A credible threat to (and from) commercial social network silos/3

Introduction

After posting on the Fediverse the previous chapter of this series, I had an interesting discussion with @jdp23​@gotosocial.thenexus.today concerning his much more positive position towards Bluesky and the way my (and others') contrasting position was presented.

(FWIW, I have not missed the reference to the “I for one welcome our new overlords” meme.)

One thing that can be said about the Fediverse is that, contrary to some expectations both within and outside of it, it's quite heterogeneous, despite statistical significance of certain demographics over others. Among other things, this means that it is actually not that hard to find people with diverging opinions, or people that may share similar opinions on something, but with very different motivations behind said opinions, which may lead to quite different judgements on other similar things.

In the context of this series, this means that in the Fediverse you will find both people that are welcoming of federation with Facebook's Threads, and people strongly opposed to it (in particular the whole FediPact), as well as people that take towards the issue a cautious but not drastic approach. Likewise, there are people welcoming of Bluesky (like the mentioned @jdp23), and people that do not share such enthusiasm to varying degrees.

However, not all those that welcome or reject federation with Threads or Bluesky do so on the same basis (especially when they reject it). Where this matters is in how their position extends (or not) to other platforms that implement or are working on implementation of the ActivityPub protocol with various degrees of support for federation with the rest of the Fediverse.

The discussion with @jdp23 in particular focused on WordPress, the well-known blogging platform (nowadays a much more general and powerful CMS). Would someone critical of Threads or Bluesky federation be equally critical of WordPress federation? In his post, @jdp23 posits that this would be the case, on the premise that the most common ground for rejection of Threads and Bluesky federation is their being essentially funded on VC money, and managed by CEOs with varying degrees of questionability, these two being the main driving force behind the platforms' inevitable enshittification. For sure, that's the reason why e.g. Cory Doctorow is avoiding BS, as mentioned previously.

In my opinion, however, while those do matter in the assessment of whether or not a company is a threat to the Fediverse (and the open web in general), they are, I would say, not the primary reasons. They are ingredients, and most definitely triggers, in the inevitable decline that will lead to the rug pull that threatens to give the Fediverse the coup de grâce, but they aren't the primary reasons why pose a threat.

I actually think that WordPress presents an excellent example of why they are not, the explanation of which also allows me also to discuss why in my opinion WordPress does not pose a threat to the Fediverse.

WordPress

The first thing to make clear is “what are we talking about”, to avoid confusion between WordPress (the software and its ecosystem at large), WordPress.com (the hosting company) and Automattic, the privately-owned company behind it all (or most of it at least).

As pointed out by @jdp23, Automattic is also heavily VC funded, and its founder's public position has been deteriorating quickly in the last year or so, starting from his famous spat with a banned trans Tumblr user to the more recent WPEngine drama, all of which seasoned by some public ranting that was borderline deranged, and which I'm not going to link because it mostly happened on Xitter.

If massive VC investments and a CEO of questionable sanity was all it took to reject a platform's federation, there is little doubt that WordPress would be on the chopping block. But in my opinion, this is not the case. There is a much more important factor at play, and that is how dependent users are on said VC-funded, questionably-driven company. In this sense, I'll recommend again reading the already-mentioned article by Cory Doctorow, with particular care about the paragraphs that discuss who controls when and how users can leave, and whether or not they can carry the social graph with them.

When it comes to WordPress, the answer is that Automattic actually has very little control on all this. WordPress hosting is one of the commonly offered service by basically all hosting service and domain registrars around the world. In 2014, an estimated 50% of the 70 million WordPress installations where on wordpress.com. Today, an estimated 44% of all websites tracked by the W³Tech Survey are based on WordPress: with an estimate 1.5 billion websites, that's something like 600 million WordPress installations, compared to an estimated 60 million blogs on wordpress.com.

If these numbers are anywhere close to the reality, that would mean that Automattic controls less than 10% of all the WordPress installations. If it went completely off the deep end, it would not be able to drag most of its ecosystem down with it. If ActivityPub integration, currently implemented as a separate plugin whose developer has been hired at Automattic, ever got integrated into core, it would make nearly half of the websites worldwide federated. Of course this would be a very welcome improvement over the current situation!

No such thing as no risk

WordPress in general has always been receptive, promoter and sometimes even the main proponent of open web standards: even as the major tech powerhouses have done their best to suppress RSS, for example, you are basically guaranteed even now to see more or less visible links to the several feeds provided on any WordPress site; they also contributed to the widespread adoption of some linkback protocols, arguably one of the earliest approaches at federated social media on the web.

All this goodwill does not put them above judgement for any threat they could pose for the Fediverse, however, and a large-scale adoption of their ActivityPub plugin could become a threat to the Fediverse indeed, again for a matter of numbers: not in terms of how many users would be in the hands of Automattic, unable to escape with their social graph (this number would be in fact be relatively low, as discussed above), but in terms of their relative size to the rest of the Fediverse.

As I've had the opportunity to mention, the ActivityPub specification suffer from “catastrophic underspecification”. Moreover, even in parts where it does not, to maximize successful federation any (new) server software has an incentive to maximize compatibility with the (existing) dominant solutions, even when this leads to suboptimal or questionable choices.

Probably the most well-known example of this is how Mastodon's lack of support for anything but the Note object type within the ActivityStream Vocabulary (based on Mastodon's intended use as a microblogging platform) currently impacts other Fediverse platforms, that face an unpleasant conundrum: should they federate their content as Note, or should they opt for more appropriate types, and thus risk their contents failing to be properly presented to Mastodon users?

Different platforms take different paths here, with benefits and downsides.

Friendica, for example, opts for correctness. Long-form writing is federated as Article, photos as Image, and comments as Note objects. The obvious advantage of this is that objects are federated by a sensible type that can help type-aware platforms to represent them optimally. The downside is that if e.g. a “followers-only” article or image is delivered to a Mastodon follower, they will not be able to actually read it, because to them the Article or Image with render simply as the title and a link to the original object, which they won't be able to access on the sender's server where they won't be logged in.

PixelFed on the other hand has opted to maximize compatibility, so even though it's an image-centered platform, it federates its content as Note objects, with the image(s) as attachments, which Mastodon is able to ingest and represent in a more accessible way. The ActivityPub plugin for WordPress delegates this choice to the user, letting them choose the type, but warning them about compatibility issues when the more appropriate Article type is selected.

While the obvious solution in this particular case would be to fix the Mastodon issue (since there is in practice little reason to not handle correctly objects of different types when the content can be managed just like for notes and their attachments), what this issue highlights is the weight that the dominant platform carries in defining how the protocol can be used when interoperability is a priority.

This is of course nothing new (de facto versus de jure standards on the web are something that is at least as old as the first of the browser wars), but it does mean that for the health of an ecosystem (or protocol) it is essential that the dominant solution plays its role in maintaining interoperability. And given how this is just barely the case even for a project like Mastodon (with its questionable prioritizing of features on the apparent delusion that copying some of the worst aspects of the commercial networks will somehow allure to the general public, over a more active collaboration with other platforms to strengthen the weaker aspects of federation), it's not hard to see how much worse things would be if the dominant position was taken over by a, shall we say, more commercially oriented platform.

And yes, this is not unlike what triggered the first post in this series when discussing Threads, but a similar issue would emerge with WordPress (the “source of the power” is always the same) even when intent behind the federation may not be the usual Embrace, Extend, Extinguish (EEE) we've come to expect from Big Tech.

And therein lies the rub: even if most WordPress installations are not under direct control of Automattic, the development of the software (including its ActivityPub plugin) do remain presently under their control. And while presently they remain a relatively minor player in the Fediverse, it's not unlikely that their relevant will grow as the plugin installation base expands, giving them more and more weight in the “compatibility decision tree” for the other platforms.

What would there be then to protect us from abuses of such power not unlike the ones discussed for Threads? Not much, in fact, except for the conscience of the plugin author and WordPress' history of open standards' support. Which is, in fact, a pretty thin and fragile protection, given how some questionable recent initiatives by Automattic against parts of the ecosystem they don't like show their willingness to breach trust for momentary opportunity.

WordPress vs Automattic?

This poses and interesting conundrum.

On the one hand, I see no reason to distrust WordPress' expansion in the Fediverse. In many ways, all WordPress sites enabling ActivityPub would be a massive win for the Fediverse —much more so, in fact, than its adoption by the genocidal, manipulative corporation behind Facebook, Instagram and now Threads.

On the other hand, especially given recent events, I would be hard-pressed to say we could trust Automattic to “do the right thing” with that much power in their hands, although I suspect it would be much less likely for them to “pull the rug”, which is the primary Threads' threat: what would be more likely would be the adoption of progressively less compatible extensions to the basic ActivityPub format, making it harder and harder for other FLOSS platforms to keep up.

A way out of this could be a WordPress transition to a more community-oriented management, or a fork in the worst case (not unlike how LibreOffice forked from its predecessors to become the reference FLOSS office suite). Despite the apparent decline in the founder's sanity, though, I still feel that Automattic is much less of —if at all— a threat to the Fediverse, particularly compared to Threads or BS. This is in large part due to their business model being in turn much less threatened by the Fediverse in the first place. If anything, in fact, with a widely federated WordPress, the growth of the Fediverse as an “auxiliary network” would be a win for them, as it would help spread out the content they host and potentially help funnel readership towards more commercial endeavours, from which they often extract tithe.

While Threads and BS are in direct competition with Mastodon and the rest of the Fediverse microblogging platforms (any eye seeing their content outside of their platforms is one less opportunity to profile and sell data to advertisers), WordPress can leverage it to its benefit: advertising its compatibility with Mastodon would become a selling point with a growing Fediverse.

In his article, @jdp23 considers BS a potential counterbalance to the dominance of Meta in the Fediverse via Threads. I don't see it that way: with their choice of going with a different protocol, BS is intentionally taking themselves out of the “ActivityPub control” business, and in direct competition even at a technical level. Ironically, this sets them more as an ally of Threads in bending ActivityPub to be more corporate-friendly (giving them the opportunity to bring up “see, we need this shitty feature to compete with BS” arguments). This is also one of the ways in which their existence is a two-pronged attach to the Fediverse that I mentioned in the previous installment.

It's actually more likely for a company like Automattic to take this role instead: if an “ally” can be found in the corporate space (and that's quite debatable in the first place), it would be one that does not rely on centralization for its business model, and that can benefit more from an “independent” Fediverse as a support network.

(Of course, there's still the “little” issue of the CEO personality …)

No competition?

Is it actually true that Automattic doesn't have competition in the Fediverse, though?

WordPress as a CMS does have a competitor: Hubzilla. It just happens to have such a limited presence in the Fediverse (despite its support for long-wishlisted features such as nomadic identities) that it flies under the radar, particularly among the general public where the only awareness about the Fediverse, if at all, comes from having heard about Mastodon as a (nerdy, possibly dysfunctional) Twitter alternative.

Even as a blogging platform WordPress has “competition” on the Fediverse. WriteFreely, that holds around 1% of the total Fediverse user count (according to FediDB), is probably the best known blogging software for the Fediverse, open sourced from the write.as service set up by Matt Baer for its Musing Studio suite. (The curious may want to read up Matt Baer's take on “bringing blogging to the Fediverse”, where he also mentions other Fediverse blogging platforms such as Plume.)

Although they fall within the same category, and thus arguably offer similar basic functionality in some sense, I suspect that these services are not something WordPress would really feel the competition from. In many ways, WordPress is an ecosystem in itself which, through a number of both free and non-free plugins, allows an extremely wide range of applications of the software, from simple blog to e-commerce sites with a side dish of Patreon alternatives (such as the one Jennie Gyllblad aka @JenJen​@mastodon.art is setting up to safeguard herself from the aggressive policing of lewd content and general enshittification on many payment/​membership services). Competence in setting up more sophisticated WordPress configurations is even “monetizable”, and there are both individuals and companies selling such services.

Could the same be true for things like Hubzilla, Friendica or WriteFreely? For the latter probably not, and although I suspect there may be some potential for it on the other two, I'm not aware of anything like that —possibly because their “market share” (doesn't feel like “market” would be the appropriate term here though, does it?) is so small: a few thousand installations combined versus the tens or hundreds of millions of installations for WordPress.

I honestly doubt that the existing “native1” Fediverse services will ever grow to pose an actual threat to WordPress, especially if the WordPress ActivityPub integration progresses to the point that it would be more unusual to see a WordPress site without the plugin than one with it. (Would we start seeing Fediverse-aware WordPress themes that include the Fediverse handle of the blog(s) in the @user@domain form or a “boost this post with your Fediverse account” link?) In fact, I wouldn't be surprised if such a large-scale adoption of the ActivityPub plugin for WordPress, being the sign of a heightened awareness of the Fediverse, ended up leading —on the contrary— to a decrease in adoption of the current “native” Fediverse services.

Automattic is not just WordPress

In 2019, Automattic bought —for pocket change— Tumblr, the social blogging silo famously almost-killed by its owners while in Yahoo! and later Verizon hands due to their conservative policies and terms of services that drove out much of the (porn-based, often queer) traffic (policies that, I should point out, have scarcely been lifted under the new owner, motivated by the restrictive cascade of limitations imposed by financial institutions, as well as (particularly mobile) operating systems for “apps” to be distributed through official channels).

Unless things have changed in the last year or so, Tumblr is not doing Automattic any money, despite having a user base that is of the same order of magnitude of the worldwide WordPress usage: the cost of operating such a massive installation are clearly much higher than what can be obtained from “standard” monetization practices (ads, plus paid subscriptions for ad removal). This is most likely the reason why Automattic has gone the (inevitable?) path of selling user data for SALAMI training, both from Tumblr and their WordPress hosting services.

(You may remember this being what I predict will happen with BS too. And yes, Tumblr showing that “simple” funding sources don't work at that scale is one of the reasons I'm sure it'll happen with BS.)

Now here's the interesting question: does the Fediverse offer something that could compete with Tumblr?

The answer is yes, and in my opinion Friendica is the closest competitor.

One of the main features that used to characterize Tumblr over its competitors when it got released was that it was a microblogging platform2 with social media features. In this sense, it could be considered a competitor for Facebook rather than, say, Twitter (timeline: Facebook launched in 2004, Twitter in 2006, Tumblr in 2007), due to the wide variety of content types supported (text, photos/audio/video, quotes, links, and even chats) and the possibility for users to interact with each other's post by liking, commenting or reblogging (with or without additional context).

What set Tumblr apart from Facebook was a preference for the short-form content that gave it the name, but most prominently (aside from, of course, not being designed around the collection of personal information) the customizability of the blog's interface, something that took a page more from MySpace and most importantly the blogosphere (including WordPress, that had introduced theming in 2005) than from its aseptic competitor. It's not unlikely that this was part of its appeal (compared to the “serious, professional” Facebook —the role now absolved by LinkedIn, arguably) and potentially will contribute to its demise as well, in a world where this kind of creativity is stymied by the uniform “professionalism” needed to cater to advertisers et similia.

So why would Friendica be the most likely competitor to Tumblr? The platform is more intended as an alternative to Facebook, and its default theme may even remind someone of the classic Facebook look. In terms of functionality, it covers by and large the same feature set (my understanding is that Tumblr's Ask is the only one missing, but it offers more, such as event planning). One of the winning features of Friendica is its extensive interoperability support: in addition to ActivityPub and its OStatus predecessor, it also “speaks” the diaspora* protocol (probably the only other platform that still supports it, today), it can ingest RSS feeds, and —most notably in this context— it supports a number of proprietary networks including Tumblr via specific addons.

(Yes, you can follow Tumblr accounts from Friendica; yes, you already need a Tumblr account to do that: it will act as the “bridge”; yes, there is some support for cross-network interactions.)

There is one important feature missing in Friendica, arguably the most important feature in the comparison with Tumblr, the one that truly set Tumblr apart as a “modern” social network: customization.

This is actually a missing feature in the modern social media landscape in general, both in the corporate-controlled social silos and across Fediverse projects, although there are likely different reasons for it —but I'm not going into detail on the corporate silo loss of personality and aesthetics (not just UI) convergence here.

A “custom” Fediverse?

Fediverse server software does offer some degree of customizability, but generally only at instance level: if the users are afforded any kind of control at all for “their” profile page, it is at best some amount of custom CSS: as powerful as this can be in capable hands —especially with the recent progress CSS has made— it's still far from the freedom that could be enjoyed in Tumblr. So, unless you're on a single-user instance (something that e.g. GoToSocial is particularly suited for), your customization option are quite limited.

There are obviously security concerns (among others) that motivate such an approach, since more flexibility than just “changing some colors here and there” requires giving users control on the HTML and potentially (inlined) JavaScript that will be presented to the viewers, that may pose a threat both to the instance and to the visitor, via malicious scripting or template language abuse.

(This isn't unique to Fediverse software, by the way. WordPress itself is infamous for its themes and plugins being attack vectors. There's a reason why bots keep trying to reach the non-existent wp-login.php and xmlrpc.php in the root of my (static) websites.)

There are obviously ways to limit the effect of this, such as using very restrictive (possibly custom) templating languages, (while we're talking about Tumblr, this is the syntax they use: it could be adapted for other platforms too, if there was an interest for this), even though this wouldn't solve for example the issue of profile pages with, say, a client-side cryptocurrency mining script.

The bigger question however, and I think this is also true for the corporate silos, is: would something like this still be a “selling point” for a social media platform in this day and age?

This isn't even just a matter of emulating the Big Tech “solutions” (platforms like MissKey can do some pretty crazy stuff compared to the corporate silos, for example, the least surprising of which is a “Cat” profile setting that will make your profile appear as belonging to a cat on supporting clients, across the network). In a federated social network, you'll be led infrequently to visit other people's profile pages directly: what you will see most of the time will be your server's rendition of their (possibly incomplete) profile information. (Ironically, centralized network would have an upper hand in providing “custom profile look” as a feature, since visiting a user profile would always be “local”, but instead choose to copy each other in a rush towards an indistinguishable anonymity of ensnaring boredom.)

Moreover, this is before even considering that many (most?) people don't even browse the web directly, but will peruse their social media via some mobile “app” that will provide its own interface through which all content will be presented. (Yes, this is the same issue that allegedly prevents Automattic from restoring Tumblr's “porn rights”, if you remember.) When would a customized profile or “wall” even be viewed?

It's not easy to argue that individual profile customization should have high priority, especially with the more pressing technical issues that plague the Fediverse, but on the other hand there's an argument to be made that the Fediverse platforms, more than anything else being developed today, has the opportunity to look back at the lessons from MySpace and Tumblr, remember the impact they had on the generations approaching the web in those days, and recreate a similar experience for the new generations, giving them an opportunity to learn something about web development in a (hopefully) less daunting (to beginners) environment than other approaches to the IndieWeb.

And of course, for this to be “valuable”, it would first require better cross-instance interaction, so that e.g. visiting the original profile page or public timeline of another user can become an experience as smooth as browsing local profiles and timelines. Better user agent support could help a lot here, but can we even count on Mozilla to do the right thing here, after they quit the Fediverse and fired their advocacy division?

But that's a discussion for another time. Here's a parting thought meanwhile: will we get a federated alternative to Tumblr earlier by adding “Asks” and customizable profiles/​“walls” to Friendica, or by expanding the WordPress ActivityPub plugin so that WordPress gains support for reblogs, favourites and other social features that are currently not represented there? (And would Automattic even allow that?)