September 26, 2023

s16e02: Consensual Hallucination; Hospitals; Warm, Statistically Plausible Generic Platitudes

0.0 Context Setting

It's Tuesday, 26 September 2023 in Portland, Oregon. We are under an atmospheric river.

Sunday night I had the exciting experience of spending eight hours in the emergency department of the hospital with the mixed happy and somewhat unfulfilling conclusion that a) there was nothing emergent and b) nobody could figure out what the thing was that wasn't emergent.

I am also opening up my availability for consulting! You can find out more about what I do at www.verylittlegravitas.com, or drop me a line by replying.

1.0 Some Things That Caught My Attention

1.1 Consensual Hallucination

I saw this recent Qwantz Dinosaur Comic about social media¹ that caught my attention thanks to its tactic of being absolutely bloody-minded in describing things. See:

• Replacing the phrase "social media" with "web forum". Okay fine, feed-based, follower/following-based social media have graph relationships in the way that flat forums don't, but the fundamental point here is about people putting text on the internet for other people to read, and then replying to the text other people have posted;
• The identification of a "main character" on social media, and the thrusting-a-sword-through-someone response of "I try to stay above forum drama"; and then best of all
• the take-a-step-back and hard-to-notice-because-you're-swimming-in-it perspective of why would you ever tie the success of your personal and professional life to posts on a web forum, of all places

OK fine, it's taking an argument to an extreme. While I can point to a bunch of examples in both my life and friends and acquaintances lives of personal and professional success and growth through posts on a web forum, I think the better point here is why would I rely on just one web forum, of all the dumb things to do.

The thing is, not even that many people use Twitter compared to that other self-proclaimed global public social network Facebook (formerly known as thefacebook for rating women at Stanford). But! Over a chat on Friday night, I figured out these things about The Unbearable Uniqueness of The Web Forum Formerly Known As Twitter.

I'll use, uh, comics writers, as an example here. Here's some of "the things Twitter does", if you want to get all Professional Development For Your Design Career about it, you can think about these as Jobs to be Done. They are all different things. In no particular order:

1. communication with a cohort or community: learning, connecting, finding opportunities, all to develop your practice and grow;
2. publicizing your work to people who might buy it, of which there are different kinds, from publishers to "people who love comics" to "people who love the thing your comic is about" and so on;
3. creative inspiration, being exposed to new ideas, and so on

Those three jobs-to-be-done are different contexts and audiences (which audiences can always intersect!).

Mastodon, as a federated social network, is capable of (1). You can totally set up an instance that's primarily for comics writers and artists. It will be most valuable at a minimum critical mass of daily active users of a minimally viable mix of characteristics, like established professionals, people starting out, people at publishers, reviewers and so on.

Mastodon would be terrible, I think, at (2), because it is lumpy and even though there are a bunch of people on it, that bunch is only (only?!) around 14 million which in the sense of being able to reach a large audience easily, is really not comparable to the reach of something like The Web Forum That Was Formerly Problematic And Yet Also Beloved. The lumpiness of Mastodon is by design. Mastodon doesn't have to be good at this. And because Mastodon is so lumpy, it's not great in (excuse me) frictionless promotion of participation in a cross-border creator economy.

Similarly, I don't think Mastodon is great at (3), finding inspiration, because it still also doesn't have the critical mass. In a depressing manner, what's abundantly clear now is that while The Formerly Beloved Web Forum had a shitty algorithm, the algorithm [sic] is a reflection of the values of those who control it, and we've now had a brilliant example of how that algorithm for displaying and prioritizing web forum posts can, say, deprioritize the number of things that might be inspiring/"interesting" or whatever "engaging" means, to instead prioritize fascist Bluecheck replies. The experience on Formerly Beloved Web Forum, were one to look at a post from someone interesting, now prioritizes replies from complete wankers who have given Elon Musk money.

If it's not clear, I think Twitter was unique in how it (unintentionally?) commingled all the above jobs-to-be-done (never mind all the other ones it satisfied to a good-enough degree, anyway). Partly by virtue of mass and its pull with media, it was also impossible to replicate. There is only so much time you can spend in so many places.

And I still think we're in an era of disaggregation. I don't think we're likely to see something like Twitter again in the next 10 years or so, which is a prediction that now I've written it is increasingly likely to become embarrassing. Economic incentives are such that social places won't interoperate, not that they interoperated much before. But each attempt to create a new one is also a splintering, I think because each attempt to create a new one is also striving to create its own identity as a differentiator. How could it not? So now you've got early adopters -- now that there's a choice -- in a way self-sorting into different places.

Bluesky has its differentiator as, I don't know, the thing Jack had an idea about? And it grew into what appears to me to now be the cool kid table at the cafeteria where exceedingly cool kids come up with shit like sexy Alf, and I mean that sincerely and enviously.

Mastodon is the place for (amongst other things) nerdy old millennials and Gen-Xers who try to ignore the Northern European Free Software No Really We're Definitely Not Racist Libertarian contingent, and also, honestly, people who love tinkering with shit that could totally work better if you tinkered with it more, instead of shit that actually "just works".

I would talk about post.news but maintain that it's still The Place For People Who Desperately Want To Be In An Aaron Sorkin Drama Where The Smart People Talk And Walk, or they were also in debate club and Model UN.

All of these are places with vibes and now the problem is there are choices as opposed to when there just weren't any real alternatives. (Yes, I know there were alternatives. They were too late, the Formerly Beloved Web Forum had already left the station by then, and they could never catch up).

Apart from one. Threads is the text-based non-reverse-chronological web forum that has one thing going for it: it is the forum made by the company that has the other biggest forums. Threads is the forum made by the company that has the biggest web forum that can trace its birth back to basely ranking the attractiveness of women. And it is inoffensive. It is a mall. At this point, I think it's likely that Threads will be a cockroach. It will be in lots of places. It won't really go away. It will unfortunately survive a nuclear war.

And I haven't even gotten into My Takes On Governments Relying On Twitter And Facebook For Communication, eesh.

---

1.2 From Seinfeld to Social Media

Take this with all the salt because I haven't watched an episode of Seinfeld, but here's a paper:

From Seinfeld to Social Media - How a Show About Nothing Paved the Way for Twitter and Main Characters

---

1.3 Hospitals

Okay, so that thing about me spending 8 hours in the ER sparked a couple of things that caught my attention.

The first was a no-good horrible user experience[sic] with the restroom doors. The short version is:

• On the exterior, the powered access push button panel at wheelchair height also communicates occupancy status (red for occupied, green for unoccupied). In case the color status isn't clear, there is also a label explaining what the colors mean, at wheelchair height, set in a size slightly smaller than the labels for the keys on an Apple laptop. (The text is not in braille)
• After opening the restroom door, it closes behind you. It makes a latching sign, but importantly, the door is not locked.
• To lock the door, you must use a door lock push panel. That panel is at wheelchair height, next to the powered door open button. These push panels are located to the left of the door, opposite the hinge side.
• This push panel is the only way to lock the door. There is no lever, no accessible toggle, nothing on the door itself.

If I say "reader, you would not believe the number of times people used the restroom and did not lock the door and people walked in on them" would you instead start thinking "Dan, it seems as if you're about to tell me that people used the restroom and didn't lock the door and people walked in on them a lot!" and you would be right, it was A Lot.

You would also point out that this entire occupied/unoccupied system of communication, as well as the labels on the push plates themselves, are problematic for those who are red/green colorblind. And you would be right.

You might suspect that, say, cleaning staff might be experienced with this problem and that when I saw cleaning staff come through they happened to show people how to lock the restroom door. You might also suspect that those cleaning staff might not have been listened to by facilities or administration when, or if, they pointed out these issues. (I wouldn't know, I didn't check).

This is a much more professional (well, slightly more) writeup of my contemporaneous account, which is another way of saying I was bored and posted the entire thing as a thread as it happened².

Anyway, another thing about hospitals after the emergency visit has sat with me for a while: remember that time when Synapse Product Development and frog and whoever took everything they learned with Disney's Magic Band³ and applied it to... hospitals?!

I mean part of this is terrible because of course I imagine the tiered FastPass Hospital Experience where you don't have to wait as long, which anyway is totally a thing that happens at the macro level, but! Can. You. Imagine? Or, you know, queue management, which is a thing that already happens when you look at a screen and can see exactly how late overstretched and at breaking point your particular healthcare provider is!

Actually now I remember why this caught my attention: I had a wristband I was wearing that had barcodes and identified who I was. They get scanned all the time every time I have an Encounter, or whatever tiny patient interaction gets rolled-up into what's probably called a Billable Encounter. I joked about it when I came home to my wife and said "oh look I've got in/out privileges so I can go back and go on more rides! Hospital Theme Park is great!"

I like these stupid things (Disney MagicBand+ But For Hospitals!) because they're prompts to think about the healthcare experience differently and better, partly because Disney spent a gajillion dollars on this and they have a whole department dedicated to what some people would describe as "making theme parks awesome and putting super cool shit in them".

I mean, now that QR codes are finally a thing, they could include a patient link to sign up to whichever EHR so a patient can find out what's happening with their visit or whatever. And from there, you might imagine a patient-friendly set of information for an admission: what's happening when, who's looking after you and so on, but that's getting into the realm of patient experience, or PX [sic?].

But I want to back up a bit.

What would an Imagineering for a hospital system look like?

The hospital I went to was Oregon Health & Science University, which is also research hospital. What would a research or teaching hospital look like that also had a (forgive me) patient experience or design discipline?⁴. Is there even a Patient Experience Research Hospital? Some U.S. hospital systems reportedly make oodles of net profit ("even though they are non-profit"). Do any of them formally have active research programs in experience design? Hell, do Epic or Cerner have research clinics? (I know at least one reader who will be able to tell me). Should I ask my friends at NHS Digital about how this works?

This all reminds me of the time I was thinking about what it would look like to have an empowered Patient Experience role at a healthcare system. One that has the explicit remit to follow a patient journey end-to-end, plus some actual teeth to get changes made. Because again, my whole thing is the seams and the gaps.

Anyway, What If Disney Ran A Hospital?

---

1.4 Warm, Statistically Plausible Generic Platitudes

Lilian Weng, who works on AI Safety at the well-renowned open and transparent company OpenAI posted on the web forum formerly known as Twitter and it was totally a thing that caught my attention. Lilian said:

Just had a quite emotional, personal conversation w/ ChatGPT in voice mode, talking about stress, work-life balance. Interestingly I felt heard & warm. Never tried therapy before but this is probably it? Try it especially if you usually just use it as a productivity tool.⁵

I am going to preface this with a very strongly worded disclaimer that what I am going to do is armchair psychology and a bunch of projection.

So.

First, I will make the joke that here there is someone who would literally talk to a stochastic parrot before going to therapy.

Then I will ask you to imagine someone who has not gone to therapy musing that a thing is probably therapy.

Then I will ask you to imagine someone who has nobody in their circle with whom they can trust warm, statistically plausible, generic platitudes as genuine and authentic communication and connection.

Then I will ask you to imagine someone who feels misunderstood and alone, and genuinely doesn't feel like they have friends who understand them, who loves computing and computers, and feels more able to communicate to a computer than a person because a computer cannot possibly be judgmental, and I am totally not projecting here.

Now, I accept that people totally anthropomorphize things and impute intent to things that do not, and consciousness to things that probably do not or are otherwise inanimate, and emotionally identify with things because they have googly eyes on them, or are called "kismet"⁶ and a product of the now clearly morally problematic MIT Media Lab.

And then what I will leave you with is my thought that perhaps we might call the conversational interface to large language models something else. Something like Artificial Conversation Engines.

---

I intended to write this yesterday, and didn't. Partly this is because I have a strong sense of responsibility to write every weekday, because that's how this newsletter started out.

To be fair, this newsletter started out 9 years ago when my responsibilities were quite different. But I'm trying to be more... reasonable and compassionate, which is to say, perhaps it's fine to take a day off when I was discharged at 5am after an 8 hour emergency visit.

I am doing slightly better.

How are you? I love getting replies, even when they're just "hi!"

Best,

Dan