Skip to main content


We should not be optimising Mastodon so it can handle more people per server. We should be optimising Mastodon so it incentivises more serves with fewer people.

(And if you take that line of thinking to its logical conclusion, you arrive at the idea behind the Small Web: https://ar.al/2020/08/07/what-is-the-small-web/)

#decentralisation #federation #fediverse #SmallWeb
in reply to Aral Balkan

Food for thought: The bigger mastodon.social gets, the less successful the #fediverse is.

Sadly, the fundamental design of Mastodon mirrors the design of Big Tech (a server architecture that can support hundreds of thousands of “users”) and thus inherits its success criteria.

I feel it’s time we at least started thinking about what the web would look like if we all had our own place on it and what it would take to get there from here.

#decentralisation #centralisation #fediverse #SmallWeb
in reply to Aral Balkan

Wouldn't a nice start be to go from: how can i run Mastodon (or any other) from ones own computer using that as a server.
in reply to Aral Balkan

I agree. I’m new to this and I’m studying the concept, but the end goal is to create my own instance. I’m just thinking with what community/topic in mind.
in reply to Aral Balkan

Optimising #Mastodon = designing flows that encourage people to leave mastodon.social for other instances, not accepting any more new members on mastodon.social, and making design changes that limit how much a single instance can scale.

A single instance that can scale to host hundreds of thousands (if not millions) of people, is not a design success in decentralisation, it’s a design failure. (It’s a design success in #BigTech.)

CC @Gargron

#decentralisation #centralisation #federation
in reply to Aral Balkan

this makes sense, and is incidentally something I was thinking about this morning too.

https://scholar.social/@badri/109273595283228563

Does it say something that I took "we should get people off mastodon.social" as an established fact rather than a new revelation? 😅
in reply to Aral Balkan

honestly I think it would help to make it easier to move from one instance to another, and then make it easier to host your own instance.
in reply to Aral Balkan

I have to agree. Not that we should make it "impossible" to have instances beyond a certain size, but it's better to have more instances than to have big instances that "work". Making the big instance more efficient is only good, if it helps making it easier for people to start their own small instance.
in reply to Aral Balkan

is your position still 1 instance per user, or has it changed over time? I've just seen that toot from 2017.
in reply to Aral Balkan

I wonder resource wise what’s more efficient, less big servers or lots of little ones.

That said there are plenty of other things that waste energy about
in reply to Aral Balkan

@Gargron I disagree somewhat; yes, scaling well is an L for decentralization, but it is a W for the freedom of hosters.

What if you want to use #Mastodon to host your own centralized platform? #TruthSocial is a good example of this; #Trump aside, it's awesome that you can use #Mastodon to both be part of a huge network, make your own network, or make a standalone social site.
in reply to Aral Balkan

@Gargron mastodon.social should not be an instance, but a list with links pointing to instances. For someone used to tw, fb, etc it is not obvious that you can choose a smaller instance without losing function.
in reply to Aral Balkan

@Gargron Sorry, but that's naive in my opinion. When people decide to leave Twitter (yay!) and look for alternatives, Fediverse tech should just work! We only saw a 0.something percentage of Twitter users migrate to Mastodon and even this relatively small amount of users caused problems (and not only on Mastdon.social). Yes, big instances may cause problems in terms of centralization or concentration, but why not focus on the opportunity to show people that Fediverse is working đŸ€·đŸ»â€â™‚ïž
in reply to Die KrÀhenpost

@kraehenpost Maybe it’s naïve, maybe it’s ~ a decade of thinking about the problem ¯\_(ツ)_/¯

(It comes down to your definition of “working.” If we manage to recreate twitter.com on mastodon.social, I wouldn’t define that as “working” but as “failed.”)
in reply to Aral Balkan

@Gargron I was thinking similar thoughts the other day. It should act like a clearing house and entry point - orienting people then encouraging them to move.
in reply to Aral Balkan

@Gargron how do you think the discovery problem should be handled in a small web perspective? The thing Twitter (and Big Tech in general) does really well is to solve that problem, and I don't see a great answer on Mastodon except for directories of individuals... Of which there are multiple in various states of maintenance.
in reply to Aral Balkan

I wouldn't be so categorical.

> A single instance that can scale to host hundreds of thousands (if not millions) of people

means that running an instance for couple thousands or hundreds people is becoming way cheaper, and as a result, more accessible.

This is the exact problem that, for instance, Matrix has. It's voracious, and a nightmare to work with, so just anyone from the street can't afford to host it. Whereas XMPP servers like ejabberd or Prosody can serve tens of thousands of users from a matchbox computer like rPi. So I can be confident that they will handle my needs easy-peasy.

And Mastodon as well is such a beast, that code optimization is way overdue. I think we should welcome it when the software is trying to get better, not bash its creator for it.

@Gargron
in reply to Aral Balkan

@Gargron While I agree with your general sentiment, "...thus inherits its success criteria." doesn't really make any sense to me.
in reply to j

@jay @Gargron Big Tech is designed to scale vertically because their success criteria is continuous (ideally exponential) growth*. It’s that ideology/success criteria that drives their client/multi-tenant server design. When you adopt that design, you inherently adopt the success criteria that created it. And if that ideology/success criteria are not aligned with your own, you have a problem on your hands.

* with finite resources (an interesting one to think about
 đŸ€”)
in reply to Aral Balkan

@Gargron By this criterion, recent developments have optimised a lot! In the last week we've had almost 200k new users, only 6k new users in a month on mastodon.social.
https://fedidb.org/network/instance?domain=mastodon.social
in reply to Aral Balkan

@gargron An arbitrary cap seems, well, arbitrary, but I would like to think there's some kind of strategy for avoiding the #LongTail trap. One #Fediverse health check that might be useful is an ongoing #bot-conducted #survey and periodic reports of the #distribution of #instance headcount.
in reply to Aral Balkan

@Gargron

Is there any server software optimized for single user instances?
in reply to Aral Balkan

@Gargron True. But in order to reach viable scale, people need a single point of entry.
in reply to Aral Balkan

@Gargron

It implies also making interoperability between servers seemless.

https://news.ycombinator.com/item?id=33426897

You can feel the problems even as a tech enthusiast... and get used to it so you don't view it as a big problem.
But normal people who are not tech enthusiasts are less likely to go through that effort and have a more difficult process
in reply to Aral Balkan

@Gargron The mega instances are great for on-boarding as they provide a gentle introduction to the concept of federation. If there's a hard limit we shut the door to new citizens of the fediverse and that's not worth it. There should be a point where dot-social users are encouraged to migrate out. A gentle reminder from the admin, or a mandatory donation to the server maintenance fund after year 1 would keep its user growth in check.
in reply to Aral Balkan

@Gargron It's an interesting challenge to solve. Probably there could be an onboarding flow which distributes new users across instances which somehow qualify, e.g. uptime * user satisfaction index * user retention vector. (Thinking loud here.)
in reply to Aral Balkan

@Gargron as a recent escapee from #twitter it was helpful to have a wide-open mastodon.social that has accepted me until I find my footing and figure out where my people are. Many I meant to follow went to instances that are now full. Having a (temporary) central clearing house might be useful though I see the sense in encouraging it to stay temporary.
in reply to Aral Balkan

@Gargron why not 50k ? Or something similar ? Having that amount of users is already pretty heavy stuff
in reply to Aral Balkan

@Gargron my issue is that the global federated timeline is too confusing and there is zero translation option. At times i had to mute people who mass post on languages i dont understand :(
in reply to Aral Balkan

@Gargron I'm new to mastodon today. I think that makes sense, but it's a pretty big hurdle to figure out which community makes sense on day 1, so I went with the default knowing I can move later. The community belonging concept feels tough to me right now as I exist and want to communicate in multiple communities, and I'm not sure how I would decide to join one today if targeted communities existed for all of my interest areas.
in reply to Aral Balkan

@Gargron These optimizations will allow me (and others) to run an instance as a virtual machine with 8GB RAM that ist still likely to support a few thousand users.
in reply to Aral Balkan

decentralisation is key - at the same I argue there's also a minimum for many people to experience a good interaction without have to 100% curate their followers. So 'small' is good - 'nihilistic' less so :-D (disclaimer: I'm an admin of mastodon.nl - currently ~6.5K users)
in reply to Aral Balkan

Totally agree. Although if Mastodon gains some more traction there will be people leaving mastodon.social because they find it does not fit their needs.

I for one am on my own server because mastodon.social often feels to left for me.

But decentralisation is key. For the good of the Fediverse and people in general.

We are not built for one monolithic community where everyone is the same.

We are all equal, not the same.
in reply to Aral Balkan

Maintaining independent instance is still quite an work, there’s need for some infrastructure as code since all instances are relatively similar
in reply to Aral Balkan

It's not ideal and doesn't fulfil the small web goals I'd like to see achieved but I am trying to set up a personal mastodon server at least. It's a start. Even if we eventually ditch mastodon, at least it's part of... I think you've described it as "decentralising yourself"
in reply to Jennifer Harrison

@GeneticJen Hey, every step is a step and that sounds like a very positive one :)

I think what I was saying was about “decentring yourself” (as in making things that don’t rely on you being at the centre to be successful) but decentralising yourself sounds good too :)

PS. Hugo (@mastohost hosts mine and he’s great, in case you don’t want yet another server to admin).
in reply to Jennifer Harrison

The experience of trying to set up a personal mastodon server and trying to explain small web ideas to others makes it clear to me how important it is that small web eventually "just works". If only you, Laura and others could skip to that step! I believe this is a worthwhile journey you're on though
in reply to Jennifer Harrison

@GeneticJen I’d love nothing more than to be able to skip to that step. And we’re not a million miles away
 but I’m still working on infrastructure :)

(I also wish it was just about writing lines of code. It’s not. It’s about attempting to make it sustainable both in the current environment and in the one we want to move towards, about trying to use bits and pieces of what exists to build a bridge towards where we want to be
 and about sharing every brick so others can build other bridges.)
in reply to Aral Balkan

@GeneticJen (Also, thank you
 “I believe this is a worthwhile journey you’re on” is the best compliment ever. And it’s definitely a journey.) :)
in reply to Aral Balkan

i agree and been wanting to move on to another instance or create one myself.
what keeps me from doing so is that i don't wanna lose my posts in this process (which'd be the case afaik) as i frequently dig up older threads for various reasons.
in reply to ottonafo đŸ‡ș🇩

@ottona Mastodon has a very nifty migration feature (as well as data backup and restore). You shouldn’t have to lose anything, including your social graph :)

See https://docs.joinmastodon.org/user/moving/
in reply to Aral Balkan

@ottona Also see: https://mastodon.ar.al/web/@pludikovsky@chaos.social/109273683628623591
in reply to Aral Balkan

@ottona Well, you lose your old toots. I did this migration process mere hours ago and it was more painful than it should. With a bit of UI/UX work it can be improved though. Ideally it should be a couple of clicks if we want to nudge people towards moving to other instances.
in reply to Juan Luis

@astrojuanlu @ottona Yeah, I went through the migration process yesterday and it was surprisingly painful.

Lost all my Toots from the old instance, and it would not let me transfer my followers because there was some backlog processing taking place at the time.

Really a terrible experience that should have more work done to improve it.
in reply to Nick :linux: :fedora:

@CorruptComputer @astrojuanlu @ottona Ah, sorry to hear that; not what I would have expected.

So the process is currently to manually export your data, run the migration feature, and manually import your data back?

(If so, yes, this is very confusing and I would likely have lost five years’ worth of posts had I not seen this and tried it myself too. I’d expect a migration feature to migrate all my data. The flow should guarantee this.)

CC @Gargron
in reply to Aral Balkan

@astrojuanlu @ottona @Gargron You can manually import your Following list, Blocking list, Muting list, Domain blocking list, and Bookmarks.

Your Followers rely on the old server being able to handle the request, which I had issues with.

Your Toots are not portable at all, and there is a feature request from 6 years ago asking for this over on the GitHub but no work has been done on it as far as I can tell. https://github.com/mastodon/mastodon/issues/981
in reply to Aral Balkan

Mastadon is sadly not fully distributed, there is such a thing like a local timeline.
in reply to Aral Balkan

Exactly. Fediverse is meant to scale horizontally.

This small instance I’m on (<800 users) was unaffected by performance issues during the Twitter exodus, which is the whole point of federation.
in reply to Aral Balkan

If you're interested in a more decentralized system then there are p2p networks like https://scuttlebutt.nz/, you can try it out by using the @manyver_se client. I guess its good to have these options :blobcatshrug:

What I like most about p2p networks is that everyone is a node in the network by default, you don't need to know how to run your own server to have that freedom.
in reply to Rokosun

@futureisfoss @manyver_se Yep, aware of them. They have their issues also. But it’s great that we have all these projects exploring the problem space in different ways. All very valuable :)
in reply to Aral Balkan

.@aral I agree but it's undeniable that the common folk doesn't want and shouldn't need to run their own instance.

A few design choices that might move people towards a more federated Mastodon landscape:

- Have a page that shows you with which instances you interact the most

- Allow moderators to limit or soft-limit new registrations

- Have generalist instances suggest users to move to other communities after a while
in reply to Aral Balkan

as a newbie - it is very hard to find instances, and very hard to find people. Neither of those are architectural issues (ish), they are fundamental to adoption though.
in reply to Aral Balkan

I've been wary of moving to a smaller instance since the murder of Snouts by a bunch of shitposting instances.
in reply to Aral Balkan

Personally, I've been working on making efficient p2p database indexes and finding a way to represent ActivityStreama in them so that we can start bridging between fedi and p2p. đŸ„°
in reply to Aral Balkan

the problem is that as long as deploying an instance is not zero effort (which it never will be), there will always be end users (most of them in fact) who rely on centralisation - simply because they wouldn't be able to do it on their own. turning this around: if you don't optimise for scale, you lock these end users out.
in reply to Aral Balkan

How about, restricting the number of users and referring them to other servers?
in reply to Aral Balkan

the twitter migrants mostly don't want to think about 'instances' and 'federated structure,' they just want to write something fairly short, push the button and have a wide audience of other people see their messages. I understand the concern, but I also think that not as many people would make the switch if we didn't have the big catch-all instances. Overall usage is important, it is a social network. Maybe people can start out there and move after getting more used to the platform.
in reply to Aral Balkan

maybe see a large generic server as Main Street. It can coexist with the smaller streets (themed servers).
in reply to Aral Balkan

@tamitha Related to those though a bit of an aside: I use masto.host and have recommended it to a number of people to bootstrap small instances. One reason is I appreciate masto.host’s Hugo’s stance that he never wants to host a large fraction of the Fediverse. He’s made commitments and promises that at enough percentage he refuses new business to encourage instance diversity. I believe he does encourage large instances to split, too. (Not for technical reasons.)
in reply to Aral Balkan

@nosherwan I reckon big is not necessarily bad as long as the culture is healthy and as long as is it easy for folks to start up their own. Not everyone wants to be a builder.
in reply to Aral Balkan

I wouldn't see it to bleak though. When I arrived here, I welcomed the presence of a bigger server to get started. Soon after I looked for peers (for me in science), got signposts, and ended up at the present server, which is much smaller.

I am sure that many others will have the same journey.

I know you advocate the "small web", and well, such journeys may make people more comfortable with the idea in the medium and long run!
in reply to Aral Balkan

I’m guessing the key difference is that you can always move instance.

This way it can fit everyone - both the specialist and the generalist.

Having large generic instances is also a good way to get started - for example I’m almost sure I’m on the wrong instance; yet I don’t really know which one would be the ”right” yet.
in reply to Aral Balkan

They got the matrix issue... Nothing wrong with optimizing but mastodon.social should also promote other servers when creating an account.
in reply to Aral Balkan

libraries setting up local instances as keepers of the public commons

in reply to Aral Balkan

I've only been here for two days, and came to that same conclusion. I joined a server with fewer users, and things are going well unless I try to interact with the largest server.
in reply to Aral Balkan

Eh, you may change the motivation behind the need to optimize Mastodon's hardware requirements, but that's something that still needs to be done, because the best way to get more servers is to allow Mastodon to run “everywhere”, and the more efficient Mastodon is, the more it can run “everywhere”. But as a consequence, it will inevitably support more people on larger servers.
in reply to Aral Balkan

In fact, if I were setting strategy for making this migration from Twitter -> Mastodon a success, I'd deliberately steer people upfront to less-populous servers with rules that are in sync with the reason most of us are coming here (low toxicity)
in reply to Aral Balkan

@michaelbrooks this is 100% where I am taking myself! I’ve been thinking about buying my own server that literally serves my personal needs. Host my websites, my social accounts, and maybe even my online streaming/video if it can.
in reply to Aral Balkan

The Small Web as described in the article sounds good, but I'm having trouble picturing how to convince people join in.

I don't think anyone today is under the illusion that Google, Facebook and others are altruistic organizations. But the majority of their users accept the intrusions of privacy and lack of control in exchange for simplicity. It takes only a few seconds and no debugging to sign up for an account. 1/2
in reply to Guy Montag

Grandma is fine with Facebook and Skype icons on her desktop. How will you convince her to manage a database?

There are already constant stories about how hackers can break into home WiFi networks, yet I know very few people willing to configure their own WAPs. That sounds much less daunting to me than saying someone has to build their own Small Web. 2/2
in reply to Guy Montag

@Montag Well, first off, Grandma could be a computer programmer for all we know, so let’s say someone who is not versed in the technical aspects of computing (like a brain surgeon, for example). With that out of the way: if they have to configure a database, you’ve already lost.

If they can sign up for it as easily as signing up for Facebook and use it without having to maintain it, you have a chance at winning.

We’re trying to build the latter :)
Unknown parent

Aral Balkan
@shine Exactly :)

Mastodon is what Mastodon is at this point and I don’t see it embracing #degrowth (although I’ll be the first to sing it praises if it does).

I’d love to see more single-tenant ActivityPub implementations (and hopefully they’ll support the migration protocols implemented Mastodon).
in reply to Aral Balkan

the thing is, a lot of people want twitter and in the public consciousness mastodon kind of took place at being the ethical alternative (it helps that is not controlled by musk)

but essentially what (i feel) most people want is:

- a free (as in beer) twitter-like thing (as in: a digital public town square)
- a non-algorithmic timeline (chronological for the win)
- incentives to not turn every discussion into flame wars

the concept of instances is one abstraction step too much for most.
in reply to Aral Balkan

I have to say - I have been very intrigued around starting a server of my own, but I don't really know what I'm getting in to. I'm not used to managing my own personal infra, servers etc. I've done so before when I was a lot younger - but have landed very much in the "these aren't the problems I want to spend time on"-camp.

Perhaps it's a mindset shift? Maybe I'll value that more and take the steps for myself, because I see the importance of small over big?
in reply to Pavneet Singh SaundđŸ€˜đŸœ

@pavsaund The thing is, you shouldn’t have to know what you’re getting into in order to do it. The fact that you need all that technical knowledge to get started is a failing on our part as designers/developers. We’re working on fixing that but in order to do so we must design technology differently (so it can be owned and controlled by people themselves not by others for them).

In the interim, there are folks like @mastohost that make it easier to host the current alternatives :)
in reply to Pavneet Singh SaundđŸ€˜đŸœ

@pavsaund I’ve gradually shifted into that camp as well. Spent years running my own Nextcloud server and got tired of all the tiny things that kept popping up. At the end I bought managed Nextcloud to be rid of them and one day realized: that’s what Dropbox has been all along.

I’m not saying Nextcloud should not exist because there’s Dropbox. My point is that running/owning alternatives to Big Tech should be as hassle free as using Big Tech itself.
in reply to Simeon Nedkov

@pavsaund I think we should promote “hassle free Mastodon” experiences instead of “run your own server”. The latter screams “work I don’t want to do” at me.

The underlying implementation can stay the same. It’s a matter of packaging the thing, meaning: you tell customers that they pay for and create a Mastodon account but under water you spin up a new instance for them.

In this scenario you are creating personal servers for everybody without them knowing (because they don’t care).
in reply to Simeon Nedkov

@simeon @pavsaund You’ll get no argument on that from me ;) https://ar.al/2020/08/07/what-is-the-small-web/

(Apart from perhaps that we should also be looking at designs that aim to make doing what you describe as easy as possible. It’s much easier to set up a system that’s explicitly designed to serve one person than one that is designed to serve 1-100,000 people. The complexity involved is orders of magnitude less in the former.)

#singleTenant #SmallWeb #decentralisation #federation
in reply to Aral Balkan

@pavsaund Agree and I love the/your Small Web concept.

Would a Mastodon-instance-per-person qualify as Small Web or are there other things to consider (as well)?
in reply to Simeon Nedkov

@simeon @pavsaund Thanks! A mastodon-instance-per-person would likely qualify but is unrealistic and would be both overkill. Mastodon’s architecture mirrors that of Big Tech. It’s a server capable of supporting one to several hundred thousand people on a single instance. That’s very different from a system designed to support just one person. The latter has orders of magnitude less complexity and requires similarly fewer resources.
in reply to Aral Balkan

@simeon @pavsaund Running Mastodon on an instance of one is like buying a hotel to live in just for yourself. Sure, you can do it, but it’s going to cost you. In the Small Web model, we all own our own cottages and it’s the collection of cottages that makes the villages.
in reply to Aral Balkan

this is the logical conclusion for human interaction and civilization in general imo. Smaller communities lead to less depersonalization, less "othering", more caring.

Aral Balkan reshared this.

Unknown parent

Aral Balkan
@max And yet, oddly, we all want to live in our own houses
 and a collection of houses creates a village
 đŸ€”

;)
@max
in reply to Aral Balkan

We should have a few BIG landing-pad servers for new people, always.

"Find a nice server" is an impossible first step.

"Make an account on any of these big servers, it doesn't matter which. Later when you notice all the most interesting people hang at x.y, move there. No biggie." is what the advice to new users should be, IMO.

#federation #fediverse #decentralization
in reply to stop genocide in gaza

@nikodemus Idea: the landing pads don’t have to be servers. They could be the start of the process of finding a server. Thereby, you don’t privilege any one server by having it be the landing pad.
in reply to Aral Balkan

I'm not sure I understand.

If someone needs email it doesn't matter which provider they get, as long as they pass certain minimum criteria.

If someone needs Mastodon, unless they go on a big server it DOES matter where they go.
in reply to Aral Balkan

yes agree - hopefully this can be the beginning of the end for the big web giants
in reply to Aral Balkan

as a new mastodon user, this is really helpful! I posted yesterday asking if there are any downsides to hosting my own single-account instance. Is that too small?
in reply to Zachary Neal

@zpneal Nope; I’ve been doing it since 2017 :)

Initially set it up myself but I don’t need another server to maintain so very happy to let @mastohost do it for me :)
in reply to Aral Balkan

But we do need to make it easier for people to get started. It's confusing to have to choose an instance before you can gain experience.

Perhaps a suitable general server could be "suggested" for people so that the don't all sign up to the "canonical" instance? They can always switch later, in their own time, when they're in a position to make an informed decision.
in reply to Aral Balkan

agreed; I think the limit ought to be around 50+-10. Working on a router now. https://hachyderm.io/@mitka/109269768449411301
in reply to Aral Balkan

Do you think that in this case, mastodon.social is the gateway to the #fediverse? People will join .social, hang around a bit, get comfortable, decide if they want to stay, and then either move on to more specialised instances, or try and setup their own? (That's my vague plan.)
in reply to Aral Balkan

and if you take that line of thinking even a bit further to its _real_ conclusion
 you arrive at the idea behind peer 2 peer networks— where every #fediverse client is a server.

p.s. i'm not saying this is or should be the goal, just pointing out how lines of thinking don't always end where one thinks they do ;-)
in reply to Aral Balkan

One thing that is needed is a beginner-level #tutorial on the subject of "dealing with the #hosting industry." I don't know the ins and outs of procuring any kind of hosting other than #LAMP stack, which, as far as I know, won't accommodate #Mastodon.
in reply to Aral Balkan

This is short-sighted in my opinion. Mastodon should be resource efficient enough that everbody can host it from home on a Raspberry Pi. Purposefully wasting resources will not lead to more decentralisation.
in reply to Aral Balkan

I’d like to do this but is there a way to preserve existing connections? My impression was no.
in reply to Aral Balkan

If I may politely disagree? Many of the same optimizations that could help mastodon.social have more users also make it so that running a different instance is cheaper.

It's worth differentiating "scaling by being able to use EVEN BIGGER hardware" (not so awesome in this view) vs "scaling through optimizations" which is a strict win. It's great if you can host a good-sized instance on a $100 machine.
in reply to Aral Balkan

I disagree, if that's what the people want, that's what you should be addressed
in reply to Aral Balkan

This does make sense to me. As I learn more about the #fediverse I'm becoming more interested in the smaller instances. I, for one, welcome this mindset shift.

The idea of setting up my own instance is appealing... but I don't think I fully understand the required investment yet.
in reply to Aral Balkan

What if newly created accounts get assigned to random servers by some algorythm or by giving people options considering characteristics of the account ( language, topics of interest, country/continent, etc )? Maybe even having the same account mirrored in more than one server?
in reply to Aral Balkan

I agree, but the hard part is the vast majority of people aren't techy enough for this.

As such...I do wonder if there is a hole in the ecosystem for an easy "create your account on one of these pre-configured domains" site, kind of like freedns.afraid.org does it.

That way users would maybe be spread out among those pre-configured instances rather than the singular monolithic one.

(scrot attached for reference)
in reply to Aral Balkan

I think we need Decentralized Identity to finally happen more than anything. There's no reason why, in this age, the Fediverse still works like E-mail. With all our data and identity stored on a single server identified with a traditional user@domain URL.

Once you can use your private key to interact with your identity on any server at any time, that's when #decentralization can really kick off. Mastodon.social down? No problem! Just seamlessly continue from another server...
in reply to Aral Balkan

While I agree that decentralization is the goal, I think optimizing both the code and the network is necessary to drive down the cost/user and improve performance (~experience). Some of that is technical, and some is social.
in reply to Aral Balkan

I haven’t found the smaller communities that are centered around the things I’m interested in. Even if I stood up a server, the people I wish were in the fediverse are sufficiently non technical that convincing them to try it and onboarding them is a big lift.
in reply to Aral Balkan

@shine I kind of had this on my to-do list for years to make my own website which already has notes, photos, podcasts, events, likes, etc. ActivityPub aware like https://aaronparecki.com/ does which would bridge the IndieWeb with the Fediverse. Some day in the future perhaps.
in reply to Aral Balkan

This thread really opened my eyes and as an instance owner I'm going to put some thought into how to tackle this problem.

I appreciate you taking the time to point this out to better improve the Fediverse.
in reply to doomy 🩀

@doomy Because they’re diametrically opposed :)

The more complexity you add to handle vertical scale the harder you make it for individuals to self host.

All that said, Mastodon is what it is. Likely too late to change course now. But hopefully we’ll see more interest in single- tenant fediverse clients going forward.
in reply to Aral Balkan

love this. set up a single 6$/mo instance for me, karlspace, and I'm loving it so far
in reply to Aral Balkan

I agree, but with that comes a need to re-engineer how hashtags can get federated. Smaller instances scoop up less hashtag content from other servers.
in reply to Aral Balkan

i've had thoughts about reworking the old concept of relays into a sort of overlay network -- joinmastodon could essentially work like an irc network in having servers that users connect to round-robin, but discovery could be handled at the central gateway service. that way if the central service goes down, you can still use your server, you just get a less complete view of trends and it becomes slightly harder to find profiles. then it should matter less where you sign up.
in reply to Aral Balkan

One thing that would help me move off mastodon.online is a way to migrate accounts that works in an adversarial environment. So, “I don’t have any access to my old account, but I do have this certificate.”

I like the idea of moving my account around freely between small instances, but I feel like I need to be picky because I don’t have the energy to deal with them shutting down or having weird admins.
in reply to Aral Balkan

Smaller instances are more likely to die out. It's better to have fewer ( in dozens or so) instances that are more reliable, IMHO.
Thousands of servers just confuse people and make the network less efficient
in reply to Aral Balkan

@max We want to drink a beer in a crowded pub, never in an empty pub. We live in cities or villages and not alone in the countryside. People are social, we want to belong to a group. That's why people flock to mastodon.social.

It doesn't help that most of the media has been reporting and promoting Mastodon as an open-source Twitter alternative without mentioning the fediverse or it's decentralised nature.
@max
in reply to Aral Balkan

TBH I think Mastodon might have a problem if everybody picks the "same" server because that is the place "to be". The distributed nature of the network should be invisible IMHO.
in reply to Aral Balkan

I would modify that principle slightly.

Yes you want fewer people/server and many servers. BUT, you also want those servers to be able to serve as backups to a few other small servers, too. A decentralized torrent like redundancy to minimize impact of single node failure. THAT's old school small web. :)
in reply to Aral Balkan

Moving from a centralized mindset to a decentralized one isn't trivial. I agree that we should foster more servers with fewer people, and here in Portugal many have registered in a local instance, but what exactly is "fewer"?
And aren't the more generic servers always going to be more crowded than the niche ones?
in reply to Aral Balkan

I don't think Mastodon should ignore the needs of large instances. If a community is built up, the platform should support that.

I agree that it shouldn't cater only to large instances, but saying that Mastodon shouldn't optimize for that use case also isn't right either imo.

The platform should support however you want to use it. Self-hosted private instances, small communities, and large communities should all work well.
in reply to Aral Balkan

if you bring this idea to its logical conclusion you get myspace with each user getting a server and the feed becomes your wall...

Where does the slider land?
in reply to Aral Balkan

way back when Opera tried to do a similar kind of thing with the browser running a few services as servers tied to your username. Worked pretty well.
Could sync browsers, leave notes, run a little blog.
As long as you turned it on, the browser connected to their server and it functioned like dns to direct you directly to the endpoint.
in reply to Aral Balkan

I’ve long felt it’s a good thing to have a limbo instance to collect the confused masses as they migrate so they learn and form a basis for understanding it. People who are ready to get it already go to smaller instances.
in reply to Aral Balkan

That was a really interesting read. And it's a very compelling notion, the small web.
in reply to Aral Balkan

This partly made me decide to create my own instance of Mastodon. It's really nice how you can have your own little corner but still connect with everyone else.
in reply to Aral Balkan

As someone who makes a point of having my own instance on here, I agree.
in reply to Aral Balkan

Thank you and everyone who added their thoughts in this discussion. Mastodon is such an amazing breath of fresh air.
in reply to Aral Balkan

I mean, Mastodon.Social is here for those who want a more traditional social media experience. I too am generally pro #SmallWeb, but most of us newcomers are mainly just looking for a twitter replacement, and so Mastodon.Social has to grow to accommodate the new crowd,
in reply to Aral Balkan

my only issue with this is when instances close. Yeah, most times they give you time to migrate your follows, but your toots, images etc die with it. There should be a way to migrate all of that too.
in reply to Aral Balkan

I've already seen a couple of mentions of new servers (instances), plus one person who decided to create a personal server.

Like... yeah, I've been on Mastodon for four years and it never occurred to me that I'd want my own server. I feel like an underachiever.
in reply to Aral Balkan

I wonder if it would be interesting to have a private Mastodon instanca running from inside one of Tim Berners-Lee's SOLID pods.
in reply to Aral Balkan

the more (small) servers people start: the population on Mastodon will get more 'diverse'. Wouldn't that cause a bigger load on each server as it needs to maintain more federated connections? Or does it work in a complete different way? (just getting curious 😉 )
in reply to Aral Balkan

a good start on this front would be an official Docker image. As someone predisposed to self-hosting, my expectation is that a mature service has a well-maintained (multi-architecture) Docker image ready to be deployed, configurable with environment variables.
in reply to Aral Balkan

The biggest issue I had setting up a server was to provide a way to send emails. The complexity of the software itself, while no laughing matter, is a problem which can be solved nicely with docker.

@aral
in reply to Aral Balkan

tinyweb. too small to, you know, overthrow things and attempt egregious badness, would be kinda nice.
in reply to Aral Balkan

Been reading about your Small web. And am very impressed with site.js. Installing tomorrow to look around. Does or will it have a plugin architecture?
in reply to Aral Balkan

it sounds like you're not necessarily against big servers, but cautioning thay they might push the direction of the protocol or make it harder to cultivate an ecosystem of smaller ones. i would say it needs to be easier to start smaller ones and i'm not sure how that becomes zero-effort or within reach of less technical people. systems like these https://easyindie.app are useful for making it simpler but it's still requires server admin; is smallweb possible without it?
in reply to Aral Balkan

Content warning: Mastodon meta

in reply to Aral Balkan

Yes. I am often asked whether I want to grow toot.bike real big. I guess no. I'll probably just close applications at one point when one dedicated machine can still comfortably handle everything.
⇧