Aether – Free, privacy-sensitive public communities (opens in new tab)

Looks good. But a native application? I would only install a native application if it comes from a source that has a proven trackrecord of being trustworthy.

Can you track how many people installed it, now that it is on HN?

You say it puts everyone on "one single universe with the same content". So it should be easy to put up an instance that provides a web interface to the content, right?

I would think that posting via the browser should also be possible without relying on a single server / authority. A user could hold a private key which they use to sign their posts. And then deliver the post via any server that offers an interface for it. Similar to how a Bitcoin transaction can be delivered via any server that offers such a service.

So it would still be decentralized but can be used via the browser.

Looks great, kudos on the `soft' launch.

* The FAQs say that the app is a p2p based app. What is the p2p protocol being used here? * How does it compare to libp2p, SSB, and others? * What's the security model, and how would the system mitigate Sybil attacks etc?

Are there any plans to have mobile apps (preferable native)? I think if people could use it on their mobile phones, it would be a game changer.

Can you post to this from Tor?

edit: oops, just found this https://meta.getaether.net/t/how-to-use-aether-behind-tor/53

Seems to overlap with what SSB is trying to accomplish. Do you see any opportunities for integrating with that or other protocols, maybe for tracking unique identities over time?

(backed on patreon btw)

Do threads refresh automatically in the Aether client, or is there some magic way in which one can do so?

c.f. aether://board/86e782e80681ac580b4d6d102b12e787c066e59f194fee57bb0bf83cc1e42fc6/thread/4932c494227ccd4f632c2ace1012f58fd79d0d531b2dcd148463491598c841d8

Is the proof-of-work likely to be too much for a mobile to be able to do? If a mobile client were to be made, that is.

Also, does the proof-of-work scale, eg how the bitcoin network gets harder and harder as the hashing power increases.

Thanks for creating this!

How does the unique username perk for patrons work? Does it mean that people can still create "normal"-coloured usernames, but there can only be one orange "verified" one?

I would like to give Aether2 a test run. The only version I can find is the old 1.2.3 release on GitHub. Others here seem to have version 2. Where can I get it?

I thought about doing something like this with orbitdb. maybe large content, like video, could be using ipfs.

will you provide .AppImage or Flatpak or Snap? or all of them?

You state that one of the purposes of this platform is to allow you the freedom to be wrong, but if people can save something before 6 months, how would you prevent someone from saving an embarrassing comment for future use against you? Or are you the only one capable of saving your own comments?

I think native UI in dedicated applications is massively overrated. It's great if it's apart of the system, if there's a config menu or something you'd expect from the system itself, but I have never had a problem with Spotify or Discord in terms of 'non native'. They can be a lot better absolutely, but native look is far worse.

I know - given enough time and money, I'd make a CLI app using something like ncurses. Unfortunately though, the reality is that I just don't have the resources.

The backend is its own process, you could definitely find the binary in the application package and run it on its own — it's not bound to Electron in any other way. Electron is just the client, the frontend graph compiler, and the backend are two separate binaries that communicate with each other over gRPC. That means you could run any of these independently, as well. [0] The client is Electron, but nothing except the UI happens in JS, everything happens in the frontend and the backend, and they're both tiny, fast, native-compiled Go apps. My backend, for example, is 4.8mb, frontend I think is 6mb. The rest of the size of the app (70mb) is just Electron, its dependencies, and the whole enchilada that comes with it. I'm not happy about it either.

For better or worse, though, since the JS part of the app is nothing but its 'skin', it's as memory efficient as Electron apps get. No real processing happens in JS, it's all Go.

[0] That said, it's not something I support right now, because if you ran the backend separately and then started the app, you'd effectively be running two backends. There is not yet a way to tell the app package that it should connect to an existing backend running locally instead of starting one.

I haven’t worried about how much memory an app uses in many years. On a MacBook Pro with solid state storage, it’s just not noticeable. I did some testing a long time ago to see how the system behaved when apps used essentially all of the memory and then some.

I actually couldn’t tell. I don’t remember anymore if I tried playing a game or not, so that might have been noticeable. But my VMs and browsers and several emacs’ connected to several clojure repls didn’t even hiccup.

The fans ran high and the computer was hot, but otherwise it just wasn’t a thing.

Is it that different on windows or Linux? At work windows is abysmal, even with an ssd, but I blame all the antivirus and corporate spyware.

—-

I do marvel at app binary sizes, just because my floppy disks used to hold jus 144k. I did a hello world in Haskell this morning and it was a meg or two. Hello world could fit on a floppy from the //e era. That’s something.

For both of your questions, the answer is proof of work.

For Sybils, that’s something I’ve considered. I’m thinking of making voting eligibility a trailing indicator where it tracks past activity to be considered a ‘citizen’ of that community, and allowing the mod team to temporaily lock down the community if they see such a thing happening.

Does any existing kind of social network prevent that kind of attack? I don't believe so, and I'd think that's pretty difficult to accomplish. Aether does attempt to make this nontrivial, I believe, by requiring proof-of-work before posting things. I don't know, but possibly that proof-of-work also applies for impeachment votes (I hope it does!). Then depending on how hard that proof-of-work requirement is and how many people are already in the community, performing such an attack would become harder.

It is brilliant. The moderation layer is a like a data view.

I wonder, what happens if, going a step further, there are different moderation teams working on the same comment stream, and readers could flip between them at will? If you ever wondered what the other side was thinking, you could flip views and experience it.

Of course, you still need some absolute site rules to cover doxxing and illegalities.

I like the way you sum this up Ixe. Your comment illicits a mental model like a manual transmission where the terrain (dynamics of the conversation) sets the context for what gear to select (which mods to have in place) , based on how much gas you are giving the engine, which is a personal choice of the driver. One driver may choose a very low gear to go down a hill (a conservative moderator that controls the decent of the vehicle) and another driver may wish to just coast down at high speed (very litte moderation). Same car, same terrain, different driver, different choice for what gear to keep it in. The mental model gets a little loose goosy when you consider that this is not a single person making the decision but rather a group, but oh well. Thanks Burak for making this platform. I'm really enjoying the ride so far!

If I remember right, Google hired him soon after he started the Aether project. I guess he's left Google by now.

So, Burak, care to share the thought behind your username? :)

> because some nodes are behind restrictive firewalls and NATs

That is just a weird turn of phrase on my part, will fix it. What I mean is:

because some nodes are behind restrictive firewalls and [restrictive] NATs.

In other words, most NATs work fine. Even if they don’t work fine, it has a process called a ‘reverse open’ where the inbound connection is requested by the receiving remote and the remote ‘captures’ and uses the same TCP socket.

Even if that doesn’t work, as you’ve discovered, you can run the backend Ona VPS.

Not all NAT's are relevant. As long as they respond to port mapping requests or use endpoint independent mappings they can be traversed.

Also note that IPv6 alone does not help because the recommended way to deploy it is to have firewalls in CPEs that block incoming connections by default, so applications still have to request the port to be opened, using the same protocols as for v4 NAT.

The payload is just text. You’re not carrying anything else.

But yes - for the very small, but not nonexistent case where a text itself can be not kosher, this is something that needs to be built in. It’s on my list.

It isn't a "democracy", votes have a different weight based on the past participation on the specific community sub.

Some type of social credit system with power at the local and global level. Give active and creditable participants in a single community more authority there by giving more votes. Global credibility (you play nice in other communities) additionally is factored into the users ability to influence. Make this credit consumable. This way a community that wants to disrupt another will have to overspend and leave themselves vulnerable to destruction. MAD in essence.

I've asked: https://meta.getaether.net/t/how-do-you-stop-hostile-communi...

I wonder why can't one just pick mods for themselves. E.g. "subscribe" to a mod, meaning whatever that mod hides, you don't see. It could even be mutual.

I dunno - when I saw 'decentralised' I assumed I'd have to download something. Isn't that how decentralised applications work?

Since Aether is a peer-to-peer network, the client syncs content and becomes a node itself to scale up the network's capacity. It also allows the user to perform actions without letting a backend track your navigation history, etc. I don't think there is encryption done on the content, only signing and verifying with keys, as there is no concept of private messages in the Aether network at the moment.

Also the application takes 1-5 minutes to "download fresh content".

The posts are echoed across the network. You know where you got the post, you don’t know whether that post was created by that specific IP. Because you could have gotten the same post from all other nodes in the network as well.

Delivery IP is not the creation IP.

> Not possible. Either source IP's are in communication, meaning you know the active posters IP, or the content is being proxied from a centralized server.

Which of these is true of a P2P network of Tor hidden services?

e.g. https://ricochet.im/ is P2P chat that uses Tor hidden services. It neither has centralised servers nor leaks IP addresses.

I agree - but anybody who ever built anything would tell you it’s a tradeoff. Yes, I could get metrics through the server logs. But I’m just one single guy that built this thing. I don’t have the resources to build a privacy sensitive tool and a privacy sensitive page view reporter that tells me how many people visited a page, so I can improve it. One challenge at a time.

Same here. At first I thought it was the hug of death, but I feel like with p2p it should only get faster as more people join?

Inbounds conns. in last 15m 0

Last inbound Unknown

Outbound conns. in last 15m 0

Last outbound Unknown

Last outbound duration Unknown

It performs an initialization. You need to exit the program. Be aware that the “close” button only minimizes to the system tray

Same for me, it's "Catching up" and appears to connect to some ipv4 nodes, but nothing so far. Edit: it werks now!

Maybe you should refresh your browser cache.

I've never seen a community that was completely open to discussion and where bad ideas disappeared because they were effectively challenged.

On the other hand, I've been in several communities where the mods had a clear agenda and the participants were helpless to change it.

I'd personally rather the people who are hostile to new ideas or critical thinking have their own communities and that their ability to harm other communities can be mitigated by a moderation system accountable to the users.

I do think there needs to be serious discussion about how to make communities that are good for people. The echo chamber and polarization effects tend to contract people into a smaller worldview. How to have a cohesive community that doesn't do that is essentially an unsolved problem in the social sciences.

Won’t this be resolved the way it does on Reddit with people ejecting to more specific communities?

I’m a feminist for example, but I’m no longer welcome in mainstream feminist subreddits, so I subscribe to various more-specific-feminism subbreddits, and explicitly debate friendly ones.

@b here. Seems the snap isn’t able to boot because of a similarly named package? Linux isn’t officially supported right now, but there was somebody downthread that managed to get it to work with the —dangerous flag.

It is a flood network.

> A result of this is that source IP of any specific public post cannot (easily) be determined.

> We should focus on changing society so it's not so critical of statements people made years ago when they changed as a person

I'm a believer in progress long term, but I seriously doubt humanity is really capable of what you're describing in a way that's applied equally to all groups of people.

It seems several orders of magnitude much easier for people to develop the habit of recording information they might need later.

Yup - check LocalMemoryDays key on your backend config. That controls the archival. That should be the largest value, and the network head and network memory should be smaller than that.

>This could go mainstream fast if it picked up the Gab crowd

The "Gab crowd" doesn't have the cultural pull or influence that this would require.

There is no mining of cryptocurrencies. You are spending CPU cycles to compute a proof so spamming the network with user actions is non-zero cost.

Impeach: call into question the integrity or validity

Seems perfectly appropriate usage to me.

I think the _scope_ of privacy here should be clarified. Content on Aether is public. There is no concept of private encrypted messaging. The privacy guarantee is that what content you view within the network, your navigation path, cannot be tracked by a server. This is because your node syncs _all_ the data, including the content that you don't navigate to. This is in contrast to the way many services track user behaviour by their username.

As discussed in other posts, it is difficult to discover origin IP of pseudonymous users, because in a flood network, when my node gets some information it is usually not from the IP where that information was generated. That is another privacy feature due to the way Aether distributes data.

Aether is more difficult to block than centralized services, since you will have to catch all the nodes instead of just one set of known IPs / domains. With great effort it is probably possible to DPI at ISPs to identify Aether traffic and drop? But being a peer-to-peer protocol, Aether can easily sync across VPN tunnels, run within private networks, or physical meshnet infrastructure independent of ISPs.

It seems there is some kind of bootstrap node running somewhere that clients connect to when they haven't got anything. The upside is that this bootstrap node isn't special, and anyone should (I believe?) be able to set up new bootstrap nodes. But I'm not sure on that part.

I downloaded the snap and installed it from the command line with `snap install --dangerous Aether.snap`. Dangerous sounds dangerous... I'm not sure if it was a good idea, but it works and my system found the app when I searched for it. I clicked the "cat's bum icon" and we're away!

@b here. Snap version is untested - it compiles but no guarantees. I think some people got it to work downthread, you might want to check which flags they used.

The same way you prevent them from becoming progressice/racist trash bin worse than reddit.

You let them speak and prove them idiots. It's really not hard. Unless you're constantly on the side that's losing, then it's pretty miserable.

But know that even though you're losing now you can reformulated your arguments and try again because no ones going to censor you. You have all the same rights because you, like they, are people deserving respect, no matter how much you look down on them or hate them. And that's what I tell them about you.

Less hate more love.

I feel like the cost of having open communication platforms is the existence of toxic communities.

The correct approach (IMO) is to educate people about these rather than to have some sort of censorship mechanism.

It isn't possible, but you can choose which subs where you want to subscribe.

There will be a moderated list of subs (maintained by the main developer), but users will be free to try to find an join even the most controversial ones.

As someone who has spent the better part of their life advocating for openness, decentralization, etc... it pains me to believe this, but I'm not sure we can have both open decentralized communities and simultaneously a community that isn't overrun with toxicity.

If the goal is shitposting and edge-waddling I think it can be done, for a while anyway. However, if the goal is to have a space where discussion is meaningful, good faith, charitable, intelligent conversation, I'm not sure we can currently do this in a completely open way and still maintain a signal to noise ratio where sane people would willfully spend their valuable free time.

In our current iteration of tech, society, education levels, culture, and politics it seems that (so far) every time we create an open space, the moment it reaches a certain level of popularity, the noise levels reach a point where it is pretty much impossible to have meaningful, contextual, and nuanced conversations.

At the point that signal to noise ratio reaches unacceptable levels, we lose a critical mass of the people who made the space "special"

I really hope we can get around this someday, but sadly, so far the absolutist free-for-all spaces really offer diminishing returns for those of us who are busy. Rather than sifting through rubble to find gold, at least for me, I can think of many things I'd rather do than engage in a conversation with someone when there is a higher than average chance they aren't there in good faith and when entire segments of the community exist who are literally only in the space to engage in and spread sophistry.

At least for now, outside of invite only communities or high barrier of entry communities, spaces such as HN or SSC --both of which are largely unknown to the general public-- seem to offer the best blend of openness and time-well-spent conversations. But I'm not sure either of these could maintain the same level of conversation at scale.

Groups have moderators, users can subscribe to specific mods' decisions.

By getting a large enough userbase, and not repeating the cycle by banishing unwanted opinions.