January 26, 2023

s14e10: Your Timeline needs a DJ; Enough

s14e10: Your Timeline needs a DJ; Enough

0.0 Context Setting

A grey, cold morning in Portland, Oregon, on 26 January, 2023. Let's get on with it.

---

1.0 Some Things That Caught My Attention

1.1 Your Timeline needs a DJ

One thing that keeps coming up with systems that provide access to scads of time-ordered data like Facebook, Twitter, Instagram and so on is the argument about the algorithm, which is broadly to say the decisions behind what is shown and in what order.

You've got your two ends of the spectrum, which are some sort of crunchy, whole-grain, granola I don't know, healthy plain unadulterated reverse-chronological. Just give me everything with the most recent thing at the top. I'd include in this, don't group anything together just to make the point because when people discover that grouping things together (e.g. if actions like retweets/boost appear in a feed) is something that they want, otherwise they may keep seeing the same thing multiple times in their feed, which gets irritating quickly.

Then there's the other end of things, which is some sort of magical algorithm that is inscrutable and may also rely on machine learning and a corpus of every single interaction you've ever made, finely tuned to attempt to create the maximum "engagement" from you, which goal might be replies, retweets, likes or so on. Probably replies and retweets, off the back of my head (make more things for more people to react to!)

And then there's a bonus room after that, which is a magical algorithm that also has humans pulling the strings, much like the (unsurprising, I thought?) admission that people at TikTok are able to juice the numbers and make sure you see some videos¹ and which honestly is just like the editorial team at Twitter dealing with Moments and What's Happening.

Every now and then someone also comes up with the idea that perhaps there could be choice in algorithm, or even that the algorithms (such that they might exist and be shareable and viewable in any useful way other than, I suspect, somewhat abstracted pseudocode) might exist in some sort of marketplace, where you can even use someone else's algorithm on the platform of your choice! A marketplace of algorithms strikes me at first glance as infeasibly difficult: I assume that the infrastructure of Twitter and Facebook and anything even approaching that size is custom-made toward implementing whichever "algorithm" or method has been chosen (and continues to be updated) to assemble a feed for a given user.

(I mean I am, perhaps naively, assuming there are differences between assembling a timeline for someone who's following under 500 people and assembling a timeline for someone who for some godforsaken reason is following more than 10,000 people).

Sorry, this is going to be another Mastodon-adjacent thing, again.

Simon Willison is just the nearest, to-hand person with an example of complaining that overnight, their reverse chronological timeline fills up (he is on the West Coast) and by the time they wake up, there are lots of things that have happened that he may not have seen:

At this point I'd be happy just to have a "random shuffle" button to give me a fighting chance of seeing a few of the things that happened overnight in other timezones!²

I lampshaded this with the title, but the obvious (to me?) connection my brain made was with DJs and people who make particularly good playlists. This, combined with a discussion in a Slack about "what is it that the BBC national radio stations are good for now, anyway?"

These are, arguably, editorial decisions trying to solve the problem of cutting down an overwhelming avalanche of information. Perhaps one way to think about this is to what end and purpose the avalanche is being cut down, and from here it's easy to jump into magical science fiction thinking, because if you're old enough you might just remember the concept of the Daily Me³, the mythical personalized newspaper you'd get in the morning that multiple organizations have attempted over time⁴.

An example of one of the most recent attempts at providing a digest for popular posts you might have missed is Matt Hodges' Mastodon Digest⁵, running a simple weighted score based on the number of favorites and boosts (i.e. retweets) to figure out what to show you.

But our imagination is always greater than our reach, and if you ask anyone about what they'd like to see in a digest or "what you might have missed", they might say things like this:

• things from these people, but not these other people
• things from these friends, especially
• not things about, say, this particular sport
• but things about this sport
• definitely not anything from or about this person
• a bit of "news"
• maybe some cute animal pictures just into the mix

This is starting to get complicated! It's a reflection of what people want and the amount of effort, ideally, people would like to spend and in most cases the amount of effort people would like to spend before they start the activity I imagine is close to zero because we are impatient dolts [citation needed] and if you ask people about amount of effort needed after an activity has completed then they will totally have different answers based on e.g. whether they have made their first cereal bowl or measured and designed and built a shed, they've mastered a piece of music, or, I don't know configured the settings for their Daily Me digest. I mean come on, he whined, I have to tell you what I like and don't like?

I suppose you could just make it a fun personality questionnaire.

Eesh.

Well at least it'd be a fun personality questionnaire that was published in the open and you could audit the source and it might be run by the administrator of your chosen Mastodon website instance(s) and there's more trust than say filling in a questionnaire for some rando research analytics company that's then going to turn around and do some very questionable, if not outright illegal, things with it.

But I am making a false argument here because at least for me, the amount I will avoid a lot of effort in configuring something is related to a) whether there is something more important to do that I'm procrastinating upon, and b) whether the decisions I make can easily be unmade or changed. Which they could be in an, uh, timeline algorithm or feed algorithm.

Or you could let someone else do it.

You know, you could let someone curate it for you.

I mean, you let people do that for you anyway (perhaps begrudgingly) when you look at any news media because (right?) these are editorial decisions.

You even let people do it when you subscribe to or read a weblog that purports to show you interesting stuff every day, like kottke.org, or you subscribe to a newsletter that keeps you up to date on what's in browser tabs.

There's what a person or people found interesting and picked out, and there's what's discernable from the activity of a number of posts n from humans y.

I guess what I was trying to get at this was: sure, perhaps a marketplace or choice in algorithms, but also, well DJs. iTunes used to have one called Genius (ha) dating back to 2008 that would create a playlist or mix from your library.

One open question then might be, if the post-based chronological social network you're a member of allows it: is it big enough to sustain, I don't know, a journalist? Someone to summarize and write up what's happened on the instance that day, that the people of that instance or community care about? Would you pay a dollar or euro or Great British Sterling Pound toward that a month? A week?

You want algorithms? You want the ten most favorited #astrophotography tagged photos in your network each morning? You want five random #mosstodon⁶ pictures each morning? I mean, these are all things that could be discrete and composable and again, that's part of what's exciting about this self-inflicted flaming molotov cocktail thrown into the room by the first person to have lost more than two hundred billion dollars: enough[^] people have moved and there's enough critical mass that maybe Those Other Places with that API key you can just get without having to ask anyone and be approved, maybe they can show a different way and maybe that way is also sustainable.

1.2 Enough

I had coffee with my friend Darius Kazemi⁸ a couple weeks ago. Darius is an all-around great person and also maintains the Hometown fork⁹ of Mastodon so you will not be surprised at all that we talked (amongst other things) about Mastodon. Again, I am sorry. I will make this one short.

I have pretty much switched my social media fast moving river of posts addiction from Twitter to Mastodon at this point, if it wasn't obvious. I don't post to Twitter anymore other than a) automated posts about my newsletter and b) to tell people that I am open for work, which I am (email me).

At this point, I follow about 2,300 people and am followed by about 6,500.

It is enough. I don't need it to be bigger. It would be nice if other people I know and have made friends with were on this network too, and I miss them. It is as if it's the 1980s, we've moved house and I don't get to play with my friends anymore, plus long-distance phone calls are expensive and not everyone is able to communicate electronically. They are around, I might be able to see them now and again, but they are far away.

Anyway. The point is this: the network I'm on doesn't need to get bigger for me at this point. It is big enough. Enough is happening and I'm finding new people and learning from new people and making new friends. This isn't to say that more could be happening, but then there's an argument that this is already too much! I've already said up top that there's a lot happening and too much to catch up with, right?

But this concept that a network is enough and it doesn't need to get bigger is enough of an unfamiliar feeling. If anything, governance (like I mentioned yesterday) needs to catch up with existing growth. These networks don't need to grow. I haven't seen any instances go around trying to acquire users. Well, I have, only in the sense that there was that one that sold itself to another one, which come on, a bit weird.

The elephant in the room here is that it's enough for me. It's not enough, clearly, for those who successfully used Twitter for activism. It's unclear whether it's enough to spur "viral" awareness of injustice, to whatever extent you might describe the events around the Black Lives Matter and Me Too movements. So it's not right for me to say the network is enough full stop, just that it's enough for me.

My friend Steve Lieber pointed out that Mastodon (as of a few weeks ago) certainly isn't enough and able to replace Twitter's ability to connect people working in comics, never mind people starting out. For all its faults, Twitter was also a better place than nothing for people starting out in an industry or practice, finding mentorship, learning and making connections.

I don't think enough happens overnight or quickly. I'm lucky or siloed or narrowly exposed to the point that my enough is "a bunch of early adopters" that I continually try to make as diverse as possible. But migrations happen in waves. There will -- and have been -- early adopters in every part, and then there's the chance for them to be followed. Like I said, I'm optimistic.

1.3 Small things for posterity

Small things, some of them already posted elsewhere, some of them in the idle thought category, that I liked and caught my attention:

• If you ever have a bout of imposter syndrome, just get chat-gpt to confidently write a bio about you, it totally worked for me and was quite disconcerting. This makes it absolutely clear that, in a way, large language models are democratizing access to the confidence displayed by mediocre white men.¹⁰
• "Financial services COBOL is economic mitochondria"

---

Okay, that's it for today. How are you doing? I should probably go for one of those walks today.

Best,

Dan