t even get into who is legible either! Does the software running behind the warning system discriminate? Does it, like certain facial recognition systems, work differently depending upon the subject’s skin colour?

So: what would you do?

This was just about “we have technology that can reduce the number of distracted driving accidents” and then “well how would we do it then?”. What would you do, and how would you do it?

OK, that’s it for today.

How have you been doing?

(Oh, and remember: you can still sign up to come to How People Work, Live! this Friday July 10 at 9am PT / 12pm ET / 5pm London)