211–220 of 356 posts
Earlier quoted context omitted.
For a properly designed distributed social network, there's no reason your data should be visible to anyone other than those you have approved.
What if people get tricked into approving malicious accounts? Who is gonna police that? How are you gonna prevent anarchy?
Do you really need to idiot proof everything? I'd say just let natural selection happen at this point.
2018 will be the year of "alternative to facebook" apps that are in no way an alternative to facebook. To be an alternative to facebook, it should at least do 50% of what facebook does, and it should be accessible to all. Anything that takes more than 3 steps to get it running it's going to keep people out. And if you keep people out, you don't have a social network, at least not anything like facebook where your gra…
It's a common mistake to attribute the success of Facebook to its features (it boils down to microblogging + threaded discussions). The usefulness (i.e. product from a marketing perspective) of Facebook isn't its features, it's the people who joined it. Look at Google+, it's not that bad in terms of features. But it lacks people, therefore it's not a competing product. A competitor would be not someone who does 50% of what FB does. It would be someone who has 50% of what Facebook has.
Earlier quoted context omitted.
"You need ... a business plan." I don't agree. What is the business plan of email (by which I mean SMTP, not some webmail provider)?
> by which I mean SMTP, not some webmail provider Cheap cop-out. "some webmail provider" is what runs 99.999% of email. The people who run their own email, or run off hobby email servers, are insignificant. You can't turn a blind eye to where the majority of users are going to be if you're implementing anything that actually needs traction in order to be successful.
Federated social networks are like email: each instance is like the "webmail provider" you can pick and chose and some of them even come with their own clients and bells and whisltes.
Protocol doesn't need a business plan.
I think it's fascinating to see distributed social networks from a tech perspective. From what I've seen so far they exacerbate the problems that Facebook has been seeing so much backlash against. 1. The whole Cambridge Analytica issue was caused by APIs that are too open. For distributed systems there are more ways to exploit the APIs and gather data on users. 2. There is a clear issue with Facebook's accountability…
I wanted to take a look at how it works but the on-boarding process is not really welcoming. Getting started [0] redirects to this weird site [1] served over plain HTTP, installer is also not signed :( [0]: https://www.scuttlebutt.nz/getting-started.html [1]: http://dinosaur.is/patchwork-downloader/
Have we reached the point where plain http seeks not welcoming. If the site above doesn't require you to login why do you think you need https? Is the only reason to hide your visit to that site from your isp?
You seem to misunderstand https... for one, it doesn’t hide your visit to that site from your ISP; they know the IP address you visited, and due to SNI, they will even know the domain. The point is to make sure you are connecting to the site you think you are connecting to.
I thought this was being done by Diaspora? https://diasporafoundation.org/ What's new/different here over this and other efforts?
Diaspora is federated, which for non-technical users is worse than FaceBook itself. The individual nodes (to which thousands of users are connected) will have far fewer resources to secure themselves than a behemoth like FaceBook. Also, uptime, badwidth etc. ScuttleButt is peer to peer, somewhat like torrents but for data feeds.
That's a weird argument - it doesn't even make sense no matter how you look at it.
1. Security is more or less centralized as per core code and core protocol. 2. Why do you imply big monolith is secure than a small instance? Where's the logical connection in that?
I think it's fascinating to see distributed social networks from a tech perspective. From what I've seen so far they exacerbate the problems that Facebook has been seeing so much backlash against. 1. The whole Cambridge Analytica issue was caused by APIs that are too open. For distributed systems there are more ways to exploit the APIs and gather data on users. 2. There is a clear issue with Facebook's accountability…
However there is no global state, users only push-pull feeds of friends and friends of friends. (In part for privacy and in part for performance, you don't need to carry all the data for a social network to be useful)
> 3. GDPR compliance about deleting data is almost impossible in a distributed system.
However that doesn't apply to individuals that aren't providing a public service.
> 4. ... Open source software and distributed software tends to be much harder to use.
Yes, however there has been some exiting discussion around what UX could become possible in a decentralized context. See https://coolguy.website/writing/the-future-will-be-technical...
> 5. Any future concern/issue will be much harder to resolve if there are thousands of different instance running decentralized social networks.
There are no instances, only peers. (Pubs are sort of easy to connect to peers and there is work being done on peer discovery to outdate most pubs; they could still be used to connect communits around topics, hobbies, etc as a way to reduce the distance between peers)
> 6. Using AI to detect abusive content or spot fake news is much harder if you only have a subset of the data.
There is no retweeting/sharing, though there is work being done on Out of Order Messaging to propagate a single message along the follow graph and there is work being done on flagging/tagging feeds/ids and posts. (Along with some semantics around how to interpret the flags/tags to improve UX and user control)
Earlier quoted context omitted.
Your concern about APIs in distributed systems seems to neglect the fact that cryptography can be used to ensure that only the right people have access to the data. Although I do agree on the fake news point you make. Edit: It seems that scuttlebutt does support optional end-to-end encryption: https://github.com/ssbc/secure-scuttlebutt#security-properti...
> cryptography can be used to ensure that only the right people have access to the data That was my assumption too, but I have no technical knowledge on the subject. If encryption was used, would that keep a rogue server admin from accessing sensitive user data?
There are no servers, it is peer to peer. The pubs are just easy to connect to to bootstrap new users into the social network and have been mostly closed off now that it has become more popular to avoid bad actors from abusing that easy way in.
You can bootstrap on a LAN / WAN without pubs.
Since it's AGPL, why not call it Free Software instead of Open Source. https://github.com/ssbc/patchwork/blob/master/LICENSE
What's the difference? I've been calling my AGPL and MIT licensed projects as free and open source.
https://www.gnu.org/philosophy/open-source-misses-the-point....
wtf were they thinking when they named it that? if any non joke product with "butt" in the name ever succeeds, I will eat my hat.
While I agree with your general premise, you might as well start picking your favorite hat condiments. H-E-B is a wildly successful grocery store chain in Texas, and also the in-house brand for many of the products sold in that store. The marketing slogan is "HEB: Here Everything's is Better!" The reality is that, while it is often better than the competing chains (sometimes sadly so, killing Krogers and other chains…
It's all about the character of the state...
I once saw in Colorado a university that had the "Alferd Packer Cafeteria" (sic) ;-)