s08e02: Software from trauma, or, fingers crossed the only blockchain episode of 2020
0.0 Context setting
I started writing this on Thursday 9 January 2020:
Thursday 9 January 2020 and it’s very cold in Portland. It’s 37 fahrenheit, which is still a silly system temperature system at the low end but perfectly reasonable at the high end. Speaking of units of measurement, I recently read some military SF as an experiment having avoided the genre for reasons of boredom and aversion to political beliefs, but also learned about the unit of grains to measure the mass of a bullet? Google says 437.5 grains equals one ounce and I’m sorry, this is just too funny for me. It’s so messed up.
1.0 The things that caught, etc…
I’m sorry, I wrote about blockchain. Hopefully the next part makes it up for you.
1.1 But blockchain
My parents came out for the holidays and there comes a time in someone’s life when their father asks them to explain what the deal is with blockchain. The backstory here is that my dad helps assess applications for funding for an organization that’s a bit like the National Science Foundation, but in the UK. Lately, as some of you might fear, an increasing number of applications have included the word “blockchain” in them. So, he asked: what’s the deal? He’d researched the technology and, well, couldn’t quite figure out what the big fuss was about. What exactly was blockchain adding?
His background is in manufacturing engineering, so a fairly typical use case might be something like “using blockchain in supply chain management”. You might, for example, have a pretty stringent requirement for components and materials in a satellite and something terrible (or, at least, not intended) might happen if the components not turn out to be what you thought they were. Like an aluminium joint designed to make sure nosecone fairings separate when you’re launching a satellite.
I’m simplifying here, but when someone says that blockchain would help prevent this, what they actually mean is that it won’t prevent this at all.
In this particular case, you’ve got a bunch of suppliers handing off parts to each other and then someone needs to trust that they got what they bought, that the parts meet the required specifications. Suppliers sign off on tests and write them down and you have a bunch of paper floating around that says that people did what they said they would do, and that the results are what they need to be. On paper.
Now, paper’s reasonably easy to forge or alter. It’s been around for ages, and a sufficiently motivated person might be able to change what’s written down. This happens in, say, pharmaceutical manufacture.
Ah, say the blockchain people. But what if I had a special kind of paper where, once you something is written on it, you can’t go back and change it? No erasures, no take-backsies? No trying to print a duplicate one and burning the original form? Well, that’d certainly be a bit better than the paper that we have right now. And, they say, what if everyone had to use the same bit of paper? That way, we only add things on to the end of the piece of paper, and everyone can see what everyone else has written.
Look:
“Sounds great,” you say, and IBM is all ready to sell you a bunch of upgraded paper.
And then you think about it for a little bit.
“Wait,” you say.
“So I believe you, and I can trust the paper a bit more after I’ve spent a few hundred million dollars and persuaded everyone in my supply chain to change how they do things. But how do I trust what’s written on the paper?”
IBM (say) goes quiet a bit and makes a hmmmmm noise and distracts you with a white paper.
Surprise: you can’t trust what’s written on the paper.
“No, no. You can totally trust what’s written on the paper. It’s special paper, see?”
“But who’s writing the stuff on the paper? Who’s entering it?”
“Oh, people are.”
“Wait, the same people who used to occasionally commit fraud and lie on regular paper?”
“Yes, well-”
“How does your new special paper stop them from lying? Does it come with some sort of singular rope-based psychoactive that compels anyone caught by it to tell the truth? Because I don’t think that scales.”
“Well, no, we can’t stop people from lying but you can totally trust the paper. Hang on, just give us a minute and we’ll get back to you.”
You wait a minute.
“Okay. Hear us out. Machine learning-”
“Machine what?”
“Deep learning. It’s a kind of-”
“OK, but how do I trust these deep machines?”
“Well, they’re very advanced and they run in the cloud and they’ll run automated testing to make sure the components you’re ordering meet the specifications, and we can trust those machine learning deep AI algorithms to write on the special paper.” The team presenting to you is collectively sweating and the enthusiastic PowerPoint in the background is stuck on a slide that says something like “TRUST 2.0” with a photograph of a black man businessman and a hispanic woman shaking hands.
“But,” you say.
The head consultant just stares at you, but they’re not as good at staring as Paddington Bear.
I get it. There are certainly situations where you need to be sure that the documentation has not been tampered with, but if you’re not paying attention, you might not realize that claims as to trust are specifically about claims as to trusting the documentation, and not the underlying action or property. We live in a human-mediated world, and ultimately, you’re likely trying to make sure that you can trust other humans. If the suggestion is that you can increase trust in a system by removing untrustworthy humans and replacing them with magical algorithms, then you have to ask, well, who’s writing those algorithms and are they employed by otherwise motivated executives at companies like Boeing or Volkswagen whose goals are not quite aligned enough with yours?
It’s a lot of money for more secure paper when you probably don’t need more secure paper, you need a better aligned environment with clearer communication, management and incentives. But, you know, that’s harder than implementing a multi-billion dollar upgrade of your supply chain technology infrastructure, and if that doesn’t work you can always blame your consultant outsourced technology provider.
I was out at a dinner last week and part of it involved a very passionate (ie: very loud and quite emphatic agreeable screaming) that the kind of people who are really, really into blockchain (especially in the financial side of things) from an ideological perspective more often than not appear to be severely traumatized humans, whose retreat into blockchain is an attempt to construct a world where they don’t have to rely on or trust other humans because of some horrible event in their past and a lack of therapy to process and move on.
It is very hard to disagree with this position. Blockchain makes sense because you can say that you can mathematically prove the paper but conveniently avoids (or on purpose, from a psychological point of view) the fact that other people in the world exist, and you have to trust what they put on the paper, too. If only, think some damaged people, other people just didn’t exist.
PS. I’ve met some great people from IBM, but right now I’m not a fan of the people producing the blockchain marketing I see in my feeds.
1.2 The shorter things
It’s interesting how much having paying subscribers has changed how I think about this newsletter.
For one, for all my protestations that I write only for myself, I’m acutely aware that it’s been a week since the last episode and that I can’t stop writing.
And for another, I keep thinking that I should use emoji in these parts to delimit each point because, well, that’s what the other cool newsletters do.
Anyway, let’s try and see if it doesn’t make feel too much like a shill:
🎧 On some Apple phones, iOS 13 will let you share your bluetooth audio with another person, if you’re all using first-party headphones (ie Airpod(s), Beats) from the Apple ecosystem. This manifests by displaying a modal dialog asking you if you’d like to share with someone, and the heuristic is roughly that it should only show up with someone who is very close to you. Like, say, when I’m on a standing room only bus and I get a prompt to share with someone’s Beats. This feels a lot like the practice of invariably younger people AirDropping (wireless file sharing) things to invariably older peoples’ phones, only with music.
🖥 Via Simon Wistow and the DisplayLink (the current standard for connecting to a monitor that isn’t HDMI and the one Apple computers don’t use) support forum, did you know that gas lift chairs can create an electromagentic interference spike when humans use them as… intended as chairs? There’s a paper and everything, dating back to 1993 which, as these things go, is very cute.
🔑 I used to be (and still am, I think) of the opinion that in general, things having the ability to talk to each other and some level of smartness would be a good thing, but I didn’t bank on the innately human ability to fuck things up either on purpose or out of incompetence. The kind of things that I do like are things that are smart but don’t necessarily have screens and are still a bit like the previous versions of the same physical thing. Which is why I really, really like the Netatmo HomeKit smart lock, which needs a better name. The Verge’s writeup sums up my feelings pretty well: it’s a lock that uses key-shaped keys, but the keys use NFC so they will work in any other lock, but are software-coded to particular locks.
🇫🇷 It is not a surprise that Netatmo is French. I feel like the French sensibility toward internet things is a great counter to the American one (broadly, don’t collect all the data, don’t put screens in all the things, make things that are stylish and subtle and integrate into the existing environment). I’ve had a fondness like many others for things like the Nabaztag that I feel Google and Amazon are still grasping toward with their ambient information always listening displays, but aren’t viable for them unless they are a means of procuring more user data, whereas the Nabaztag is (was) pretty much an output-only device.
⌚︎ Another great company in this area is Withings, whom I already quite liked before they came out with their first smartwatches that don’t have any displays in them. I’ve written about them before: watches that look like regular watches, analog faces, no screens, but all the MEMs-based (or whatever) motion sensors and bluetooth and so on that mean they can talk to your phone, but they still look like a regular watch and don’t grab your attention. They reflect a very different philosophy and if I were to stereotype, it’s what a smartwatch might look like in a country that’s very happy for you to take a bunch of vacation, versus a country where if you take vacation you’re a loser and you’re not committed to your job. Withings have recently come out with an updated smartwatch that can detect afib, like the newer Apple watches, as well as sleep apnea, because it has an spO2 sensor. This smartwatch does have a screen but I’ll allow it because it is small(er) and discreet and not the entire face of the watch.
🖲 The underlying theme here is of one of just enough internet, wonderfully coined and put forward by Rachel Coldicutt, formerly of doteveryone. There’s so much to unpack from just the phrase, but most recently, I started thinking in a tweet about:
just enough personal health monitoring (there’s so much here, too, I think)
just enough screens
just enough bureaucracy
The just enough bureaucracy thought of course led to just enough automation, and yes, all of this is subject to the caveat that you don’t know what just enough means unless you’re sure about what your goal is. Just enough for what?
OK, I said these were supposed to be short ones, but here’s some really short ones now:
💊 I randomly tweeted about DARE, the 1980s American program to educate teens about drug abuse, but imagining a contemporary program to educate people about social media disinformation. Only I forgot to make clear that I was imagining a program with the equivalent success of DARE, which is to say, not very much, because such a program would be difficult to get right. (Or: it’s easy to imagine legislators suggesting or requiring a program like DARE, but for the implementation to be horrific and also potentially backfire). [Aside, apparently I can’t make these short]
👓 The US military is getting ready for replacing its air tankers with a new one, and predictably the new one which has technology up the wazoo, is not working very well. Or as well as you’d like it to. The interesting bit to me is that the Boeing KC-46 uses digital displays for the boom operators (the ones who extend the phallic fuel and life-giving extension from the tanker for other planes to suckle from) so that it can operate in near/total darkness which, as any mother who has breastfed at night might tell you, is a bitch. Latching can be hard. Anyway, the technological solution to this need to refuel at night involves stereoscopic 3D displays and, presumably, active 3D glasses. But they don’t work that well because, well, optics and lenses and image processing software. Thankfully, you can probably fix this all in software, which is something Boeing is hopefully good at. Too soon?
🏛 A couple of interesting posts from Bruce Perens, most recently on ethical [software] licenses and earlier on a hippocratic [software] licence. Perens concludes that software licenses are pretty ineffective tools for social change for reasons that are both structural and practical, but the kicker for me is his (accurate, in my opinion) observation that if you have a social goal, perhaps your best bet is to engage with your system of government to enact better laws, rather than reaching for the instrument that is closest to you and easiest for you to control, i.e. the license terms you choose for your software (e.g. “you’re not licensed to use this if you don’t support vaccinations”). This opens up the entire can of worms around how a population can and should be involved in government, which is not a software problem, but also not a problem that software cannot help with, but only if software stops believing that it can solve everything and makes everything better.
👮♂️ I saw an ad on Amazon’s front page touting their Ring Video Doorbell 2 through a suggested user interaction, “Alexa, show me the front door.” Left unsaid, though, was that depending on who you are, you might get a different result. Here’s some we can imagine:
Local advertiser: “Alexa, show me the front door.”
Alexa: OK, I can show you 10,000 front doors in the Portland Metro area. What would you like to see?
No? What about:
Law enforcement: “Alexa, show me the front door.”
Alexa: “OK, I have access to 2.6 million front doors. Which front door would you like to see, in what area, and over what time period?”
Or:
Data science marketer: “Alexa, show me the front door.”
Alexa: “OK, I have six hundred and thirty thousand minutes of video with person segmentation and predicted age of participant occurring between 4pm and 9pm local time on October 31, 2019. What would you like to see?”
Or:
Data science marketer: “Alexa, show me the front door.”
Alexa: “OK, I have package delivery frequency records and predicted package source based on vehicle identification for 1.5 million front doors over the 6 month period. What would you like to see?”
This doesn’t feel quite right, that there’s the ability for any number of people to examine and query data based not on just my front door, but, well, all of them. But, you know. Progress!
I don’t know about you, but it’s been incredibly difficult sliding back into work. I keep trying to approach it like the Electric Monk’s horse, creeping up to it without it noticing me. It’s not like I haven’t got things done, it’s just that doing so in what feels like an effective way for a long stretch of time has felt the last week like trying to pick up a perfectly frictionless surface.
Which is to say that I am finishing this newsletter episode on Sunday, three days after I started writing it, out of a mixture of anxiety and sheer bloody-mindedness. And, I note, the kindness of a partner who knows that getting this done will help me and is protecting time and space so I can get it done.
Happy new year, still! It’s not February yet, so I feel like we’re still allowed to say this.
What did you think of this one? Interesting? Not interesting? If you did find it interesting or useful, it’s always nice to see people talking about it or forwarding it on. Not that I’m doing it for the ego boost, right? I mean, not only that. Obviously Bret Stephens is the only person who googles himself.
And yes, I continue to love notes, even if they are just to say “hi”.
Best,
Dan