August 12, 2020

Death from fake decentralization

BitTorrent was not the first such network. Why did he survive and the rest perish? How did he become the main a tool for sharing large files on the Internet, despite the fact that previous systems were easier and more convenient for users? What lesson can be learned from this?

</p>

Why did BitTorrent survive and eDonkey die? What does this mean for the future of electronic payments?

I suggest looking at how these networks work:

eDonkey 2000

In eDonkey 2000, nodes containing a list of availablefiles are a kind of "server" and form a network. If a client wants to share a file (to become a “seed” in the BitTorrent language, take a triangular diagram of this network), then he connects to one of the servers and reports that he has a file for exchange. Clients interested in this file (“peers” in the BitTorrent language) send a request to the servers, find the sources of the file (triangle and each other) and start downloading.

Like modern cryptocurrency networks, this“decentralized” system: directory servers are served by volunteers. On the Bitcoin or Ethereum network, miners or full clients do this. The word “decentralized” is enclosed in quotation marks, because although the network is distributed around the world and seems invulnerable, it has one problem that killed eDonkey.

To download a file, you need to connectto the network of directory servers and find all sources of the file. The edonkey protocol had a very primitive request mechanism. At that time, distributed hash tables (DHTs) were not yet invented: a request for a file and peer availability was sent to each server. Each client looking for a file sent a request to every directory server it knew increasing the load on the entire network. It turned out that this load is not necessary at all.

If you carefully look at the figure, you cansee three separate networks: a server network, an orange file network, and a green file network. In theory, only two directory servers are needed to process requests for these files:

</p>

In order to find out the source of the file, the included servers are not needed. You do not even need to know that other servers exist.

Bittorrent

This is BitTorrent. He has a little problem:

How do customers find the appropriate “servers” or, to use the BitTorrent language, trackers?

This is where the torrent file comes into play: it contains the address of the nodes that carry information about the "sids" (the original sources of the files) and the "peers" (all clients who try to download the file).

Of course, a new level of complexity arises here.and some inefficiency regarding the exchange process: first you need to find a verified source of torrent files, then make sure that this source has the right torrent file. Hundreds of trackers and sources can exist for one file, but if your trusted source does not have them, then you are out of luck. But that's not all BitTorrent problems.

Union as the best form of decentralization

If you are looking for an orange file, then you do not caregreen file, its tracker and sources. This means that you will never send a request to the green tracker. But other people may need millions of different files, and in extreme cases, there may be millions of independent trackers that don't need to know about each other, whose traffic does not intersect. This is a much more efficient and infinitely scalable system. With the advent of DHT, it has become even more effective. Now participants can independently play the role of trackers, reducing the load on other trackers. However, such a system has low availability.

  • Participants need to look for the tracker in trusted sources.
  • They need to rely on the tracker in that it provides the correct file.

But, as practice shows, these problems are insignificant. Even despite constant problems with the law, there are well-known proven sources of torrent files.

Lesson for electronic payments

Is it possible to use this system for electronicpayments? In systems in which, unlike Bitcoin and Ethereum, all participants do not need to know about each new client of the network? On systems that store most of the information about transactions in a trusted fragment, providing a low cost for these transactions? On systems in which participants trust fragments? On systems that can work legally? Or maybe even in systems where fragments compete with each other and try to become better, cheaper and more effective?

It turns out that such a system already exists. This is a banking system.

Take your time to claim that the banking systemis offline, slow, expensive and inefficient. Most likely, these are the problems of your bank. Even banks with imperfect technologies in such "developing countries" as Thailand, not everything is so bad:

  • Transactions are instant and free, regardless of whether you or the recipient is a client of one bank. Both you and the recipient will receive a text message as soon as you click the "pay" button. Is free.
  • You can pay by phone number (PromptPay). Yes, "like paypal," but for free. Try same with bitcoin.

In addition, there are many developingpayment processors that work "on top" of the banking system, for example, LINE Pay or WeChat Pay. They allow you to instantly and free of charge make payments using your mobile phone, pay for tickets for trains, breakfast, coffee. They do not use blockchain - they do not need it.

Cryptocurrency systems work too slowly, expensively and inefficiently (and no, they cannot be improved without breaking their basic promises).

Cryptocurrencies have one property that nobanking systems: as Simon Morris wrote in his articles, this is a built-in ability to break the rules. But let's be honest: do you want other participants could break the rules in the system with your money?

</p>