Let's not get carried away here. Twitter is great. I use it too much of the day. But the world hardly needs Twitter.
I think he strikes a good tone in both emphasizing the importance of the work they are doing at Twitter, and how they need to do better. I think less than perfect objectivity can be forgiven.
The benefit of any place or tool for social activities is only tangentially related to the benefits provided by that place or tool. The value is the people that are using it, who can (and do) move all the time.
Anybody interested in the relationship between a service like twitter and the people that create it's value by using it may be interested in the talk given by an admin o fark.com a few years ago, where they discuss how they almost destroyed their community ("you'll get over it") from a failure to remember why people came to their site.
If there is one lesson that Twitter or any other social media business needs to learn, it's the one discussed in talk: involve the people that use your service at least somewhat in the changes you make to the service, and absolutely don't surprise them with sudden change, or they will find some other place to hang out.
 https://www.youtube.com/watch?v=YnVeysllPDI (language warning, if that matters (it's veyr mild))
Hmm, the world needs networked communication; the specific protocol running at the endpoints isn't particularly important. TCP/IP was used because... it was already used by the people involved.
Further, the world needs easy communication; the specific mechanism to communicate isn't particularly important. Networking was used because... it was already used by the people involved.
LOL, sure as hell. Twitter was blocked in most of these countries (and Turkey too). The outcry was huge, but guess what... They 'survived'.
Twitter is just a medium. The preferred medium for some. There are countless alternatives nowadays, no one needs twitter.
Seems like what's really "needed" is a distributed Twitter-like platform, using a DHT or something. Keep it small and simple enough, and a client could run on many devices 24/7, updating every a few times an hour, or more if the user's actively using it.
User identity might be the hard part, because a long hash isn't very human-friendly. But perhaps that could be solved by a centralized directory authority mapping user ID hashes to Twitter-like usernames. The mappings could be cached by clients in case the directory went down or got blocked. Sort of like how PGP keyservers work, letting clients get keys when they need them and add them to their local keyring.
No, you can set up endless blogs (using Wordpress or Joomla or any other CMS with Twitter plugins), and include a twitter stream in it. You can post to twitter by email or using similar plugins.
A government cannot block all those sites.
Blocking twitter is not enough to block all content coming from and going to twitter.
But then the users have to do the same thing, preemptively finding such proxies and jumping between them. I'd guess that that would be a good enough result from the government's perspective.
And that's just read-only stuff. Post by email? DPI could drop packets containing emails addressed to Twitter, even if they're being sent through a mail server in another country. And, sure, you can proxy stuff, but again, the point here is the masses. If 95% of people are blocked from access, that would probably prevent mass protests from happening; or at least keep them from happening quickly and unexpectedly.
But besides all that, when the government cuts off the country's Internet access entirely, creating a nationwide intranet, none of those blog proxies are going to help. The only thing that is going to truly work in such a situation is peer-to-peer stuff.
This is why it is so important to create protocols for communication instead of single-point-of-failure services. While it is still possible for bad actors to interfere with traffic, it is much harder to shutdown an entire network of federated servers. In contrast, a service can be disabled by simply throwing legal papers at the person running the service (or otherwise threaten them, bribe them, etc).
> DPI could drop packets containing emails addressed
Which is why that email should have been encrypted. Even non-authenticated opportunistic encryption would make that kind of DPI much harder and more expensive.
We've had encryption for a long time, and there is absolutely no excuse for to not have encryption in all network software. Unfortunately, a lot of software has been badly negligent in their duty to keep their users safe.
> cuts off the country's Internet access entirely
...they should be left with a split network, still able to communicate internally. It is pure hubris to believe that a single service will always be available in any situation; when international links are involved, some amount of loss-of-service should be assumed. This is why email allows multiple MX records - servers fail, so build in backups, some of which can be local.
> peer-to-peer stuff
Yes, though the model to copy is federated services like email.
The power of the internet is that it IS peer-to-peer. No 3rd party has to grant authorization to listen(2) for connections or connect(2) to a remove host. This has been forgotten because of the damage that NAT does to the network (you aren't a full citizen ("peer") on the internet when you have to ask the NAT device to forward your packets.)
Now, after about a decade of failing to heed the warnings about NAT, the digital imprimatur is being constructed and running proper peer to peer (no 23rd party) is harder than ever. Hopefully IPv6 can reverse this trend.
> Yes, though the model to copy is federated services like email.
I disagree. For one thing, DNS. How many average users will be able to reconfigure their systems to use a DNS server besides the one their ISP provides? And when the national government cuts off the nation's external Internet access, what good will 18.104.22.168 be?
It seems obvious to me that, for this kind of use-case, peer-to-peer, DHT-type services are what's needed. Federated is better than centralized, but it would still be easy to censor and block. Even P2P protocols could be blocked by ISPs blocking inbound connections to their customers, but barring that, it would be difficult.
Twitter is a news source not a newspaper.
While I agree, I also hate this. A Twitter trend can inspire good journalism, but it is not in and of itself journalism. Too many media sources slap a tweet on TV or on a website and call it journalism.
Twitter simply connects the event's with the journalists :)