Twitter to decentralize something
manton.org
manton.org
So if Jack wants you to come build it, and you’re a P2P person, you should go work for him in even if you don’t like him, because he has both the willingness to pay for it, and the broadcast power to make it mainstream. And those are the main problems that needs to be solved, not tech.
I know this because I’m going through the same thing myself — I work on Aether (https://getaether.net), which is close to what I think he is talking about.
If we want ourselves to solve this, and not rely on @jacks of the world, we need to find a way to make P2P into a viable business. For me, in the end I decided to create a Pro version for use within companies, so that the P2P version can be fully free of monetisation concerns.
For example, there are tens of millions of images uploaded to Twitter daily. They all have to be stored, and then retrieved for nearly arbitrary tweets and arbitrary users.
All this content has to be stored somewhere and retrieved fast enough to be useable.
p2p tech is fundamentally incompatible with American morals. You will never see a p2p based firm be successful in the US any more than you could have had the industrial revolution in Ancien Regime France.
1) It needs to work reasonably fast when the suddenly popular post is seen by, say, 10000 people in a short time. My home uplink won't handle that.
2) It needs to work when my computer is turned off or disconnected from the internet.
3) (optional, but still important) People on the other side of the world should be able to get popular data from a cached copy that's physically closer, both for latency and bandwidth reasons.
"Store what you post" isn't sufficient, all that I post needs to be stored also by someone else (e.g. a CDN) - but that immediately rises all kinds of issues about liability. If only I am storing what I post, then the system can afford to not care if that image turns out to be child porn, but if someone else is storing what I post, then that's a problem for everyone.
https://www.wired.com/2017/06/pied-pipers-new-internet-isnt-...
There's a lot to unpack about the issues of p2p monetization. Probably a few books worth.
Before we get into the social issues I think the original sin of the internet was that IP did not negotiate payments as part of the fundamental protocol. Every layer above that suffers from the same.
This made sense in a government/corporate/university environment where the protocols were invented, but not for the wider world. I'm not going to try and give solutions because all of them are flawed in some way, but better than the current system where only surveillance advertising and altruistic services are viable business models.
The internet is happily chugging along on routers and switches who know nothing about the global topology of the network.
When you reduce the payment problem to 'how do we get two routers to keep track of the data passed between them' it is a non-problem that even the first generation of imps could have solved).
Networking is just not worth lying about until you get to sums for which lawyers can get involved, at which point you have to deal with lawyers.
That isn't really what causes the problem. Payments don't need to be inside the network layer of the internet.
The problem is that a digital payment system that works like IP routing doesn't exist at all, even independent of the internet. Existing banks could implement it trivially -- you walk into the branch with $100 in cash, give them the money and they give you a secret passphrase. Then the banks talk to each other so anyone anywhere in the world who walks into another bank and has the correct passphrase can get the money.
Except that there are laws against that. Know Your Customer laws that require them to know who you are and know who the person you're sending the money to is, even if what you're paying for is both legal and private. "Consumer protection" laws that require them to allow you to reverse the transaction, which means they can't just give the money to the other party, they have to sit on it for a while.
The effect of these laws is to make small transactions between random individuals incur prohibitively high transaction costs and subject honest people to capricious denial of service attacks by risk-averse financial institutions without effective recourse. It also makes people afraid to pay for things they don't want an official record of them consuming, which specifically applies to political speech when your views don't align with the party currently in office -- or the party who may be in office next year.
In other words, it interferes with people actually paying people for creating content on the internet. That leaves the creators with only advertising to support them, which implies central control by the ad network.
Fixing that requires a way to anonymously, irreversibly send small payments with low transaction costs. Which is currently de facto prohibited by law.
The cost of getting it wrong is so low on the internet that you can use the most blunt statistics to still come out ahead.
You don't need packet-level accounting. You need, for example, to be able to charge $0.02/view for a video while getting to keep almost all the money, so that a creator with 10,000 views/day can make a decent living. Then you offer viewers a deal -- no charge if you host the video. Only a small fraction of the viewers take you up on it ($0.02 is not a huge incentive), but a few have unlimited data and are happy to get any price for something that costs them nothing, or it makes them feel like good supporters to be distributing your work, and that's all you need in order to have unlimited scalability and negligible hosting costs.
> You don't care if someone scams you a thousand times, you do care when someone scams you a few billion times, or when you're talking to millions of people.
This system isn't supposed to be a savings account, it's for small transactions. If there's significantly more money in your account than you typically spend on small transactions in a month then you're doing it wrong, and the amount in your account would be the upper limit on what you can get scammed out of.
People would still keep the bulk of their money in a traditional financial instituion that knows their name and can reverse transactions. But a payments system that works differently than that should be able to exist.
You absolutely do.
What you don't need is accurate and global accounting of packet transactions. A simple cryptographic local ledger between each pair of routers will be more than enough. The routers can then decide how the data is charged, or if the balance isn't closed somehow, just start dropping packets coming from that link. We are not talking about people doing videos - a largely vapid and useless form of entertainment only propped up by legacy add platforms like google because video adds are much more effective than text based ones - we are talking about people running routers in the center of the network. Once we have decentralized the lowest levels of the internet then we can start thinking about the higher ones.
After all the choice between lord Google or lord Verizon isn't much of a choice.
> Only a small fraction of the viewers take you up on it ($0.02 is not a huge incentive), but a few have unlimited data and are happy to get any price for something that costs them nothing, or it makes them feel like good supporters to be distributing your work, and that's all you need in order to have unlimited scalability and negligible hosting costs.
So again, the main business model is altruism with no way to actually monetize the transport layer. Replacing the kings of layer 4 with the kings of data is not a victory. It's just a different tyranny. One I don't see being much better than the current one.
Only when you can make money in each layer of the network can the internet truly be free (as in freedom).
>If there's significantly more money in your account than you typically spend on small transactions in a month then you're doing it wrong, and the amount in your account would be the upper limit on what you can get scammed out of.
Yes, that's why you shouldn't do it that way.
You don't spend 'money' you have agreements with your peers over how much data they are willing to receive from you. The clearing process is completely independent. Cash, cat pictures, happy feelings, revolutionary zeal. All are acceptable currency if the owner of the router agrees with you.
At this point you're essentially arguing that network neutrality is bad and what we really need is the opposite of that.
Realize that in your system, it's Verizon and Comcast who have all the power. They're sitting on millions of customers who have no other network path to them, so they would get to charge monopoly prices to anybody who wants to communicate with them, and the customers themselves have no other choice because the incumbent ISP has a local monopoly (or the sort of totally inadequate competition that DSL is now putting forth against cable/fiber).
And if you're paying for sending data but not for receiving it, it would give ISPs all the more incentive to keep doing and increase doing the things like selling highly asymmetric connections that upload much slower than they download, which make it harder to create competitive P2P networks.
> So again, the main business model is altruism with no way to actually monetize the transport layer.
How is the network not already monetized? Is there some alternative to Verizon and Comcast where you're paying no monthly fees, short of building your own network like Google does? Even then you still have to pay for it, you're just paying for land, labor and equipment instead of service.
You pay your ISP to get you to a peering point. That's monetized. The other endpoint pays their ISP to get to the peering point. Also monetized. What's left? The peering link? The actual cost of that is such a small percentage of the total that you can round it off to zero, and people only really try to charge for it when they're rent seeking because they have a monopoly path to a large number of customers.
> Yes, that's why you shouldn't do it that way.
Because occasionally people who make bad security choices will be encouraged to do better after suffering an uncovered loss of something like fifty or a hundred bucks? That seems like a small price for having a payments system that has low transaction costs and protects privacy.
> You don't spend 'money' you have agreements with your peers over how much data they are willing to receive from you. The clearing process is completely independent. Cash, cat pictures, happy feelings, revolutionary zeal. All are acceptable currency if the owner of the router agrees with you.
How is that any different than it already works? Anybody can configure their network to accept traffic or not from whoever they like. What's missing is an efficient method for clearing small payments -- but that's a general problem, not anything specific to internet links.
Shameless plug, we are working on Interledger[1]. Which is designed to be an open-clearing system based on packetised payments. It works really well for low value, low-friction payments. Our initial use-case is targeting web monetisation which doesn't work on existing payment rails due to the costs.
Part of the work coming up this year will to be going after the retail payments space and I think the challenges you mentioned are definitely going to be the ones we are going to have to solve next.
1. www.interledger.org
That also makes it relatively unsuitable for long-term storage of stale information. Active communities will last, but archived ones with no new content won’t, and eventually be gone.
I mean, can some company come along and build "archived searchable aether" as a service?
Histories are useful only if they relate to an active entity and you have the option to kill that entity and move over to some other name - and that new name would be impossible to link from the old one.
I know it's hard to thwart people like Oxy, that are stacking all sides. Providing both VPN and proxy services. Then using those services to sell anonymized scraping as a service. Then collating online data obtained from scraping with offline data to create ridiculously detailed profiles about people. (Then leaking them all by the way. As though collecting them wasn't bad enough. [1]) But my ideal distributed social media would be architected to defeat something like Oxy. Because Oxy, PDL etc are already behemoths and the government has more comprehensive data on you than even Oxy.
I realize though that aether is probably just something to "keep the honest people out" so to speak. But my vote for a "nice to have" would be the ability to thwart some of the data mining and linking baddies that are out there.
For example, one of the problems with Ethereum's Whisper protocol (decentralized messaging) has been that nodes aren't incentivised to process messages. There are other problems too and some efforts are being made to generally improve the tech[+].
(Context: I have a Patreon here: https://patreon.com/nehbit)
I generally try to only charge it on the months I push a new release on, and otherwise keep it paused. I’ve found that’s a good way for me to not stress out when my ‘day job’ gets too hectic.
https://hn.algolia.com/?dateRange=all&page=0&prefix=false&qu...
You make things sound far too benign. Where do death threads fall in this classification, are they "free speech" or "objectionable material"? Where do disinformation bot armies fall? There's a lot more to content moderation than just "that person doesn't want to see boobies or the f-word".
Disinformation campaigns? Well again, if you don't want to be misinformed, pick your gatekeeper. Probably all the better. If somebody's going to protect me from being misinformed, I want to know what their biases are. I don't trust Twitter or Facebook; it's not a simple and objective task.
I think your point that this stuff is complicated is exactly my point. It is complicated, and people have different preferences. No one company can be the gatekeeper that satisfies the entire planet's preferences (at least, not under the current models).
I don't want to over sell it, but that kind of investment in something which is unlikely to yield revenue from a public company is at least unusual.
At twitter's scale, 5 people are a team you put together to decide if they are going to migrate to X library/framework for an internal tool.
This is not a task where you want to be frugal in terms of resources.
Exactly the kind of task you want a small team doing the initial research and planning on otherwise it'll be bikeshed central in every meeting, surely.
I certainly hope not. While these were successful it does not mean that their projects were good or that they contributed for the good of mankind rather than for its downfall.
I guess we need a way to share things we've liked?
Retweets should be representable through RSS entries.
Hashtags are just searchable text.
See https://aaronparecki.com for an example and https://indieweb.org/friendly and https://indieweb.org/Webmention
Twitter is fighting to stay relevant, they dropped to Top 35 website from Top 14 in last several months.
This is an opportunity for them to become a global standard, and that is worth it.
The most important piece is that they can scale when they pull it off. I've been working on decentralized tech for a while (I run https://github.com/amark/gun) and can say it takes a lot of attention that most projects (dare I say "blockchains") have no scaling abilities at all.
I don't think they would ever do that. It would be really dangerous to implement and would possibly expose an unfiltered stream of partially illegal content.
I mean, can you imagine "validate if this is an unsolicited dickpic", or "verify if this is a dox attempt", or "verify this is cp"? This is one area that can never be exposed to non-employees.
New social network? Blockchain!
New news platform? Blockchain!
I get what blockchain is. I get how it's used in cryptocurrency transactions. But will someone please explain why it's being incorporated into industries that always have done (and IMO can continue to do) just fine without it?
I skim through this [1] list, and see no clear benefit to incorporating blockchain into most/all of these? Most of them seem to be using it as 'secure database' or something.
[1]: https://www.fool.com/investing/2018/04/11/20-real-world-uses...
That is all.
As long as data is stored in Twitter managed database, it would never work.
By open we mean everyone gets full permissions, tovarisch!
And a few open data centers to handle the load to the database and to the storage.
The user pays the storage cost upfront. However it is suitable only for file like content, not mutable content like databases AFAIK.
1: it doesn't define enough metadata ("this is an image", etc)
2: it doesn't handle backtracking well, if one of the recipient servers misses a message (servers are just supposed to retry a given number of times).
3: it has no security/authentication concept.
All of these could be bolted on regular e-mail, but calling it e-mail at that point would be a travesty, and we might as well call that a new protocol. And if we are writing a new protocol, we might as well take a few liberties to address the shortcommings of the first (one could argue against the last part).
More competition in this domain is a very good sign IMHO.
Let's see what Telegram Open Network will offer.
Just set up AWS EventBridge (or Azure EventGrid), or any of a dozen other options that don't require you to administer a server. And then use AWS Lambda, Azure Functions, Azure Logic Apps, Zapier, etc, etc. Don't maintain a server that runs a bunch of apps.
We care about what we're managing, and care about being responsible for it, we just want to manage it at the right level of abstraction. Let the cloud provider manage and be responsible for the hardware.