Well, it's still you + gmail.com. It could be you + fastmail.com, it could be you + posteo.de, it could be you + <insert other e-mail provider>. You still chose gmail over another provider, like you could choose cucumbers.pizza over another provider in the case of the Fediverse.
[1] https://www.searchenginejournal.com/gmail-rewards-users-with...
Not to mention, I do know a fair number of people with non-gmail personal addresses—iCloud, Outlook.com, Yahoo etc.
I have an email address. That unique identifier is also what I'd want to use for the fediverse. So how do I most easily do that?
(This isn't going to be a universal expectation; email at your own domain is not incredibly common. But it's also not exceedingly rare, especially among technical folks, or among folks with a personal "brand" and social media presence.)
It’s marketed as easy, but once you start digging into the details, it quickly becomes more complex than I was expecting.
I’ve been looking into Mastadon, Pixelfed, and Matrix to use in this way, but it seems pointless since I don’t want to host my friends and family’s content, and they’re not going to sign up for some random odd sounding instance. I’m then left with an empty social network or communication tool that’s little more than a tech demo.
It’s also a shame that they also helped enforce top posting, especially considering how ludicrous it is in the context of threaded emails.
Congratulations, we reinvented Facebook.
Anything with any hierarchy in its namespace will follow this path. Only pure P2P has a chance of avoiding it, and pure P2P is very hard to build and scale.
It also faces a serious protocol ossification challenge with no good solution (yet). Federated protocols face it too, but less as there are fewer cats to herd.
That's true of pretty much any system, though: The Iron Law of Oligarchy[0] always applies.
How is that anything like reinventing a monolithic social network like Facebook?
As a separate analogy, consider a server for a MMO. MMO servers live and die based off the usercount, and lower count servers are considered less desirable for other incoming users. So what ends up happening is that players congregate around the server that has the most users, which in turn drives up costs for said server requiring them to take in more money (through donations or ads or subscriptions) and soon they become the standard for other servers to follow.
Even if other servers end up hosting, few reach the same numbers because most new users are instead redirected or drawn to existing servers with a high user count. So we're back to stage 1 where the fediverse hasn't really solved anything at all. Said competitor may then choose to instead jump off the fediverse and take their users with them (because you become associated with certain other negative groups which are bad for your image), the fediverse starves because your average user doesn't care and then we have Facebook again.
As a counter example, if the behavior you described really did happen in federated systems then why hasn't it happened to email?
I think you'll find it agrees with your assessments. DNS and SSL Certificate Authorities centralize an otherwise sensible decentralized system.
It's out of date because work has continued. Here's some hints as to how we can improve the situation:
- Decrease importance of server you're on by allowing easy account migration. One easy way to do this is to use mutable data to represent your activitypub profile in a content-addressed store... you can look at mutable links in IPFS as an example (the work that we're doing on Datashards is also relevant). (ActivityPub actually does support other URI types that are not https so this is no problem...) Don't like the server you're on? Update your actor profile to point its inbox URI at another place.
- Support hosting over more p2p, easily self-hostable, NAT-punching and secure systems where you know you have a secure connection because the name of the server is actually the fingerprint of the key. That's actually what tor .onion servers and I2P servers are. There's no reason you can't run ActivityPub over such servers, and some people do, but it isn't widely supported because...
- ... because of the way we've chosen to do names. The right answer isn't webfinger (which isn't in the ActivityPub spec but is what's popularly deployed), it's petnames: https://github.com/cwebber/rebooting-the-web-of-trust-spring...
- And now you need a way to handle anti-abuse in a system that doesn't assume that domain names and instances are all too important. OcapPub outlines some of that (sadly unfinished, but the core ideas are there) https://gitlab.com/spritely/ocappub/blob/master/README.org
- On top of all that, maybe add store and forward messaging to support nodes being offline. This can be done and we have plans but I need to write them in a more visible place.
So in short, as one of the main authors of the biggest fediverse specs out there, not only do I agree, work is happening.
Everything I'm reading in your comment and in the links you shared, I agree with. I think the main difference is in our choice of tools.
For the network mapping (for trust and petnames), I believe that a graph structure is required. Having this at a foundational layer is important for scalability and mathematical simplicity imo.
Other than that, I think we agree on offline-first and multi-casting being requirements. And bonus points for keypair management through shamir's secret sharing and multi-device key management.
Can I contact/keep up with you directly somewhere to continue the conversation?
I can’t agree with that because Gmail was far from the first to provide this. Growing up all my friends and I had Hotmail addresses.
Gmail won because of unlimited storage and to a lesser extent a far superior UI. The invitation system also gave it a “cool” factor others never had.
I don't understand this sentence. What is the "something else" here that Gmail uniquely lacks?
The unique domain that I would otherwise have to associate myself with. "gmail" is so ubiquitous that it just doesn't mean anything anymore. It's just... the protocol.
Tell someone that your email is "[email protected]" and even _that_ is saying something significant about yourself that you may not want to say.
A big aspect of appearing "cool" is not letting on that you care so much. Social networks and messaging systems should allow people to be "cool" if they want to be.
That doesn’t explain how Gmail became ubiquitous. Back when it started, Hotmail, Yahoo Mail, AOL were very popular (among many others). Since it was not yet ubiquitous, Gmail’s ubiquity could not be an explanation for success.
Of course, I am, but no need to reinforce that.
...yes, of course; why else would you do it? ;)
I'm semi-kidding; there are very real practical benefits, but I really do run email through my own domain, at least in part, because it sends a message.
Many people on HN just straight up won't understand what I'm talking about.
To a normal person, "@gmail.com" doesn't say _anything_. Give someone a different domain and they will literally look at you funny.
The irony is that because email is so broken as an open platform, this normalized default so prone to collision has become the legitimate one, and everything else is filtered out...just like what tends to happen in other domains of society. It's easy upfront, but we're finally getting past the initial adoption of these technologies, and overall quality of service - from initial UX to issues like these - is going to become an increasingly large factor.
Also, if someone asks me for my YT channel, I just say the channel name and they know where to look, youtube.com.
With peertube you have to: give your channel name, remember then it's NOT youtube, tall then the correct instance, hope they won't type peertube on google and try to search for you in the first instance that comes up.
Also, searching for your account on any Peertube instance should work fine.
BUT, there's a huge problem. When you have a big creator on the platform, like PewDiePie for example. Every instance would have someone creating a fake PewDiePie. So, either the original creator would go to another platform, or they would need to use bots to create they users on all possible instances. So... creator will not have any incentive on using Peertube, as there's no 100% way to know who's the original.
Remember when Mastodon started, the fake IDs were already a problem: https://mashable.com/2017/04/06/you-cant-delete-your-mastodo...
You shouldn’t need Youtube, Twitter, or some other centralized provider to attest to your identity; you should be able to, with centralized providers only announcing or caching that attestation (just as keybase provided for with proofs). Maybe there’s an easy way to tie this together with FIDO, U2F, etc. I need to do more research.
But I tend to think that a true web of trust model makes the most sense. There may be 250 pewdiepie's on the network, but I'll just follow the one that (1) is closest in my network (followed by my friends) and (2) has the most follows (AKA highest trust).
We already follow this logic of #1 and #2 today even on centralized networks. Web of trust just makes sense to people intuitively.
Who do I know that already knows you? How else can I determine that you're legit? I'm gonna snoop your profile and see what looks legit. Ah, looks like a lot of normal people follow you already.
Of course this isn't full proof. But it's not full proof on centralized networks either. We just need reasonable certainty.
The argument is that, at some critical point, X became the default choice, and stayed that way.
Why it was gmail was interesting at the time, but isn't anymore; now it's literally just "people pick gmail because it's gmail".
Before I had gmail, I used my ISP email address. While using my ISP email address, I set up an email server with a dynamic hostname -- homeip.net at the time. It 100% worked.
There absolutely was a choice. You just had to be technical to know it existed in the first place.
It seems strange to me that gmail would be the default anywhere.
I agree, it's hard to choose from so many awesome domains.
"Which server do I choose to get in bed with?"
It's not just a question of which server do you trust to be a good actor and to not shut down tomorrow. It's also a question of which domain name do you want associated with your user account?
Do you want [email protected]? Or do you want [email protected]?
I believe that gmail won email not because of security and good user experience. I think they won because it removed the complexity of this problem.
When I sign up for gmail, I can just be myself. I don't have to be myself + something else.
Consequently, this is the strength of P2P networks. Remove the domain question entirely. I _am_ just a long string of characters (which is like my essence), but people call me by my name. Just like in normal life.