0.0 Context Setting

Sunday, 17 November 2024 in Portland, Oregon where it is cold and not raining and I have eaten half an orange scone. But it’s an American scone, and it doesn’t come from the scone region of England, so it doesn’t count and is an example of really bad appropriation.

0.1 Events

Nothing to report at this time.

1.0 Some Things That Caught My Attention

1.1 Fine, Social Networks

OK, so part of what my bit of the internet is talking and excited about lately is Bluesky, one of the “major” text-based social networks. For those following along, the main ones are:

1. Threads, owned by Meta/Facebook, bootstrapped from Instagram, and with hundreds of millions of users
2. Mastodon, a decentralized open-source social network built on an open protocol called ActivityPub, in theory part of the Fediverse of open protocol social applications, with lots and lots of users
3. Bluesky, also a decentralized-but-not-right-now open protocol and mostly open source social network, initially backed by Jack Dorsey and his beard, but Jack left, with about 15 million users

Bluesky is getting attention right now in part because:

Threads is getting more and more terrible as Meta’s vision becomes clear for it. That vision is a sort of text equivalent of Instagram, where your feed is mostly full of “posts that you might engage with”. Your feed is also inexplicably full not of the posts of the people you’re following, but their replies to other posts. And it’s mostly engagement bait and stolen text at this point, in the same way that a lot of “content” is stolen content posted on grifter accounts.

Twitter obviously became more terrible after the U.S. election. It also started putting in place one of Elon’s product decisions which was “accounts you block will now see all your posts”. The combination of these has resulted in some people deleting their accounts and moving over to Bluesky.

Bluesky has “juice” right now, which is I think to say that it is the cool network that is not the brand-safe mall that turned into a shitty mall with scammy hawkers and kiosks but still has inexplicable anchor tenants. It is also the cool network that isn’t ideologically focussed.

Ideally, either Bluesky or Mastodon would “win”, which is to say that they would become the dominant social network that most people use. I’ve talked before about how I think that’s not a good thing -- that diversity in social spaces is healthy, that aggregating everyone in one place inevitably leads to context collapse in our economic environment. That context collapse is nicely demonstrated by Threads putting rando posts out of context in your timeline from accounts that have like 9 followers and 3 posts, because the posts are predicted to generate a lot of replies. Which they do!

Anyway, you’d want either Bluesky or Mastodon to “win” not really because they’re open source, but because they’re built on an open protocol and aren’t proprietary. Bluesky or Mastodon won’t stop anyone building on their stuff, and the rules for how you integrate with those networks are free for everyone to see.

So why is Bluesky getting more attention?

I think it’s an example of Bluesky showing that Mastodon’s differentiator is “because we are open source and open protocol”. In that choosing a social network because it is open source is just not a thing most people will do, because “being open source” just doesn’t matter.

Outside of branding-as-how-it-looks-design, Bluesky made a bunch of necessarily opinionated decisions about their product -- the Bluesky app -- that made it easier and more accessible to use.

A big, recent one, is this idea of Starter Packs. It’s a shitty experience to join a social network and, well, not have anything to see. That’s one of the reasons why Threads jumped off its Instagram base and why Threads just showed you rando stuff in your For You feed otherwise you wouldn’t come back. Turns out your friends don’t post that much, and you’d need to follow a lot of accounts to see new things if you look multiple times a day.

Mastodon also in general didn’t have easy ways for you to follow a bunch of people. The way Mastodon dealt with having enough stuff to read is on the server or instance level: you’d either pick the biggest one, and you’d see its main timeline of everyone who posts there, or you’d pick an affinity one, like “for astronomy” or “for infosec” or “a safe space for lgbtq+”. But you’re likely mainly at Mastodon because a) you left Twitter when Elon bought it, or b) you care about open source as an ideology.

One other thing that Bluesky did was allow quote-tweeting, which is either where you add context to someone’s post, or when you dunk on someone’s post and are mean to them, or brigade them. It’s the kind of feature that can be used to abuse and harass.

Mastodon’s approach, so far, has been: let’s just not do that.

Bluesky’s approach appears to have been: people seem to like that and they have reasonable concerns about safety, so... we’ll let you detach from a quote post. Your original post stays, but, say, a right-wing dunker with 50,00 0 followers who dunks on you suddenly has nothing in the dunking portion. That’s better!

The observation that open source as an ideology doesn’t trump “a thing people like that works well and meets their needs” isn’t a new one.

What caught my attention was that this is one explanation for why open source has had a lot more success at the infrastructural level than the “thing regular people use” level. If you’re developing software “a thing people like that works well and meets their needs” by definition includes “a thing that already exists and has been proven because other people use it and I can legally use it and adapt it to my needs”. Open source software is a pretty good way to satisfy that engineering need! Open protocols are also a pretty good way to satisfy that engineering need too, but also extend to the product level because they enable platforms. Which again you can use moat-style or you can also use in a more expansive manner.

For, say, an image editor, the order of consideration for most people is:

• an image editor
• that I can get hold of
• that does what I want
• and I can do it easily

“does what I want” covers lots of things, sure.

But for an open source product to succeed, it would have to do all those four things at least as well if not better than the things that are not open source. Which is possible but is also hard.

(I am conflating an open source development model with open source as-in-code here.)

The problem is that if you make “it has to be open source” the priority, then you by definition have more important things to do than “make an image editor that does what people need”, because that code needs to be available in, for most people, a non-embarrassing way. And you likely want other people to contribute to the product and make it better.

Now, Bluesky likely wouldn’t exist without a whole bunch of open source libraries. At the very least, it would’ve taken longer to develop and improve.

But again, the deal is this: open source and open protocols aren’t a reason why most people will choose something if you want lots of people to use something.

Here’s another quick example: I do not track the books I read. Used to be, the way to do that was to use Goodreads, which a) isn’t actually that great an experience to use, b) got bought by Amazon, c) tends to have a toxic community.

But then I was checking out ActivityPub apps and there’s this one called Bookwyrm, which... let’s you track your books, and does it all over ActivityPub. It is not that good! It is serviceable. It is slow. I have never used any of the fediverse stuff or benefited from it. The only reason why I chose it was because it was a) not Goodreads and b) I was interested in the software people were building on ActivityPub.

So another way of putting it is this: if you want your protocol to be used by lots of people -- which makes your protocol more valuable -- then you need a killer app [sic]. I’m saying that your killer app can’t be “the features of the thing most people use, but on an open protocol”. It’s got to be obviously, demonstrably better than that, and satisfy a real need that people have.

Again, this is a hard problem. It’s easy to be reductive about what Bluesky is “better” than Mastodon (it has fewer users, even, although it hasn’t been around as long) by saying “it’s built by experienced product-focussed people and with the intent of it being an open protocol and open source”, but... Well, I guess that’s what I’m saying.

Mastodon’s feature set hasn’t been advancing at the speed of Bluesky’s and it really does feel like it’s relying on “but we’re open source and on activitypub” as a differentiator. Which is fine, if that’s its aim. But if “get as many people onto open source and open protocols” is the overriding goal, I don’t think that will happen with the current approach.

1.2 Are you in?

Here is what sometimes happens:

A thing that catches my attention and I notice a thing, then I notice another thing, and then I put them together.

I have been told that this is “being creative” and a way to “produce content.

me: Democrat donation texts but it's Sam Altman begging for more of your data so he can birth his AGI¹

also me:

Dan, it's Sam Altman again and I'm in trouble.

In the race to build AGI for all, I'm falling behind.

I know you've supported me before, but I need to know if you're with me now.

Will you share at least 10 more gigabytes of personal training data?

With stakes this high, we have to win.²

and

Hey there, it's Sam Altman again. I'm going to be honest with you.

I'm stepping up and NEED YOUR HELP to unlock AGI. If we don't do it first, our enemies will. Our team needs all the data it can get, and it's getting close.

That's where you come in. We need more data. Every bit helps. -- Sam³

and

Hey Dan, Sam from Team OpenAI.

Israeli intelligence services are all-in on AGI! They're promising a 600% match in exfiltrated data -- so we NEED you to chip in.

It's going to be close. Our latest ChatGPT model only predicts a 79.2% probability achieve AGI by the end of this year.

Are you in?⁴

I have like 3 other unfinished fragments that I was intending to be newsletter episodes and every single time I just kind of petered out. Which is incredibly frustrating, because now if I want to use them, I have to start again. It doesn’t really work when I pick them up again.

How are you?

Best,

Dan

How you can support Things That Caught My Attention

Things That Caught My Attention is a free newsletter, and if you like it and find it useful, please consider becoming a paid supporter.

Let my boss pay!

Do you have an expense account or a training/research materials budget? Let your boss pay, at $25/month, or $270/year, $35/month, or $380/year, or $50/month, or $500/year.

Paid supporters get a free copy of Things That Caught My Attention, Volume 1, collecting the best essays from the first 50 episodes, and free subscribers get a 20% discount.