Meta and the fediverse: sorting heat from light

As I alluded to in my previous post on the Fediverse and Mastodon (as before, used interchangeably here, deal with it), the prospect of Meta’s new “Project 92″/”Threads” being ActivityPub compatible has led a not-insubstantial quantity of existing Mastodon users to react with incredible fury and calls for the entire thing to be blocked by all other servers (“defederated”), in theory thus smothering it at birth while also protecting existing Mastodon users’ personal data, in this reading meaning their profile information and posts. Those who have tried to adopt a “wait and see” approach, or correspond with Meta to try and figure out what their intentions are, have been pilloried and in some cases received death threats.

It’s worth, here, analysing exactly why Meta might be interested in doing this, because quite a lot of heat has been generated talking about this subject over which there are very few actual concrete facts. To be clear, my purpose here is not to defend Meta specifically (a company which has, shall we say, a rather interesting moral compass and approach to data protection to say the least) but to point out that so much of this discussion is borderline conspiracy theory, and conspiracy theory that makes basically no sense whatsoever in the cold light of day, other than if you assume that Mark Zuckerberg is actually Satan and is the font of all sin. Which, again, may well be true, but still.

Let’s deal with the main claimed motivation that Meta might have for this:

“Meta are going to harvest all our posts!”

Meta, here, are presumed to be deeply, eagerly fascinated with the posts and profile information of Mastodon users, to the point that it is worthwhile to them to build a brand new social networking service, integrated with both Instagram and Mastodon, in order to harvest this “data” through some means. This, in my view, relies on some extremely questionable assumptions, to wit:

  • the posts of Mastodon users can be used by Meta to derive some sort of value from them, likely by marketing to them;
  • Meta can do so despite these users never actually using a website owned or controlled by Meta;
  • the ~2million active Mastodon users that exist are such a disproportionately interesting and lucrative marketing cohort for Meta, compared to the several billion existing active users of its social networking sites that it already fully owns and controls, that it justifies going to all this time, effort and cost to advertise to them, somehow, through some means;
  • the only way in which Meta can harvest all of these (apparently extremely lucrative) posts and bits of profile information is by running its own ActivityPub-based social networking website.

These are all exceptionally dubious, but without all of them being true, this doesn’t make sense.

First of all, a bunch of aggregated text data about posts is not particularly useful unless Meta is developing e.g. a large language model on the level of ChatGPT. It’s not clear why they’d want to do that or what they’d gain from it.

Secondly, Meta are not interested in “data” in a vacuum. Their whole game regarding keeping and analysing the data of its users is that they wish to use this information in order to target ads, so that it can say to potential advertisers “look! Your demographic of 18-35 year old men are on Facebook! Give us money!” They’re not interested in data in the abstract; unless they can actually use this data to sell ads to someone, there is no benefit to them in having it.

In Mastodon’s case, the theory is that I, Bloonface, on, might have my posts analysed to decide that I fit in a demographic that someone might decide to advertise to. But unless I used Threads, Meta would have no means of displaying adverts to me, or selling ad space on my page, or putting (or reading) a tracking cookie on my computer. There’s no secret backdoor in the ActivityPub spec that can be used to deliver adverts to be displayed in users’ feeds on other servers; and if Meta tried to create one, it would be easily defeated by those servers simply ignoring it.

The reality, of course, is that they have no means of marketing to fedi users, or displaying ads to fedi users. And frankly, fedi users are not that interesting. They are simply not worth the effort of advertising to. Even assuming that the only way Meta could obtain all these posts and profile titbits about people was by running their own server infrastructure – again, very dubious assumptions – Meta could quite conceivably exceed the entire active user base of the entire fediverse within days should they decide to start up a Twitter-alike in anger, simply through cross-sale to its continent-sized existing user base. Whether it uses ActivityPub or not. And these would be users on its server, using its platform, who it can advertise to. In what sense should they give a shit about fedi? Frankly, you guys aren’t that interesting.

The final point is that Meta does not need to run its own ActivityPub server to achieve all this stuff. ActivityPub is not secure in the slightest and is actually kind of a privacy nightmare. Frankly, there are worse and more obviously harmful actors already using AP than Meta, and it’s quite difficult to argue that it’s somehow worse for users for Meta to have their posts than it is the bunch of Pleroma-running block-evading Nazis to whom they are almost certainly already being syndicated. (I have another post brewing about this also, you’ll be excited to hear – my mentions will be less so.)

“Meta are going to embrace, extend and extinguish ActivityPub!”

The other option is that Meta are supposedly going to “embrace, extend and extinguish” Mastodon through Threads, presumably by pretending to co-operate with ActivityPub and then screwing with the protocol in order to render Threads ultimately incompatible with it, thus gaining control of it and able to dominate it. This, too, relies on some fairly dubious assumptions:

  • Mastodon is a meaningful competitor to Facebook and Instagram;
  • Mastodon is such a meaningful competitor to Facebook and Instagram that it cannot simply ignore it;
  • Mastodon simply cannot be competed with by Meta, but must be controlled if not irrevocably destroyed;
  • Mastodon is so obviously better than anything that Meta can come up with, that Meta needs to resort to anti-competitive behaviour to beat it; and
  • Whatever Meta might do to ActivityPub cannot be reverse-engineered by other server users.

The first two points can be easily refuted simply by looking at the user numbers. Facebook has almost three billion monthly active users. Instagram has about two billion. At its peak, Mastodon has had about 2.5 million (and it is now well below that peak) Mastodon’s entire active user base, at its highest water mark, was under 0.08% of Facebook’s. Pre-Musk, Twitter had 253million; still a drop in the bucket compared to Facebook, but a hundred times that of Mastodon’s peak.

Based on those facts alone, Mastodon is simply not a meaningful competitor to Meta. Full stop. Meta does not need to somehow destroy Mastodon to “win” because Meta is already winning. Again, Meta could sign up just 1% of its existing Facebook user base to use Threads – a pretty shitty conversion rate however you spin it – and wind up with a user base ten times Mastodon’s peak. Meta doesn’t even need to think about ActivityPub. Sorry guys, but once again, you are just not that interesting.

The other side of things is that, as discussed in the previous post, Mastodon is convoluted to use, offers little in the way of a USP that matters to most users, is appallingly difficult to scale, is hostile to users in terms of both UX and culture, and is especially hostile to the sort of actual regular social media users that Facebook and Instagram have. Meta doesn’t need to resort to anti-competitive means to beat Mastodon in the marketplace, it simply needs to provide a better product to end users that is not convoluted, isn’t overtly hostile to users, is reasonably well-moderated, has a decent UI and offers a decent USP over other competing services (in Meta’s case, a strong network effect gained from its abundant cross-selling opportunities between its other big social networks). If it does those things, it has a better product. It doesn’t need to beat Mastodon with unfair competition when fair competition will do the job just fine.

The most common thing cited as precedent for Meta doing this is XMPP/Jabber. A popular post that made the rounds talks about how Google “killed” Jabber through embracing, extending and extinguishing it via Google Talk. Unfortunately, the reality that the author Ploum glosses over in that post is that prior to Google picking it up, nobody gave a shit about Jabber, which was a complete also-ran in the market place compared to the then-established competitors (MSN Messenger, AIM and Yahoo!, with BlackBerry Messenger briefly becoming a big thing here in the UK). And then, after Google dropped Jabber, nobody gave a shit about it again.

But then, next to nobody gave a shit about Jabber even while Google were using it. Ploum himself admits that most Google Talk users were just using Google Talk to talk to other Google Talk users. I was online throughout the entire era and didn’t know a single person who used XMPP. Tim Chambers had similar experiences from actively trying to switch people to it, finding it a tough sell because it wasn’t actually any better for his friends than the services they already used, and in some cases was worse.

Jabber was not a successful competitor that Google needed to crush through anti-competitive means, it was a tiny little also-ran IM service used, almost to a man, only by FOSS users who believed in it ideologically and were willing to forgive its faults for the sake of that ideology, had a lacklustre feature set even in 2008 and by 2013 (when Google stopped federating with other XMPP servers) was even more hopelessly outclassed in terms of features. For example, I can’t say for certain, since like I say I’ve never used it and I don’t know anyone who does or did, but it appears that Jabber never supported inline picture messaging. This is something that Facebook Messenger and Windows Live Messenger could do in 2011. Something about that tells me that the reason XMPP didn’t dominate the world has more to do with it actually kind of sucking, and less with Google deciding not to throw continued effort and resources at something that ultimately wasn’t very relevant to their users and also sucked.

Perhaps the actual lesson to learn from Jabber is that a decentralised also-ran social service that is and was never actually a meaningful contender or competitor might have issues that can’t be blamed on a big bad corporation adopting it, and that instead it should focus on providing features and satisfying use cases that users value. One could also be somewhat unkind and suggest that Google simply provided a convenient conspiratorial explanation for the more obvious, Occam’s Razor truth that Jabber just plain sucked from an end user’s point of view whereas the actually popular networking protocols that their friends were on didn’t. One might also hasten to point out that Windows Live Messenger, Skype, AIM and indeed Google Talk itself were also similarly left by the wayside because of competitive pressure, and didn’t need to have some nefarious outside party intervening to do so.

To bring this back round to Meta and Mastodon, then, Meta does not need to “embrace, extend and extinguish” Mastodon to beat it in the marketplace, the same way that Facebook Messenger did not need to “embrace, extend and extinguish” Windows Live Messenger to be obviously better and more useful than it. Indeed, Meta does not need to beat Mastodon in the marketplace at all, because it already has; as far as social platforms go Mastodon is a tiny network that is not particularly competitive with Meta and poses no real threat to them. You might as well be talking about Kellogg’s looking to shut down a one-woman granola stall on Norwich market.

Indeed, as the (shall we say) spirited reactions to my previous post often stated, a great deal of Mastodon’s users do not want mass adoption. In which case, what’s the issue with Meta “embracing, extending and extinguishing” anyway? Nobody can prevent you from using vanilla ActivityPub and keeping Mastodon as the tiny, irrelevant niche social platform it is at present – it’s just that Meta won’t be joining you on it. Isn’t that what you want?

So what else is there?

“Meta are evil bastards and they’ve done lots of evil things so they’ll clearly do the evil thing I just imagined they will, for no other reason than they are evil!”

I agree that Meta are not a nice company. I use what services of theirs I use basically under duress. But I’m not sure what relevance that has to any kind of cold discussion over why Meta might want to link Threads with ActivityPub.

Is the idea here that Meta Platforms Inc., a large, publicly-traded corporation, is going to harvest non-specific data from a tiny, niche social media platform, or work to kneecap Mastodon (in a way that won’t work), in ways that involve a great amount of time, effort and money, solely because it is made up of horrible people that want to do bad things because they’re evil?

Have you seen the inside of a big corporation before? I can’t say I’ve got any personal insight into Meta’s corporate governance, but I very much doubt their board meetings resemble Dr Evil discussing a plot to take over the world, and everyone simply going along with that. (For a start, Nick Clegg is a far worse person than Dr Evil.)

In reality, launching a service like Threads is a massive undertaking with a lot of moving parts. It’d involve, at a minimum, engineering time, marketing spend, legal discussions, hiring, branding discussions (probably involving external parties) and trademark registrations. Probably a lot more than that. In a company the scale of Meta, you’re talking millions – possibly tens of millions – before you’ve even shipped a product. All of this time, money and effort spent, essentially just for the lulz.

All of this in a publicly-traded corporation that is accountable to its shareholders, to whom it has to provide regular financial and strategic reporting which cannot include a simple line item for “nefarious ne’er-do-welling”. And this would be happening while these same shareholders are already not all that impressed with Meta to begin with, because it turns out the hare-brained metaverse bollocks they’ve bet the farm on might be the only product less obviously useful to end consumers than a decentralised microblogging service. And microblogging platforms are themselves not a famously profitable undertaking, which is why Elon Musk had to be legally threatened into completing his purchase of one just as soon as he realised that actually it was a complete waste of money.

In short – are you fucking high? At least Meta’s other various acts of nastiness had some sort of positive result for them. What does this accomplish? What actual benefit would any of this accomplish.

So why is Meta going to use ActivityPub?

Here’s the thing; I haven’t the faintest fucking clue.

It does not obviously help Meta in any way to have interoperability with ActivityPub. Threads would probably gain a lot of users whether it had ActivityPub or not, simply by virtue of a popup appearing in the Instagram and Facebook apps advertising it. While I’m sure it’s comforting for the people who subscribe to one or more of the above conspiracy theories to believe that actually Mastodon is really interesting and consequential, the reality is that Meta doesn’t need them.

The Jabber analogy works in a different way here, in that it’s not particuarly clear why Google Talk needed to interoperate with XMPP at all. Google gained a user base for GTalk in spite of its Jabber integration, not because of it. The users it would have found most useful to its end goal – ultimately, revenue – were not the sort of, bluntly, nerds who loved XMPP because of its intrinsic XMPP-ness.

In the same way, what Meta really needs for its purposes is not Mastodon users, but the sort of people who your average “normie” social media users follow, and whom Mastodon does not have and is largely hostile and/or unappealing to: sports stars, celebrities, journalists, brands, influencers. Basically none of those people or companies use Mastodon; they all use Instagram. They also all use Twitter. Network effects matter, and Mastodon doesn’t bring much of one to the table.

The best explanation I can come up with is that federation with existing Mastodon servers, and their content, allows Meta to effectively “bootstrap” Threads, rather than it being obviously a ghost town. But then they don’t need to do that either. Most social networks started with nothing and nobody. And Meta is starting with the prospect of cross-sale to people who people might like to follow. Integrating it with Instagram – or essentially turning Instagram into a front end to Threads – instantly gives it a user base that is likely to outstrip both Mastodon and Twitter combined in a matter of weeks, even days.

But the main point here for me is that a gap in why Meta are doing this does not automatically make bizarre and unsupported theories or assertions about why they are doing it any more likely. The key ideas that get thrown around don’t make any sense, business or otherwise, and/or rest upon Mastodon being a far more important and consequential platform than it is. Given that there is, to put it mildly, an information gap, and so much of this is speculative, it’s not entirely clear why “wait and see and then block them if they actually do something bad” – or even “find out what this thing actually even is” is as unconscionable as trailed.

9 responses to “Meta and the fediverse: sorting heat from light”

  1. Not The LBC Guy Avatar
    Not The LBC Guy

    Scale does interesting things to an organisation. Things that would be a very healthy line of business for a company with a million customers become trivial distractions at the billion user scale. But similarly, investments with the potential to bankrupt a million-customer company can be smart long-term bets for a billion-user company.

    Rather than the Google/XMPP metaphor, I would look at something more like the CIA during the cold war. A huge org with vast resources and a singular mission to protect the USA’s dominance. It doesn’t take a huge leap to draw parallels between that position and Facebook’s place at the pinnacle of the social network hierarchy.

    Sure, the CIA had massive projects like Azorian, but it also did a million small projects just to prevent the prospect of the US Government being weakened. To the outsider the agency’s reasons for doing this are opaque and maybe even nonsensical but someone in there had to justify the idea and get a budget, tied back to that mission of securing the existence of the USA (as existing).

    So what counts as an existential threat to Meta? Well there are your obvious big threats like Google finally figuring out how to leverage its userbase, but once you have a team working on that how else do you fortify the borders? You look longer-term. You look at anything that has the potential reach billion-user scale, because billion-user scale means enough of a network effect to start draining eyeballs from Meta’s services. And it’s cheaper to nip any potential threat in the bud when it’s in the single-digit millions than the tens or hundreds of millions.

    Through the lens of this being a protective move for Meta, it’s not about the cost of doing it, it’s about the potential cost of not doing it. If (and I accept it’s a big if) the Fediverse reaches enough scale to undermine Meta’s network effect, that could cost Meta everything. Even a few tens of millions to prevent that possibility is a sound investment.

    Of course, ActivityPub can resist technical changes. Of course, the Fediverse can make reactive policy changes. But those are weak protections in the face of Meta’a ability – as you pointed out – to simply flood the network with its own users. They’ll use scale as an argument to have a seat at the table, to bully through their PRs, to implement their own extensions to the protocol and say adopt these or become the fringe implementation of your own project.

    So that’s my guess at the why. It’s a pretty standard embrace-extend-extinguish play, focused on either smothering the Fediverse or making it so tied to Meta services that there’s no practical distinction between the two.

    1. This doesn’t make sense – all you’ve done is reiterate the “they’re going to embrace, extend and extinguish!” thing in different words, while ignoring the reasons that they do not have any cogent need to do that.

      This sounds like Bitcoiners in 2013 going “in 2023 when everyone uses Bitcoin, then the big banks will be screwed!” while failing to notice that a) nobody used Bitcoin then, b) nobody had any coherent reason to switch to Bitcoin from the money they already had and c) they had no coherent plan to overcome point B. The whole thing falls apart if you don’t assume that “the Fediverse reaches enough scale to undermine Meta’s network effect”. Even you grant that that’s a big “if”, so it can’t really be presented as an inevitability, or something that makes any sense for Meta to respond to with such a vast influx of money, time and resources.

      As it is, Mastodon has failed to reach even one percent of Meta’s total user base, despite its most obvious competitor and comparator falling over. Why would they care about EEEing ActivityPub when it poses no coherent competitive threat to them or their interests? That’s the simple question nobody seems to be able to answer, except with handwaving about some hypothetical future where actually, Fedi is a hundred to a thousand times the size it is, through some means, nobody seems to really know how.

      1. Not The LBC Guy Avatar
        Not The LBC Guy

        Well, this is where I think you misinterpret my argument. The question is not whether the Fediverse can or will reach that scale (my personal view is that it theoretically could, but it won’t, and I’m perfectly fine with that).

        Instagram wasn’t anywhere close to Facebook’s scale when (then-)Facebook bought it, but it was a forward-looking move. What’s the forward-looking move with a potential competitor that you can’t buy?

        So the question is, are Meta’s leadership looking far enough into all possible futures that THEY consider it a possibility that the Fediverse reaches competitor scale? If they are and do, then it makes financial sense to spend at most a few tens of millions today to remove that threat from their future, rather than billions in the hypothetical future where it has reached a competitive scale. And like I said, scale brings its own perspective – tens of millions might seem like a vast investment to you or I, but it’s a rounding error to Meta’s turnover.

        If it’s valid to question the why behind my reasoning, then I think it’s valid to question the why not behind yours. As you say, there’s no obvious external reason that Meta would choose to get involved in ActivityPub. In that case it only makes sense to try and infer a purpose from Meta’s prior behaviour in similar circumstances. And they have a history of moving very early to defend their dominance. I think it’s quite naive in the absence of a why to assume the move is designed to favour the Fediverse rather than Meta.

  2. My sense of why Meta is using ActivityPub is that it’s for the same reason that Meta is using HTML, CSS, and JavaScript.

    It’s a standard that’s there, that does most of what Meta needs, and that doesn’t involve spending money on reinventing the wheel.

    It’s good enough for the job and may have the side benefit of interop with WordPress, Tumblr, and others who have already said they would jump onto it.

    From a financial standpoint, it means they don’t have to invest time and money into building a protocol. Building a protocol does nothing in creating value for Meta so spending money on building vs. using an existing standard does not make financial sense.

    Looking at it through the valley lens, there is one well funded competitor to Twitter right now: Bluesky.

    But Bluesky is developing its own protocol so using ActivityPub immediately neutralizes that potential threat.

    This leaves Mastodon as the only issue and, for all the reasons you enumerated here, Mastodon doesn’t matter. The Mastodon business model (if there is one) is not seen as a threat to Meta’s ad based business. If using ActivityPub also neutralizes that, then Meta will be ok with it, but Mastodon is not something they are losing sleep over. More of a collateral damage situation if it’s killed.

    That’s the primary reason for using ActivityPub.

    1. David Dudovitz Avatar
      David Dudovitz

      I fully agree that is the main and obvious reason. As a side effect, though, there is one tiny part of Mastodon that does not matter in any way to the bottom line of Meta Platforms, Inc., but might (emphasis on might) matter to Zuck personally: people who migrated from Tech Twitter. As far as I can tell, they make up the main body of people who left Twitter for Mastodon back in November, and Zuckerberg might personally care about having them on his service in some fashion despite it not profitting the company at all. Given that they are unlikely to leave Mastodon, using ActPub for cost reasons might allow him to get tech people to be federated into his service as a bonus.

      1. I really don’t see what purpose having a bunch of tech Twitter people would serve for either Zuckerberg personally or Meta more generally.

        Moreover, the Venn diagram of “tech people who really like ActivityPub” and “people who’d use a Meta-run social networking site” isn’t one with much overlap.

        It might be flattering for tech people to think that Meta is really bereft without them posting. But I think they’ll do just fine without them.

    2. It’s a standard that’s there, that does most of what Meta needs, and that doesn’t involve spending money on reinventing the wheel.

      Facebook does have form for this. They used XMPP internally (i.e. without federating with anyone or exposing federation endpoints) to power the first few iterations of Facebook Chat. (They may well still use it internally in a heavily-modified form, I don’t know.)

      The other thought that crosses my mind is that there’s just some engineers within Meta/Instagram who have been poking around with ActivityPub on a “huh this is kind of cool” basis and somehow managed to wind up with that being pissed around with at about the right time for it to become a shipping product. Not a few tech products and services have been released to the public after just being a bunch of engineers pissing around with something they thought was interesting. But admittedly, that doesn’t really pass the “would spend tens of millions of dollars on” test.

  3. Threads is getting to compete with Twitter because they see an opportunity right before Election 2024 as Musk experiments.

    Using ActivityPub is a selling point. If they can get a few instances to federate them, it gives them some credibility that they are embracing open standards.

    The user base of Mastodon too small and too different to monetize successfully.

  4. The reason why meta is using ActivityPub with threads is simple, because they can.

    It doesn’t hurt or aid them in any meaningful way as it stand so why not? It’s a selling point they can tout threads’ app stores page to seem trendy and forward looking and it really doesn’t matter if the existing decentralised community don’t want them, their ecosystem’s existing userbase will be enough to sustain threads atleast for a little while.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.