4 ms·
The problems we're trying to solve: Recurring outages on major networking sites such as Twitter and LinkedIn, along with incidents where Twitter members were m
by mechanical_fish 17y ago
The problems we're trying to solve:
Recurring outages on major networking sites such as Twitter and LinkedIn, along with incidents where Twitter members were mysteriously dropped for days at a time...
The problems with the proposed "decentralized" solution:
Most users wouldn't trust their laptops (much less their mobile devices) to store content and serve it up 24 hours a day, so they would need a cloud service... A backup service is also necessary... impersonation and unsolicited communications (spam) are hard to solve in decentralized systems... nobody can be banned from a decentralized social network...
Has anybody asked a customer whether this tradeoff makes any sense?
And, incidentally, the idea that "RSS never goes down" is preposterous. When the blog goes down, the RSS goes down. When the blog changes URLs, the RSS goes down forever, or until you track it down again by hand, whichever comes first. I have a tiny handful of RSS feeds compared to most people, but I still have to routinely garden the darned things. And when the RSS feed isn't full-text, or has a weird error [1], it's not exactly down, but it does kind of suck on a permanent basis.
---
[1] For me, Roger Ebert's RSS feed can't be read properly in iPhone NetNewsWire. The first paragraph is full width, and then
it tends to shrink
like a crazy
EE
Cummings
poem,
until we're down to one word per line.. Only that one feed has this problem. I wonder whose bug that is -- NetNewsWire, or the blog software, or the configuration, or Ebert? I wonder why I would want to have to routinely deal with such bugs on a feed-by-feed basis?
- wmf 17y agoAn RSS feed can go down, but RSS itself cannot go down.
- tedunangst 17y agoThat's about as helpful as saying Twitter can go down, but HTTP cannot go down.
- hellotoby 17y agoBut what he's trying to say is that when Twitter goes down, it takes everyone with it. When a blog goes down it only takes it's RSS feed down with it, not every RSS feed in existence.
- Periodic 17y agoThe whole point is that a single failure will not bring down the whole system. As it is right now, regularly no one will have twitter access. With a decentralized system you might lose specific people, or even a whole region or maybe even communication between regions, but never the whole system.
- mechanical_fish 17y agoPart of the whole point of Twitter is that either your posts are seen by all of your followers, or (in the increasingly rare cases when Twitter is down) by none of them. If you want a decentralized system where some of your recipients will get the message, but others won't, shouldn't you just send email?
- pyre 17y agoDecentralized doesn't mean client-side. Think about it this way: What if Twitter wasn't a centralized company, but a standard protocol for software running on servers to communicate with each other? Each user would run their own 'microblog' on a server somewhere. When someone 'followed' them, they would end up subscribing to an RSS/Atom but in a way that was more similar to rssCloud or PubSubHubBub (i.e. the user1 wants to follow user2 so user1's software registers a 'callback url' with user2's microblog; now when user2 publishes, an update is pushed out to user1's microblog software) Users wouldn't even necessarily have to fool around with 'hosting packages.' This opens up a market for companies to provide 'targeted hosting' (hosting of a particular software package) or to create their own software package that conforms to the basic tenets of the protocol while adding things onto the interface (like games/advertisements/quizzes/etc). So the 'average user' wouldn't necessarily have to get their hands 'dirty' trying to figure out the difference between Shared Hosting and a VPS. You can have servers in a decentralized setup. The whole point is that there isn't a single point of failure (in this case Twitter or Facebook). Not only that, people could (presumably) have better access/ownership of their own data and be able to back it up in case of disaster (in which case they could restore everything back to normal or migrate to a different server/service).
- halo 17y agoI don't really get your point - that's an extremely helpful analogy. The web is based around HTTP, a deliberately decentralised protocol with no single host that could take the entire web down at once. If Twitter goes down, or even a prominent web-host, the entire web doesn't go down with it. This is an extremely useful property, because even if your favourite site is down, you can still use other parts of the web. On the other hand, Twitter, Facebook and other social networks are centralised. Unlike the web and HTTP, if they go down, you can't do anything to do with those services.
- pyre 17y agoThe point isn't "if Facebook goes down I can't access Facebook." The point is "if Facebook goes down everyone on Facebook can't access Facebook." If social-networking was a series of web-software packages running on servers (whether self-maintained or managed by a 3rd party -- something like what Wordpress/LiveJournal/Blogger are to blogs) that communicated with each other, then if my social-networking node goes down I obviously can't access it. But other people's nodes are still accessible by me and by them. It's just my node that has dropped out of the network (granted it's possible that your node dropping off of the network could also take your permissions with it -- i.e. you would only have access to public content until your node was back up).