s5e11: Things That Have Caught My Attention
0.0 Station Ident
It's a Friday. I still haven't seen Avengers: Infinity War yet (this movie, at least, is one that demands spousal solidarity).
1.0 Some Things That Have Caught, Etc., Etc.
* Software continues to eat the world, world continues to be reminded that software isn't automatically good and creates a whole new bunch of interdependencies and room for ambiguity in behavior and expectations. In this instance, when someone sold their Volkswagen back to a dealer and the owner retained access to the onboard, connected telemetry system *after* she had sold it[0].
* The other day, Twitter disclosed that (all?) users' plaintext passwords had been exposed in its logging[1]. This happened pretty much the day after Github disclosed that they had accidentally written plaintext user passwords to their logging too. What made this *drama* though was when Twitter's CTO (unadvisedly) reminded everyone that Twitter wasn't under a duty to disclose this information but was doing so anyway (presumably because, as we're all learning, companies Value Our Privacy Very Much as we get closer to GDPR-day). The general theory here is that Europe's GDPR regulations (which have been on display for the last few years at the planning office, etc.) are about to kick into force and Large Internet Corporations legal and compliance teams are, well, doing their job and making sure that the relevant regulations are being Complied With. So, you get a bunch of companies doing very specific auditing due to impending GDPRpday that might, say, expose inadvertent logging of plaintext passwords[3]. My response here is a very high-level: oh look what happens when you've got regulations that *law and a reasonable enough stick and threat of enforcement* actually require organizations to follow through with security practices like "knowing what you're logging" or, at the very least, "checking to see if you know what you're logging every now and then".
* Let's just accept that I was reading up on a specific Star Trek: The Next Generation episode and happened to find that Data made the claim that "there hasn't been a complete systems failure on a starship for over seventy-nine years"[4] ("Evolution", Season 3, Episode 1). This is a big deal! I mean, we know that the technology isn't particularly reliable on Federation Starships because of the humanity's Hold My Beer gene and propensity for messing with things, but really! Seventy nine years is not that long a time! There are probably people saying, "Oh yeah, remember that one time Starfleet Engineering pushed that shipOS release and it bricked the entire fleet" and the Vulcans are all going "seriously, what is it going to take for you guys to adopt formal verification of software as a species" and the answer is "never" because Wesley gets to check out code from 1701-D's main branch and Geordie accepts his pull-request.
* I am not watching Westworld Season 2 for reasons, but (as the saying goes), some of my best friends watch Westworld. I did watch season 1, to a certain degree. I am aware that Westworld's storytelling technique involves multiple ambiguous chronologies and the show and individual episodes move fluidly between different time periods. Does this remind you of anything else? That's right! Non-chronological interest-based algorithmic timelines and feeds![5] This is just a connection I've made - but I remember when Hollywood movies started adopting videogame conventions and feeling like they were heavily influenced by videogame tropes - Battle: Los Angeles (2011)[6] especially felt like it was watching someone play a scripted single-player Call of Duty game. So I'm wondering if, say, post Lost (2004)[7] with its free movement between multiple time periods and timelines, *combined* with increasing everyday familiarity with non-chronologically-ordered feeds is making it easier to put non-linear storytelling in front of them and have them freak out like when Memento (2000)[8] (also a Nolan!) came out.
* Mat Honan reminded me[9] that Amazon have made an Alexa for Kids (during the course of which Tom Simonite had a great story about Alexa diligently and non-plussedly educating a group of preschoolers about the history of FARC after misunderstanding their requests for farts[10]) and Honan has a great article about it[11]. There are now enough Alexa (plural?) out there that the phenomenon of "the funny things kids say to Alexa" is pretty well documented as well as the earlier "Alexa is teaching my kid to be rude" observation. This isn't to say that Amazon haven't done *any* work thinking about how Alexa works in a kid context (Honan's article shows that they've demonstrably thought about how Alexa might work and that they've made changes to the product to accommodate children as a specific class of user) but the overwhelming impression I had after reading Honan's piece was that, as a parent, I still don't think Amazon haven't gone far enough in making Alexa kid-friendly.
They've made some executive decisions like coming down hard on curation versus algorithmic selection of content (see James Bridle's excellent earlier essay on YouTube, that something is wrong on the internet[12] and recent coverage of YouTube Kids' content selection method still finding ways to recommend, shall we say, videos espousing extreme views). And Amazon have addressed one of the core reported issues of having an Alexa in the house (the rudeness) by designing in support for a "magic word" Easter Egg that will reward kids for saying "please". But that seems rather tactical and dealing with a specific issue and not, well, foundational. I think that the foundational issue is something more like this: parenting is a *very* personal subject. As I have become a parent, I have discovered (and validated through experimental data) that parents have very specific views about how to do things! Many parents do not agree with each other! Parents who agree with each other on some things do not agree on other things! In families where there are two parents there is much scope for disagreement on both desired outcome and method!
All of which is to say is that the current design, architecture and strategy of Alexa for Kids indicates one sort of one-size-fits-all method and that there's not much room for parental customization. This isn't to say that Amazon are actively preventing it and might not add it down the line - it's just that it doesn't really exist right now. Honan's got a great point that:
"[For example,] take the magic word we mentioned earlier. There is no universal norm when it comes to what’s polite or rude. Manners vary by family, culture, and even region. While “yes, sir” may be de rigueur in Alabama, for example, it might be viewed as an element of the patriarchy in parts of California."
Some parents may have very specific views on how they want to teach their kids to be polite. This kind of thinking leads me down the path of: well, are we imagining a world where Alexa or something like it is a sort of universal basic babysitter, with default norms and those who can get, well, customization? Or what someone else might call: attentive, individualized parenting?
When Alexa for Kids came out, I did about 10 seconds' worth of thinking and, based on how Alexa gets used in our house (two parents, a five year old and a 19 month old) and how our preschooler is behaving, I was pretty convinced that I'm in no way ready or willing to leave him alone with an Alexa for Kids in his room. My family is, in what some might see as that tedious middle class way, pretty strict about the amount of screen time our kids get (unsupervised and supervised) and suffice it to say that there's considerable difference of opinion between my wife and myself on what we're both comfortable with and at what point what level of exposure or usage might be appropriate.
And here's where I reinforce that point again: are you okay with leaving your kids with a default babysitter, or are you the kind of person who has opinions about how you want your babysitter to act with your kids? (Yes, I imagine people reading this and clutching their pearls at the mere *thought* of an Alexa "babysitting" a kid but need I remind you that books are a technological object too and the issue here is in the degree of interactivity and access). At least with a babysitter I can set some parameters and I've got an idea of how the babysitter might interact with the kids because, well, that's part of the babysitter screening process.
This stuff is really interesting (can you tell) and not least of which because recently on Hacker News someone had posted about how they'd been attempting to implement The Young Lady's Illustrated Primer[13] the networked computer mediated Victorian-style governess/tutor from Neal Stephenson's Diamond Age. (Relatedly, I see that in the tech industry's process of Implement Everything That Neal Stephenson Wrote About, we've now reached the Diamond Age phase). (Relatedly relatedly, it's somewhat amusing to see that even in the Hacker News conversation there are differences of opinion as to the lesson that Stephenson was attempting to impart with the Primer and the relative benefits of the different versions as delivered to the three girls).
I'd also been rewatching Douglas Adam's Hyperland[14, 15], his 1990 BBC Two documentary on hypermedia and hypertext. (Aside: the documentary includes the concept of "micons" - moving icons, a bit like what if button interface elements but also gifs, to which the answer is: they'd be really annoying; and second, that there's an interview with a Media Lab group who are in one part quite pleased about how diverse they are and with the benefit of hindsight and wokeness, we are able to say: yes, well done, you had some white women in your group). The fun ven diagram party between Hyperland, The Young Lady's Illustrated Primer and Amazon's obvious dream of Alexa for Kids is that of an apparently-intelligent software agent that learns a child's needs and wants, meets them and helps a child to grow to their potential. I had forgotten how in Hyperland Adams was convinced (as many of us were, at the time) of the promise of intelligent agents. Those agents would (as Tom Baker says) "work tirelessly on our behalf" and unequivocally be working for us, which I have to admit I laughed out loud when I heard it again in 2018 against the backdrop of "well, you're the product, innit" and our current trend of talking about How The Internet Actually Turned Out. Anyway: these intelligent agents would passively and actively learn our interests and proactively act on our behalf. In reading Adams' biography[16], he retells a story of an intelligent agent that would know what your music tastes were, would know where you were because of your GPS location, would be accessible on a mobile device, and let you know in advance that the little sleepy village you were driving into had that rare LP from that band you love that you don't own. (We are not quite there yet, but we can certainly tell eBay to let us know when someone's got one for sale and mobile device app notifications have certainly learned how to be irritatingly location-based in the pursuit of transactions).
Intelligent Voice Assistants are a bit like the Intelligent Agents all over again. Back in the 90s (hello, Magic Cap fanboy here), we were thinking the agent software would literally move from host to host in some sort of fever dream of Sun's where Java needed even more of a point. Whether in our mobile devices or because they're hooked up to one of the largest retailers on the planet, they learn what we need (apparently), what we want (apparently) and are increasingly trying to get us to buy it before we want it. The thought of my children growing up with Alexa in the way that Charlotte, Nell and Fiona did with the various Primers isn't necessarily one that I thought I'd have to deal with, and I say this as someone who's had relatively serious discussions with friends and colleagues about how you might go about making a Primer.
--
Anyway, those are some of the things that I found interesting.
How are you?
Dan
[0] Former VW owner discovered digital access to her car months after it was sold - The Verge
[1] Twitter advising all 330 million users to change passwords after bug exposed them in plain text - The Verge
[2] GitHub accidentally writes clear text user passwords to logs | Hacker News
[3] م. محمد الدوب on Twitter: "GDPR is making people check their logs and it's hilarious so far with all these companies finally actually checking what they log.… "
[4] Evolution (episode) | Memory Alpha | FANDOM powered by Wikia
[5] dan hon on Twitter: "With storylines from the past, present and future mixed together does Westworld’s storytelling technique remind you of anything? That’s right. Algorithmic interest-based non-chronological timelines and feeds."
[6] Battle: Los Angeles - Wikipedia
[7] Lost (TV series) - Wikipedia
[8] Memento (film) - Wikipedia
[9] Mat Honan 📰 on Twitter: "I wrote something about this recently that you might enjoy https://t.co/l1dXc3TXvg… "
[10] Tom Simonite on Twitter: "I recently witnessed Alexa educate a group of pre-schoolers about the guerrilla movement FARC after it misunderstood their request for farts.… https://t.co/hfEQXOlthE"
[11] Amazon Created A Version Of Alexa Just For Kids
[12] Something is wrong on the internet – James Bridle – Medium
[13] Designing a 'Young Lady’s Illustrated Primer' from Diamond Age | Hacker News
[14] Hyperland - Wikipedia
[15] Douglas Adams Hyperland - YouTube
[16] Amazon.com: Wish You Were Here: The Official Biography of Douglas Adams eBook: Nick Webb: Kindle Store