Live data from Hacker News

Mastodon 2.5 released

blog.joinmastodon.org

71–80 of 105 posts

Re: Mastodon 2.5 released

#71
post #67
post #30

Earlier quoted context omitted.

> Have you tried Mastodon? Yes. > You still pay someone else to run the server (or not, if you prefer), you can just choose who you pay, and which server. Sending toots does not require that you set up your own infra, re: the email example. Right, you either pay with your time or your money, as opposed to FB for which you pay nothing at all. Specialization of labor reduces costs, federation increases them. > This is…

If your criteria for success is a large centralized for-profit monopoly, then of course Mastodon will never be successful, since that is the definition of failure in the context of the federated network. And since we have that now, you of course should be perfectly happy with the status quo.

No, my criteria for success is a large network that is meaningfully better than FB, which we absolutely do not have and Mastodon will not deliver.

Re: Mastodon 2.5 released

#72
post #50

Earlier quoted context omitted.

> This looks like a networking protocol, but this is actually process for electing the main node in the network, which will eventually revert it back to the mean of centralization. Let's see how Mastodon scales. I don't entirely understand why you assume that one node will come to dominate such that it will be equivalent to a centralized system? The term "federation" is borrowed from political science for a reason. J…

> I don't entirely understand why you assume that one node will come to dominate such that it will be equivalent to a centralized system? One, or a small number, and because this is the history of these types of things. > The term "federation" is borrowed from political science for a reason. Just like in the USA, multiple federated states who share broad-strokes policies maintain interoperability while also maintaini…

> One, or a small number, and because this is the history of these types of things.

It's worth teasing this apart.

There is a befuddling belief on the part of some decentralization advocates -- and of some decentralization detractors! -- that decentralization must imply an even distribution of power or influence.

That strikes me as unrealistic in the extreme; meaningful networked ecosystems essentially never converge to uniform distribution. Power laws are the norm, not the exception.

I try to avoid "decentralized"'s linguistic baggage by using the term "federated" instead. Email is federated; Gmail is its behemoth... and that's okay.

(As for Mastodon: today, the majority of the network's traffic resides on the largest 5 of 3500+ publicly announced instances. It's power laws all the way down.)

Re: Mastodon 2.5 released

#73
post #36

Earlier quoted context omitted.

Right email is the exact model I'm basing my thinking on. You don't see people running their own email servers, you see 4-5 companies in the world gobbling up the VAST majority of the network, even with weak network effects.

I've been thinking about this recently myself and also used email as my go-to example -- and it's not lost on me that, if Mastodon really follows that path, we're likely to end up with a few immense instances that provide added services, either in exchange for a small fee or for ads (or, let's be honest, through data mining). I suspect one of the biggest potential places for improvement will be providing a semi-forma…

Hey thanks for the really thoughtful response!

> The current approach presents itself as a kind of socialist/libertarian communal utopia, and it can be depending on how a given instance is run -- but in practice, each instance is a monarchy with absolute power over not only your presence in the kingdom but which other kingdoms you're allowed to talk to. Your social graph can be broken at any point because your admin has decided, for any reason, to stop federating with an instance you have followers on, or vice-versa. The net effect is that you can be punished for transgressions, real or imagined, that you weren't a party to (and might not even be aware of). And I don't think that has a technical solution, just a policy one.

This is so spot on! It's easy to say Mastodon is solving all these problems, but the reality is that it's just not big enough that it has faced any real challenges, and the challenges it really faces are mostly philosophical, not technical. AFAICT it hasn't scratched the surface on the philosophical governance issues.

Re: Mastodon 2.5 released

#74
post #50

Earlier quoted context omitted.

> I don't entirely understand why you assume that one node will come to dominate such that it will be equivalent to a centralized system? One, or a small number, and because this is the history of these types of things. > The term "federation" is borrowed from political science for a reason. Just like in the USA, multiple federated states who share broad-strokes policies maintain interoperability while also maintaini…

> One, or a small number, and because this is the history of these types of things. It's worth teasing this apart. There is a befuddling belief on the part of some decentralization advocates -- and of some decentralization detractors! -- that decentralization must imply an even distribution of power or influence. That strikes me as unrealistic in the extreme; meaningful networked ecosystems essentially never converge…

> There is a befuddling belief on the part of some decentralization advocates -- and of some decentralization detractors! -- that decentralization must imply an even distribution of power or influence.

No, it's that if you observe that power laws are the norm, as you stated, then you start to see decentralization as a process, not a state of being. The idea that a network (or governance structure) "is decentralized by design" makes no sense. If you know decentralization is temporary, it's hard to take Mastodon etc. seriously.

Re: Mastodon 2.5 released

#75
post #74

Earlier quoted context omitted.

> One, or a small number, and because this is the history of these types of things. It's worth teasing this apart. There is a befuddling belief on the part of some decentralization advocates -- and of some decentralization detractors! -- that decentralization must imply an even distribution of power or influence. That strikes me as unrealistic in the extreme; meaningful networked ecosystems essentially never converge…

> There is a befuddling belief on the part of some decentralization advocates -- and of some decentralization detractors! -- that decentralization must imply an even distribution of power or influence. No, it's that if you observe that power laws are the norm, as you stated, then you start to see decentralization as a process, not a state of being. The idea that a network (or governance structure) "is decentralized b…

Alas, I don't quite understand the point you're trying to make (and I'd like to!)

Perhaps this will clarify things: I think "decentralized by design" is linguistically ambiguous, because "decentralized" has a bunch of odd baggage -- do you think it's reasonable to say that email is "decentralized by design", and why (or why not)?

Re: Mastodon 2.5 released

#77
post #58

Earlier quoted context omitted.

I'm stunned that people still today don't realize that there is no, big and centralized free services. How would that work on the company side? Who pays for giant servers and software development and maintenance? People are not too sharp in 2018.

I resent the "not too sharp" comment, and the implication that I hadn't even considered the idea that you're paying with your data coming from all three of you. I'll respond here to all of you that responded generally with "you pay with your data, it's not free." Ok fine, but you don't know what you're paying with on Mastodon, because you have no idea what any one node might do with your data, because there is no cen…

That's fair it was a smart ass-ed comment, didn't mean to offend, but that one was tee'd up there, couldn't resist :)

I haven't tried Mastadon yet, but the fact I can run my own server for my family or close friends at a very low cost seems very appealing. To me this seems like the most likely use case, I can't imagine just going around joining random nodes, so I'm not too woried about the "virtue" of the whole network.

I'm happy to know that there is a solid, open source option like this out there.

Re: Mastodon 2.5 released

#78
post #74

Earlier quoted context omitted.

> There is a befuddling belief on the part of some decentralization advocates -- and of some decentralization detractors! -- that decentralization must imply an even distribution of power or influence. No, it's that if you observe that power laws are the norm, as you stated, then you start to see decentralization as a process, not a state of being. The idea that a network (or governance structure) "is decentralized b…

Alas, I don't quite understand the point you're trying to make (and I'd like to!) Perhaps this will clarify things: I think "decentralized by design" is linguistically ambiguous, because "decentralized" has a bunch of odd baggage -- do you think it's reasonable to say that email is "decentralized by design", and why (or why not)?

Sorry, I'll attempt to clarify: most human networks obey power laws in terms of centralization. For instance, the vast majority of the content is produced by a tiny minority.

Sometimes humans make networks in which this is no longer true, that is content is produced basically equally among all, or at least most, members of the network. These are decentralized networks.

These types of networks are not sustainable or large for a broad host of reasons, and so they form only temporarily. But they are desirable, and valuable, which is why they still form at all.

Email exists well within our very centralized federated political and cultural system, and so no, it is not meaningfully decentralized. We do have some small, decentralized networks (artist communes, maybe qualify?) within the US, but they do not scale. Mastodon will also fall prey to the same issues: as soon as it is big enough for anyone to care about, regulation will come, and regulation will require significant centralization.

Re: Mastodon 2.5 released

#79
post #47

Earlier quoted context omitted.

> I’d say it’s succeeding as much as facebook is. Maybe I'm defining succeeding too narrowly, and sorry to be rude, but what universe do you live in where this is even close to true? Some quick googling shows the network hadn't even cracked 100k users in Feb: https://mastodon.social/@Gargron/99469840739197657 > FB is essentially a growth engine designed to drive MAU numbers as high as possible Right, and it does that…

Facebook is like daytime TV with 100 channels and nothing interesting to watch. That may be succeeding in financial terms but there is a lot of room for other services where the aim is not to earn as much money as possible and I stead cater to intelligent people instead of the masses.

This implication you're making about "intelligent people" not using Facebook does nothing to further your argument, and just makes you look like a dick.

If you just want to refer to smaller, more specialized or niche communities instead of mass appeal, then just say that.

Re: Mastodon 2.5 released

#80
post #33
post #23

Earlier quoted context omitted.

I think you're maybe forgetting that the goal of Mastodon, or any federated (as opposed to distributed) system, is not to have every single user run and administer their own instances, but rather to enable users to freely select from a wide variety of compatible interconnected peers. > People don't want to moderate themselves, they want a larger power to do it for them. I think people mostly want a smaller power to d…

> I think you're maybe forgetting that the goal of Mastodon, or any federated (as opposed to distributed) system, is not to have every single user run and administer their own instances, but rather to enable users to freely select from a wide variety of compatible interconnected peers. This looks like a networking protocol, but this is actually process for electing the main node in the network, which will eventually…

> This looks like a networking protocol, but this is actually process for electing the main node in the network, which will eventually revert it back to the mean of centralization. Let's see how Mastodon scales

Would you say the same about email? Gmail may be huge, but I still wouldn't call email a centralized system. I imagine Mastodon would have a similar fate, if it gets big.

Post reply on HN