Thursday, 19th October, 2023, sitting on the sofa, in the family room, in the basement, listening to music1 on the frankly ridiculous in-ceiling speaker system we inherited with the house, in Portland, Oregon, USA.
Today's things that caught my attention are pretty much the brief notes I took during my coffee chat with Paul Rissen2 (Hi, Paul!) and Adam W. Flynn.
First, something that caught my attention when we were talking about startups and their preoccupation with scaling. I like to say that scaling is considered harmful, which is to say that there are certain types of scaling -- namely the Californian Ideology kind -- that are harmful. Toxic, even. Consumptive, zero-sum competitive, very much eating. The up-and-to-the-right growth-focussed scaling.
There's other scaling! I though of the phrase scaling through time, which honestly feels like a wanky way of saying persisting or trying hard for a long time. I suppose the distinction here is a resistance to using shortcuts, like flooding the area with money. Or, more possibly, that scaling through time means sustainable growth and persistency, institutionalization. Scaling through time as generational. Which brings to mind Deb Chachra's new book, How Infrastructure Works3.
... or perhaps scaling on another axis. Not through time, not scaling longitudinally, but scaling... through depth? Another (wanky?) way to think about this might be "extending downward/upward through the Wardley Value Chain"; figuring out to what extent vertical integration might be needed for what you're trying to accomplish.
This, because of my recent exposure through my work with Rewiring America and stories about how complex the Big Greening is from just a residential point of view: local permitting and inspection regimes, distribution, installation, incentives, rebates, financing and so on. Scaling the navigation of seams. Is scaling even the right word? I suppose in a different sense, in the terms of climbing or surmounting, yes, myriad bureaucracies, but not in terms of growth. Perhaps in terms of making systemic change?
I wrote this title and I'm not even sure faster is the right adjective here.
Rissen and I were talking about tools for writing. Tools for narratives, and more specifically, interactive narratives. Interactive storytelling. Which on the one hand, interesting! And on the other hand, a weary sigh born from decades of experience.
So, some notes.
Scripto[^scripto], which I've written about before I'm sure, and covered by The New Yorker5. Scripto, the online scriptwriting service born out of The Colbert Report, and as I understand it the New, And Better Than Final Draft thing for working together quickly. You can tell where Scripto came from and how it's spread -- the standout clients/customers are Stephen Colbert's Late Show, John Oliver's Last Week Tonight, Trevor Noah's The Daily Show. So sure, it's spread from a professional/peer group. But also I wonder about its correlation for writing at speed, not just collaboratively. So. Scripto as not a word processor but a word-collaborator, in the way that I suspect Google Docs is still half in the Word Processor world and the first tentative, successful foot into the word-collaborator tool. And entrenched, for reasons.
There's something here in writers rooms -- at least, how I understand writers rooms to work. I remember the story of how Dan Harmon would make Community, the tv show, and the story (in Wired, of all places) about his story circles and the whiteboards (not that whiteboards are anything particular to Harmon's rooms). Whiteboards instead being the collaborative, real-time, low-latency, collectively visible, physically situated work surface (uugggggh).
What would not be better than a whiteboard, but sit alongside a whiteboard? The lazy example might be the orchestra with tablets/screens/e-ink screens for music networked to the conductor: take notes, get the conductor's notes and so on.
It feels fortunate that now the WGA have figured out a deal around the use of so-called artificial intelligence, in that it's used as a tool for writers and not a substitute for writers. How might these automated text processing engines be used for and benefit writers? Off the top of my head, some ideas:
For starters, what might help with people entrusted with continuity and show bibles? Would it help, for example, to use retrieval-augmented generation or embedding with all your scripts and notes?
As an aside, I had the idea of getting a dump of the Memory Alpha Star Trek wiki, producing embeddings based on that text and using retrieval-augmented generation plus speech-to-text and text-to-speech to essentially... create a version of computer, that would answer questions about the Star Trek universe. Which feels like it could be pretty... awesome, if it worked? Or worked well enough? Because I've wanted for a while to feed my newsletter archives into something like Datasette and ask questions about what I've written.
When I worked at Wieden+Kennedy, my copywriter CD partner would occasionally moan about not being able to remember the brilliant line or idea he'd had from the other day. Or hour, to be honest. He wanted something that he could use to record those ideas, or be listening all the time. You could have that now, modulo the stupendous privacy issues.
But what if you scoped that sort of assistant and that sort of surveillance? What if a room could remember, and you'd only be able to ask questions to the room while in the room? What if a writers room was listening all the time, what if only authorized people could ask questions? Would that be helpful? You could run that experiment now: what was it we said about this character yesterday? Or, rather, you can do this already. Zoom will sit in on your meeting. It will transcribe. It will let you ask questions and it will summarize. It's perhaps more skeezy in Zoom's sense because frequently the consent just isn't there, and it's not clear. But I love the idea that there's a room that has memory, and I understand that power structures and organizational structures mean that, really, you don't have a choice: if all the meeting rooms remember, then you're just going to have to go in them.
(Which also raises compliance issues! I don't even want to get into compliance issues. "This room forgets everything that happened in it after 72 hours". Or, even "this room has ears and a mouth", heh, or "this room has ears and can type")
I wrote above about how these language calculators and word-generation-calculators might be useful in continuity.
Professional indexing is a job. Indexes are super useful! They help when there's a book and you want to look up a specific topic!
It would be interesting if ebooks came with embeddings and you were able to ask questions about them. It would be interesting if ebook readers were able to do retrieval-augmented generation on the text of the book you bought.
Note: this isn't generating new text. It's not writing in the style of the author. It's a tool - sort of a new/different way of summarizing or providing context. In an automated way. You don't have to wrap it in personality or pretend it's human, but yes, the interface can totally be natural language.
I would like to see that. I wonder if I should talk to Cory Doctorow about running an experiment with it :)
... and then there was my coffee chat with Adam W. Flynn, continuing my conversation about how LLMs and other so-called artificial intelligences can be used by people.
There was this concept of caring and relationship management. When does a tool for maintaining a relationship become skeezy?
My wife's birthday in my calendar and a reminder isn't skeezy. In a calendar on my phone and with reminders? Not skeezy.
Carnegie's techniques in How to Win Friends & Influence People can easily be described as a pattern, a recipe, and these days, as an algorithm. Carnegie has always felt skeezy to me.
Why? Why so icky?
Thinking aloud, it's about intent. And in Carnegie's case, it's about the framing: how to win friends, as if friendship is a competition, as if you've got to collect them all, as if there's winning and losing, and instead not positioned as how to maintain or care for relationships that you value.
There's something here in authenticity and scaling, too. Customer relationship management in some ways attempts to care-at-scale in order to sell. Intent might be on a spectrum -- selling to genuinely help a customer, versus cold pressure selling. Is it skeezy to keep notes on fifty, a hundred, more customers? Prospects? I think so. It's transactional, it can't be anything but. It isn't an authentic relationship.
Flynn told me a story of someone who had all these tactics about dating, one of which included "go and be interested in the things your date is interested in", which just strikes me, again, as inauthentic. Surely you want to be interested in the things you are interested in, and not force it? Otherwise the entire exercise is framed as such, as something to be won or lost. The intent there isn't connection. And anyway, if you do end up faking interest in something in order to find and win someone, I guarantee that over time, if you don't actually develop a genuine interest in that thing, then you're going to bank and accrete resentment.
We came across this because I was talking about the parts of my work that involve understanding clients better. Understanding their motivations, fears, what would make them successful as a way to understanding their actions and working with them.
So then, the horrible idea: an Eliza for coaching you before a meeting. What is it that you want to achieve? With whom? How can you be effective in your meeting with this person or these people? What do you need to remember? Is this terrible? Is this just an aid or a crutch in the way that you might have an aid for a disability, whether permanent or temporary? (The other part, about wrapping this in Eliza, I thought was deliciously potentially Evil, playing on our habit for empathizing with things that pretend to be and sound human and totally going for them.)
Then the point became whether you could be authentic at scale. And that the question some people might have is: how can I be authentic at scale? To which my flip answer would be: by not being authentic at scale, dum-dum. It's the wrong question. It brings to mind competitive caring, thanks to the economic context we're in.
At this point, I start thinking about prosthesis for care. How might we make caring easier? But that's not the answer! Down that route there's a red alarm and a klaxon because that's the techno-solutionist approach: how can we use technology and tools to solve a societal, inter-personal, relational problem.
But I would like to think there's a difference in expecting a technology to solve a problem and instead be a tool or an aid. The difference is in blindly assuming or pushing that technology is always good, and I can point to the tired example of forums enabling people to organize to be abusive versus forums enabling people to organize to recognize chronic health symptoms and support each other, to call for and achieve change in the healthcare-industrial-medical-commercial-academic complex. I'm with Doctorow in believing and being optimistic about technology as a tool for organizing and for amplifying the intent and actions of care, of responsibility.
I also, though, remember something I remember from a book that I read recently: that in many cases, intent doesn't matter. The action or the result does. Well, perhaps it isn't as binary as that, but what matters may be firmly and strongly rooted in the effect and outcome rather than intent.
So: tools better aimed at intent-outcome matches, better aimed at framing. Better aimed at... useful connection? Caring connections? It's here where the Dunbar number, however discredited, is something that feels like it should be right, that it wants to be right. And from there it's a jump to the early discourse around Facebook and social networks: the desire for connection, the automation of social grooming with poking and likes, the desire to scale social grooming and connections. But again, the context collapse of those relationships; I wrote before about attempts to deal with different gradients of friendship, of relationship, that social networks have been doomed to reinvent circles or other methods of organizing people who matter, in different ways, in different contexts to you.
I maintain that the collapse doesn't work, that it's harmful, that sticking everyone in the same feed comes with some pretty bad effects. At least, there's an example of that in Facebook, and yet there's a counter-example in the mixing and meeting-new-people of Twitter. But again, the framing: Twitter was about following and not friends.
Like I've written before, these are different spaces. LinkedIn is the drinks at the conference. It is a space with a lot of people, networking. It is a different space. Instagram is a different space, the apps are different, they look different (at least, still). You know that you are in one space and not the other, and perhaps one of the smartest (sigh) things Facebook did was to not integrate Instagram and WhatsApp into its parent platform, to keep them separate, knowing and acknowledging that people need and want different contexts.
It's here that Flynn introduced me to the concept of intimacy gradients in architecture, which is something I'm almost embarrassed to not have known about, but instead will take the "wow! I get to learn about this today!" perspective. The porch is different than the foyer is different than the living room is different than the dining room, the kitchen, and then finally the bedroom and bathroom. Different spaces. Social networks wanting to collapse space because collapsing space made it easier to gather data. More people in one place was always better, up and to the right.
The reason for more people in a social network was down to the dogma of Metcalfe's Law, that the utility of a network increases to the square of the number of connected users. That is a great thing! But a telecommunications network is the backbone, not the topmost layer. I think the law works for the underlying network but not necessarily the application or the presentation.
The utility for Facebook increases to the square of the connected users.
The utility for LinkedIn, for that space of professional networking might do, too. But the utility for the front-end of Facebook? For maintaining those types of relationships? It raises the question of what is Facebook for? Is it supposed to collect friends? Is it supposed to maintain relationships?
This is also why I keep coming back to Facebook Groups, as one of those distinct features of Facebook that actually implements different boundaries, that allows privacy. Facebook Groups is the living room to which people can be invited, the meeting space that can be private or semi-private, the place that makes sense for a book club or a tenant's association.
So what does that make the feed? Your own weird living room? A sort of new place? A new place where you kind of have control, but not really, because you've been guided toward adding as many people as possible? A place which smushes together both information from "people" of varying degrees of relationship, and yet also "pages" and businesses and groups and so on?
What space does this map on to? Is it the living room that also has an area for mail? For spam? Is it a living room that also has a phone in it? A photo album? Why are all these people coming in here? Because you invited them in, duh. Because, of course, Facebook is more useful the more people you're connected to?
I think this comes back to one of the other things on my mind right now: the swing of the pendulum back to (slightly) smaller, more specific spaces, more context-driven spaces.
On the one hand, I get the utility. I get that to save time, a unified inbox is useful. I get that having one place to deal with things is useful, but also the lack of interop means you don't have a choice. You can't choose the places into which you bring these relationships or this information.
Anyway. The long story short is that today was a very stimulating coffee meeting day.
One last quick thing. If I were a cool English teacher at a school then I would totally set Disco Elysium as a text for my class. It's canon material. The thing is, I'd need to set it as the text for something like two years.
I WAS GOING TO WRITE SHORTER THIS WEEK. I AM SO SORRY.
How are you?
Things That Caught My Attention is a free newsletter, and if you like it and find it useful, please consider becoming a paid supporter.
At this exact moment, the theme to Dungeons & Dragons from the film Dungeons & Dragons: Honor Among Thieves, by Borne Balfe ↩
How Infrastructure Works by Deb Chachra: 9780593086599 | PenguinRandomHouse.com: Books (archive.is), Deb Chachra, October 2023, Penguin Random House ↩