Hacker News Reader

Comment by tptacek | Hacker News Reader

It's also unethical to make promises to users, like privacy for posts published to a semipublic social network, that software can't possibly keep.

It's going to be interesting watching the norms of the fediverse collide with the expectations of ordinary software users. Mastodon and ActivityPub have a lot of promise as an alternative to centralized social networks, and there are good reasons to use them even if you don't buy into the ethos of the fediverse. If Mastodon is going anywhere, those users will soon drastically outnumber the keepers of the fediverse norms.

View on HN

On the one hand it's bizarre from a technical standpoint to insist that Mastodon can't search or quote-post, when the software is federated and open source (so I can change my own instance to enable search and quote-posts), and when an API is available (so I can roll my own search index without even running my own instance).

If you don't want your stuff archived and indexed by strangers you don't trust, then don't post your stuff in public to a federated network operated by strangers you don't trust.

On the other hand it's not really a technical problem, it's a social problem, contrary to everyone who will try to solve E2E federated group encryption in this thread.

It's a social problem because if I fork the project to enable search and quote-posts, any instances running that fork will get defederated by the community that doesn't want search and quote-posts.

Maybe the Eternal September of new people will want these features, and a fork will become more popular than the original branch. The people will decide what they want and vote with their instances.

As for me, I want self-search, so I built my own full-text search service that indexes everything I post and everything I tag as interesting. I've opened it up to a few friends, but we won't publicize it since it would be a pretty unpopular thing at the moment.

Hmm? Masto comes with decent self-search now, you don't need to build it.

Yes, Elasticsearch is an option, but many (most?) instances haven't installed it, including mine.

And even when it is installed, it doesn't do what I want.

In addition to my own posts, I want to index the public posts of those I follow, and I want to organize by topics that aren't found in the text of the post, and I want to "follow" search terms similar to how the latest Masto allows following hashtags.

None of this is available with Elasticsearch even if it was enabled on my instance.

[Edit: I misunderstood what the parent comment meant. My apologies. My original comment's text is left intact below, but it is inherently misdirected because of the misunderstanding. It still can be, of course, argued that a proposed solution is not realistically attainable - it's quite a tectonic shift in the collective psyche. But the parent is correct.]

I'd disagree. It's fundamentally a [techno]logical problem - a flawed design that only "works" because it's not happening at scale. The pen is mightier than a sword, and ink lasts longer than memory. Those millennia-old maxims still stand very true.

It guaranteed to get only "worse" as the system scales up and gains more "bad" actors that don't conform to the behavior. And it never will be solved socially (only mended) just like email spam was never socially solved, despite giant collective efforts like RBLs, community spam filters and IP reputations. Only shoved under the rug or plastered over, to be repeatedly revealed when a next drama happens.

I'd be happy to be wrong (if you know that I am - please give me an example), but I think that social and legal methods are never the solution, they don't ever fix anything (unless we're talking about humanity radically changing - then, maybe). They provide certain safeguards against honest people not accidentally screwing up, they have chilling effects, and they provide some recourse after the trust is broken - but they never fix the core issue.

Sometimes there is no way to change the design. Like - as a possibly silly example - we can't fix theft or murder, so those are social problems because we don't have technologies (such as achieving post-scarcity or immortality) to fix them. But the information security here is essentially math problem and it can be fixed.

Of course, whenever it will be is a matter of cost. Chances are, no one would ever do it, because it'll be too complicated (=costly) and the social approach would be good enough. Like how the world can't even switch to the IPv6. But, again, something like that doesn't really solve anything for real, merely lowers the frequency of occurrences to a tolerable level.

> but I think that social and legal methods are never the solution

Just to be clear we're talking about indexing + searching what people post on public internet feeds?

IMO there are only social solutions to this problem. Any legal one would probably kill any federated service that can barely compete with centralized ones as it is... or (more likely) not be enforceable in the first place.

There's two social solutions:

1) don't post things you don't want recorded forever on public networks

2) stop treating every mistake humans makes in their personal history as deal breaker for life

The first is a matter of personal responsibility, which is not a very popular concept in certain communities popular online. The other is more rational and practical but is still a very serious problem in western culture. Countless people have been 'cancelled' (or attempts have been made) based on 10yr old internet posts the posters themselves denounce.

If you want technical solutions use Signal, not a federated Twitter competitor.

> Just to be clear we're talking about indexing + searching what people post on public internet feeds?

Yes. I've only brought legal here, as there is some movement towards using legal frameworks for privacy ("right to be forgotten" etc.), as I totally misunderstood what kind of social solutions you propose.

> There's two social solutions: [...]

D'oh! Now I see. And I'm honestly sorry, as I completely misunderstood what you meant saying it's a "social problem".

I wholeheartedly agree now - either of those would be an ultimate solution. Not just a solution - I also do believe that this is how things should be.

It can be argued how realistic it is - I mean, it's quite a giant change to how society sees things. But I hope - at the grand timescale - it's well achievable. Many societies had drastically changed during the 20th century, so who knows where we'll end up...

Yeah, agreed, the "right to be forgotten" legal could be very expensive to enforce on a federated internet. It's only feasible when Google is all you have to deal with. Even then it's a big hassle.

Search and quote-post are very different in this regard.

Quote-post can be a problem because it encourages bad behavior among people who aren't actively malicious. If the most popular Fediverse software doesn't support it, there won't be a widespread quote-poste -> pile-on phenomenon even if a few servers support it without getting widely blocked.

Search becomes a problem when a group of bad actors use it to harass and dox people. It doesn't require widespread adoption, just a small number of bad actors with access to an index. From what I've read about previous organized far-right harassment campaigns, I find it likely that bad actors are already indexing Fediverse posts, but not publicly talking about it. Excluding search features from the most popular software will not help with this. Blocking servers that admit they're indexing posts for search will not help with this.

> there are good reasons to use them

What is one of them?

The medium is better for expressing thoughts (you can write a decent-sized blog post in a single Mastodon post). It has most of the effective bits of Twitter, and many of the effective bits of the peak, RSS-soaked blogosphere. If the tooling gets better, I can see it being better than the golden age of Google Reader.

It also makes sense that you can stand up your own Mastodon, in the same sense as you could stand up WordPress back in the day. Most people won't, but the fact that you can, and that the ability to do so is baked into protocol, keeps development interesting. It also makes it more interesting for third party clients, which Twitter has spent the last decade trying to eradicate. So that's a second (and probably third) reason.

Perhaps ironically, these are also reasons that I think the norms of the "fediverse" aren't going to matter in the long term. I think the software approach is important, and the notion of a single "decentralized social network" is not.

I thought you were giving reasons for "ordinary software users" to go out and use Mastodon in its current form. I was hoping that perhaps their shitshow onboarding UX got updated to accommodate the influx of users over the past few weeks. Nope-- they just changed the "Create Account" link to "Apply for Account" in their hall of instances. :(

But regarding Mastodon as a potential alternative to the the current centralized social networks-- fair enough. I even saw recently that Ted Unangst even made a stripped-down Mastodon server in Go which I'd like to check out when I get a chance.

Yes: there are reasons ordinary users might prefer Mastodon to Twitter; as I said:

* You can write full posts in Mastodon

* You can own your own data, run your own service, or select from a variety of hosted options

* The ecosystem encourages good third-party clients and tooling

The tooling isn't there, but then, I don't think anybody expected the events of the last few months, so I'm inclined to give it a quarter or two. As a casual user, there are ways in which Mastodon is superior to Twitter already, and most of my problems are with the clients; Tapbots can't keep up with the beta demand for Ivory, and there's 2-3 other promising clients, so I'm optimistic.

I wrote Mastodon off for years, but: it's pretty good? I miss the blogosphere, and Mastodon is a step back towards it.

To add on to this, remember that there is a Fediverse plugin for Wordpress as well.

And lots of other popular sites, like Tumblr, are adding Fediverse support as well.

And Drupal

What about the inherent bubble-y nature of Mastodon?

Do you see it as beneficial to have a series of large groups of balkanized servers optionally interconnected rather than a general public network everyone contributes to. I hate the label of "public square", because Twitter should already be opt-in in terms of which content you see based on who you follow, even though it's one big network (ignoring the @jack era AI-feeds which fought this idea).

A key difference in Twitter vs Mastodon UX is that users would have to make different accounts for each "Balkan country" server they join. If one server blacklists or is blacklisted they won't be able to participate in others. And this has implications well beyond US polarized politics once it reaches global scale.

I perceive this to be the biggest critique people have of Mastodon and I simply don't care about it at all; it just passes directly over my head. My guess is that it's a consequence of people mostly "belonging" to community servers, with defined (and evolving) community norms. Blog hosts didn't have those, and I don't think Mastodon/ActivityPub servers will for long either. It's going to get easier to run your own server, and there are going to be hosting providers that do that work for you; ultimately, "Mastodon instances" (or some pin-compatible equivalent) are going to be commodities, like Blogger blogs.

I'm on a server (infosec.exchange) that I like just fine for now, but I'd expect to be publishing on my own server within a month or so. At that point, what's the "bubble"? It seems like exactly the same situation as RSS, except I'm not depending on Google to keep Reader going.

I'm not bullish on "the fediverse" as a federation of communities with some global norms (like religiously posting alt tags on images) and lots of local norms (like never posting AI art, or things like that). But having actually used it now, I think the underlying technology is a lot smarter than I gave it credit for, and I think it stands a decent chance of being important.

Right, "bubble" is probably a loaded word to use here.

I personally would prefer going back to the "old" internet where we had niche forums. Mastodon provides that while letting forums connect with a wider set of communities.

I guess my only critique is that it's not often pitched this way. But it will become apparent to most people later on.

The "connected communities" thing doesn't resonate with me. To me, the promise of Mastodon is just integrated publishing and RSS-style aggregation, plus a set of conventions that captures most of what people did with Twitter. I think it's a weird quirk of the evolution of Mastodon/ActivityPub that I'm on someone else's server right now; it's a little bit like being on someone's huge group blog. Group blogs make sense, but not at the scale Mastodon communities operate at.

I think that if Mastodon sustains the growth they've seen over the last few months, then a year from now, it's going to be less common for people to primarily publish on communities like journa.host and FediScience, and more common for them to publish from huge Blogger-style community services --- with very few weird norms, much more comparable to 2021-era Twitter than to Mastodon.social --- or their "own" instances, just like everyone does with blogs.

> Group blogs make sense, but not at the scale Mastodon communities operate at.

That's interesting. Like I said, this bit will become more apparent to users as it grows, since in the current form that's all it really is (a few sets of big networks). I think you just sold me on the general idea of Mastodon.

Back to your earlier point, it might really come down to the UX issues around communicating on Mastodon itself, if it's going to pull this off. But that doesn't mean there won't be another service that can in the future. Otherwise that's the big thing that scared me away, since I'm a hyper-critical of UIs.

> and there are going to be hosting providers that do that work for you

There are hosting providers doing that work for you, though several of them are overwhelmed too at the moment. There'll be more.

I run my own, and it's not hard but it's certainly still too much hassle for casual users, so I think there's definitely space for more hosting services.

> I'm not bullish on "the fediverse" as a federation of communities with some global norms

Yeah, a lot of these norms will collapse in the face of a broader user base that want Twitter 2.0, and don't care about a lot of the ideological stuff going on with Mastodon and other current Fediverse software.

> But having actually used it now, I think the underlying technology is a lot smarter than I gave it credit for, and I think it stands a decent chance of being important.

Treating ActivityPub w/WebFinger as a globallly addressable signed pubsub queuing system in itself is very interesting. You can build so much more on top of that than "just" "Twitter 2.0" or Mastodon. Of course there are some examples of that, like Pixelfed, but I think we're just scratching the surface of reimagining services on top of a layer like that.

Human interaction is inherently bubble-y.

Nobody wants to get shouted at by people preaching stuff they don't like or feel harassed by all day long, so we seek together in social groups that by and large are united at least along some axes of interests, and avoid too much talk of the topics we know are controversial within those the groups.

Some of us then sometimes seek broader forums to try to spread our groups ideas. But a lot of people actively avoid those kind of situations. Many others use different identities in different groups because they don't mesh well.

"Balkanized" servers are only balkanized to the extent people wants them to be (yes, admins can ride roughshod over users, but users can vote with their feed and migrate their accounts and follows elsewhere). Most people want to be widely federated. Some want forums where they can say things nobody else wants to listen to or wants to self-isolate (e.g. Truth Social, Counter Social and Gab all run - or at least ran, not kept up - Mastodon, but either don't federate at all and/or are widely blocked)

> A key difference in Twitter vs Mastodon UX is that users would have to make different accounts for each "Balkan country" server they join. If one server blacklists or is blacklisted they won't be able to participate in others. And this has implications well beyond US polarized politics once it reaches global scale.

If you want to participate in things so offensive to so many people that they're mutually blocking each other, then yes you either have to do that or find an instance that haven't defederated either. Or host/rent your own and not block anything other than what you, yourself choose.

With Mastodon you have that choice. With Twitter you risk getting outright banned site-wide.

> (yes, admins can ride roughshod over users, but users can vote with their feed and migrate their accounts and follows elsewhere)

This is true when you think it's about HN users as the prototypical user. But I'm just imagining having to explain to my mom why she can't participate in [X, Y, Z] set of Mastodon communities, doesn't she already have a Mastodon account? Why does she need accounts for each? Why are her old "tweets", followers, DMs, etc not available on the new server?

This is a much larger barrier than advanced users can appreciate... maybe not right now but it will increasingly be in the future.

Ultimately it's very possible groups of interconnected communities is a superior concept to centralized social networks. But this should be more of an obvious part of the UX w/ Mastodon. It's fine if it's properly communicated AND if porting your old tweets/accounts/followers/etc over to new servers is easy... but even then it's an adoption hazard (people avoid complexity they don't understand).

Currently it's communicated as a Twitter of FB competitor, which may very likely come back to bite them (not to mention the UI sucking).

> But I'm just imagining having to explain to my mom why she can't participate in [X, Y, Z] set of Mastodon communities

"X and Y don't like each other, so they refuse to talk to each other. Think of it like how you can't talk to people on Facebook from Twitter". People siloing themselves is not a new concept for people.

Ironically, the new thing with Mastodon is that you don't need accounts for each of X, Y, Z to talk to people there even if X, Y and Z defederate from each other, as long as you can find "neutral territory". Only for a very small set of instances is this a challenge.

Being used to silos seems to also be behind a lot of the confusion over having to pick a server - people end up fretting over the choice when it matters a lot less than they think. I wish joinmastodon defaulted to just suggesting a random "generic" instance, and hid the rest behind an "advanced users" button.

> Why are her old "tweets", followers, DMs, etc not available on the new server?

Her follows and followers are available on the new server if she moves the account. When I moved to my new instance, most followers had auto-migrated within minutes. It can definitely be made simpler, but given the number of non-technical users I've seen migrate, it's not that big a barrier.

That the old posts are not I agree is an issue - it is likely to get solved, as the objections to migrating those with the rest are minor and stupid (you can migrate those too, but it requires a friendly admin willing to do a data import, or running your own instance, so for all practical purposes most users can't)

> (not to mention the UI sucking).

There are multiple UIs competing to improve this, before even considering the Mastodon alternatives that all can talk to Mastodon (e.g. Misskey, Pleroma), so this is improving rapidly. Personally I use the advanced interface and I think all it takes is some custom css and it'd be superior to Twitter in most ways (the advanced interface is somewhat like Tweetdeck; I run it with custom CSS that makes column widths configurable). Because of the total open nature of the API, there's far more room to get something good here than there is for Twitter, which is fairly awful.

> A key difference in Twitter vs Mastodon UX is that users would have to make different accounts for each "Balkan country" server they join.

Sure, but this is functionally no different from the days of forums when you'd have an account on many different forums to discuss the topics of interest to those forums. Just like forums, they may all be running the same two or three codebases, but the communities are different. The cool new thing federation adds on top of this is that communities can organically merge and split while retaining a basic level of cohesion.

And since they're all running the same protocol, it's straightforward for clients like Fedilab to make multi-account use easy.

> What about the inherent bubble-y nature of Mastodon?

So, I'd kind of question whether this is a real thing in any important way. In the early days of the Twitter exodus it _did_ look like it was going that way a bit, but in practice most instances quickly sorted out moderation to a reasonable degree. There certainly are largely defederated instances, but they're probably not of much interest to most users.

I would have some concerns about what happens if Twitter goes down for a protracted period and there's a much larger less self-selecting exodus, but for the moment at least fediverse balkanisation feels more like a theoretical problem than a real one which is likely to impact significant numbers of users.

One other potential looming challenge, I suppose, is Tumblr. If that does add ActivityPub support, it'll be, in effect, a very big, extremely poorly moderated, yet culturally very important instance, and how instances treat it could provoke a split.

> and the notion of a single "decentralized social network" is not.

I've been running my own Pleroma instance for a month or so and I'm firmly of the opinion that a single network is actually a Bad Idea. The fact that ActivityPub is federated means that it's also prone to abuse from other nodes similar to email spam, which is just about as simple to execute in the fediverse right now as rolling a new domain name. Small-time casual admins like myself won't be able to keep up with a global abuse firehose which hasn't really happened yet to my knowledge but will sooner or later.

The solution in my mind (and what I've already done) is to use a node whitelist instead of relying on blacklisting, which means smaller networks based foremost on interpersonal trust. Which means a constellation of much smaller but also stronger social networks that reflect actual communities as opposed to just the internet at large. This is super easy to do with ActivityPub servers and my prediction is this is the way forward although I expect people will also try to write clever abuse filters and fall into that perpetual cat-and-mouse game since that's what happened with email.

Reply on news.ycombinator.com