Distributed protocols that combat economy of scale

newer
Snowden sasys FBI can unlock iPhone

older
USG Support of Motion to Compel...

rysiek

6 Mar 2016 6 Mar '16

3:41 p.m.

Hi, as "A Miner Problem" shows, even a distributed, p2p solution like BitCoin suffers from a secondary centralisation flaw. Simply put, economy of scale lets the biggest player or a cartel of thoise basically undermine the "decentralized" part and take control over the network. A similar thing has happened with e-mail and GMail. GMail basically dictates the rules in the e-mail world, simply because they're the largest and have the most users. Again, economy of scale is to blame. So let me pose a question here: is it possible to design a protocol that does not succumb to economy of scale-based secondary centralisation? Is it possible to design a protocol that does not lend itself to economy of scale? -- Pozdrawiam, Michał "rysiek" Woźniak Zmieniam klucz GPG :: http://rys.io/pl/147 GPG Key Transition :: http://rys.io/en/147

Attachments:

signature.asc (application/pgp-signature — 931 bytes)

Show replies by date

agave

6 Mar 6 Mar

4:20 p.m.

On Sun, 06 Mar 2016 16:41:05 +0100 rysiek wrote:

...

Hi,

as "A Miner Problem" shows, even a distributed, p2p solution like BitCoin suffers from a secondary centralisation flaw. Simply put, economy of scale lets the biggest player or a cartel of thoise basically undermine the "decentralized" part and take control over the network.

A similar thing has happened with e-mail and GMail. GMail basically dictates the rules in the e-mail world, simply because they're the largest and have the most users. Again, economy of scale is to blame.

So let me pose a question here: is it possible to design a protocol that does not succumb to economy of scale-based secondary centralisation? Is it possible to design a protocol that does not lend itself to economy of scale?

There's always things like TorChat for instant messaging and IPFS for content distribution. There's no concept of any centralisation in either - they're completely peer-to-peer. I'm not sure why you drew a comparison to e-mail - it's inherently centralized, just among several servers.

rysiek

9:13 p.m.

Dnia niedziela, 6 marca 2016 11:20:54 piszesz:

...

There's always things like TorChat for instant messaging and IPFS for content distribution. There's no concept of any centralisation in either - they're completely peer-to-peer.

Such was also the idea in BitCoin. As we can see, there's a secondary (not protocol-level) centralisation happening there. Cartels emerge, because economy of scale makes them viable, and indeed profitable. Yes, TorChat, IPFS, Tox are good examples. I'd be hard-pressed to identify any economy of scale possible there, I guess.

...

I'm not sure why you drew a comparison to e-mail - it's inherently centralized, just among several servers.

It's not centralized in a way Facebook is, for example. While e-mail was obviously never a p2p system, it was not a centralized system. The idea was decentralized, federated among *many* servers. But economy of scale did its magic and now we have a small number of companies controlling almost all of it. -- Pozdrawiam, Michał "rysiek" Woźniak Zmieniam klucz GPG :: http://rys.io/pl/147 GPG Key Transition :: http://rys.io/en/147

agave

9:52 p.m.

On Sun, 06 Mar 2016 22:13:41 +0100 rysiek wrote:

...

It's not centralized in a way Facebook is, for example. While e-mail was obviously never a p2p system, it was not a centralized system. The idea was decentralized, federated among *many* servers.

Oh, I see what you mean.

...

Such was also the idea in BitCoin. As we can see, there's a secondary (not protocol-level) centralisation happening there. Cartels emerge, because economy of scale makes them viable, and indeed profitable.

Hmm, I'm not sure if there's any particular way to solve that problem for BitCoin. I can't imagine any ways of establishing consensus among a completely decentralized protocol other than by popular consensus of the participants, and popular consensus can be manipulated easily as the post you mentioned shows. Maybe decentralization ought to be left to communications only?

rysiek

11:21 p.m.

Dnia niedziela, 6 marca 2016 16:52:13 agave pisze:

...

...
Such was also the idea in BitCoin. As we can see, there's a secondary (not protocol-level) centralisation happening there. Cartels emerge, because economy of scale makes them viable, and indeed profitable.

Hmm, I'm not sure if there's any particular way to solve that problem for BitCoin. I can't imagine any ways of establishing consensus among a completely decentralized protocol other than by popular consensus of the participants, and popular consensus can be manipulated easily as the post you mentioned shows.

Indeed, and the whole mining thing was supposed to protect BitCoin from such manipulation. Granted, it made it hard, but at some point, with enough money/value involved, economy of scale made it viable. And that's what bothers me, and that's my question: Is it possible to design a protocol in a way that makes economy of scale not work? The BitCoin example is particularily complex, as there's no good way to tell, on the protocol level, "all those nodes are part of a single cartel, controlled by a single entity". But maybe there is a way of basing this on behaviour? If nodes X, Y, Z behave in a very similar manner, lower their weight in consensual decisions? Apologies for being vague; I just feel there's something to it, but can't really put my finger on it. I see economies of scale driving a lot of not-so-great outcomes (centralisation of e-mail providers; centralisation of ISPs and the general Internet infrastructure, introducing single points of failure in many places). And that makes me wonder what can be done about it.

...

Maybe decentralization ought to be left to communications only?

How so? -- Pozdrawiam, Michał "rysiek" Woźniak Zmieniam klucz GPG :: http://rys.io/pl/147 GPG Key Transition :: http://rys.io/en/147

agave

7 Mar 7 Mar

2:52 a.m.

On Mon, 07 Mar 2016 00:21:30 +0100 rysiek wrote:

...

The BitCoin example is particularily complex, as there's no good way to tell, on the protocol level, "all those nodes are part of a single cartel, controlled by a single entity". But maybe there is a way of basing this on behaviour? If nodes X, Y, Z behave in a very similar manner, lower their weight in consensual decisions?

That's a good idea. It would need to be fine-tuned, though; I'm reminded of the days when I used Windows with antivirus software that would very often false-flag and delete things that I needed. Although, I suppose the consequences wouldn't nearly be as severe.

...

I see economies of scale driving a lot of not-so-great outcomes (centralisation of e-mail providers; centralisation of ISPs and the general Internet infrastructure, introducing single points of failure in many places). And that makes me wonder what can be done about it.

...
Maybe decentralization ought to be left to communications only?

How so?

I was going to say that because the simple sharing of information from one person to another doesn't rely on messy things like consensus that come from managing a distributed database like the BitCoin block chain, it's thus immune to economies of scale and fits snugly and easily into the decentralization model, but I didn't really think of the physical infrastructure of communication. Mesh networks do effectively address the problem of centralized infrastructure, though, but it's unfortunate that they aren't yet popular or saturated with resources enough to serve as a viable alternative in most cases. freifunk.net in Germany is a good example. I guess this goes back to your example of GMail again. Mesh networks aren't viable yet because tons of people don't use them yet, and tons of people don't use them yet because tons of people don't use them yet. The network effect is very unfortunate.

Shawn K. Quinn

1:11 a.m.

On Sun, 2016-03-06 at 11:20 -0500, agave wrote:

...

On Sun, 06 Mar 2016 16:41:05 +0100 rysiek wrote:

...
Hi,

as "A Miner Problem" shows, even a distributed, p2p solution like BitCoin suffers from a secondary centralisation flaw. Simply put, economy of scale lets the biggest player or a cartel of thoise basically undermine the "decentralized" part and take control over the network.

A similar thing has happened with e-mail and GMail. GMail basically dictates the rules in the e-mail world, simply because they're the largest and have the most users. Again, economy of scale is to blame.

So let me pose a question here: is it possible to design a protocol that does not succumb to economy of scale-based secondary centralisation? Is it possible to design a protocol that does not lend itself to economy of scale?

There's always things like TorChat for instant messaging and IPFS for content distribution. There's no concept of any centralisation in either - they're completely peer-to-peer. I'm not sure why you drew a comparison to e-mail - it's inherently centralized, just among several servers.

As used today, with many people using services like GMail, Hotmail, etc it could be considered centralized. However, in the sense that I can set up my own email server, and have someone email me at a given address without having to get approved by a central authority first, it is decentralized. Email doesn't even strictly require DNS to work, as there is an email address syntax (rarely used these days, but still technically valid) for addressing email directly to a given userid at a specific IP address by dotted quad. -- Shawn K. Quinn

juan

6 Mar 6 Mar

9:34 p.m.

On Sun, 06 Mar 2016 16:41:05 +0100 rysiek wrote:

...

Hi,

as "A Miner Problem" shows,

Sorry is there any other source/analysis for popescu's claims?

...

even a distributed, p2p solution like BitCoin suffers from a secondary centralisation flaw.

Actually, if the cartel exists, what it shows is a flaw in the bitcoin protocol. A look at the original bitcoin paper suggests that the author(s) didn't bother too much with the network side of things. Didn't forsee the use of specialized hardware. Et cetera. "The network itself requires minimal structure." "The network is robust in its unstructured simplicity." Looks like it isn't...

...

Simply put, economy of scale lets the biggest player or a cartel of thoise basically undermine the "decentralized" part and take control over the network.

...

A similar thing has happened with e-mail and GMail.

Good example. Because google isn't the result of 'economy of scale' but it's a great example of what the state can accomplish using a 'private' facade.

...

GMail basically dictates the rules in the e-mail world, simply because they're the largest and have the most users. Again, economy of scale is to blame.

So let me pose a question here: is it possible to design a protocol that does not succumb to economy of scale-based secondary centralisation? Is it possible to design a protocol that does not lend itself to economy of scale?

grarpamp

7 Mar 7 Mar

2:05 a.m.

On 3/6/16, juan wrote:

...

Good example. Because google isn't the result of 'economy of scale' but it's a great example of what the state can accomplish using a 'private' facade.

http://news.slashdot.org/story/16/03/06/1834211/eric-schmidt-gets-a-job-at-t...

juan

3:23 a.m.

On Sun, 6 Mar 2016 21:05:08 -0500 grarpamp wrote:

...

On 3/6/16, juan wrote:

...
Good example. Because google isn't the result of 'economy of scale' but it's a great example of what the state can accomplish using a 'private' facade.

http://news.slashdot.org/story/16/03/06/1834211/eric-schmidt-gets-a-job-at-t...

But thanks jesus and marx, the good government is there to regulate the evil corporations. Just ask stallman and the rest of lefty geniuses...

Zenaan Harkness

7:23 a.m.

On 3/7/16, juan wrote:

...

On Sun, 6 Mar 2016 21:05:08 -0500 grarpamp wrote:

...
On 3/6/16, juan wrote:

...
Good example. Because google isn't the result of 'economy of scale' but it's a great example of what the state can accomplish using a 'private' facade.

http://news.slashdot.org/story/16/03/06/1834211/eric-schmidt-gets-a-job-at-t...

But thanks jesus and marx, the good government is there to regulate the evil corporations. Just ask stallman and the rest of lefty geniuses...

Oh come on! That's so facetious! You should thank Google, Facebook and Citigroup for government, at least they pay for the elections... it's all very above board.

Georgi Guninski

2:18 p.m.

On Sun, Mar 06, 2016 at 04:41:05PM +0100, rysiek wrote:

...

So let me pose a question here: is it possible to design a protocol that does not succumb to economy of scale-based secondary centralisation? Is it possible to design a protocol that does not lend itself to economy of scale?

Maybe it is possible to design such, but design is not that important. You _must_ implement it in the real world(tm). Some of your threats include: people technology (backdoors, bugs, sickyouaruty) TLAs governments Won't be surprised if bitcoin is illegal in parts of the real world(tm).

rysiek

10 Mar 10 Mar

10:12 p.m.

Dnia poniedziałek, 7 marca 2016 16:18:31 Georgi Guninski pisze:

...

On Sun, Mar 06, 2016 at 04:41:05PM +0100, rysiek wrote:

...
So let me pose a question here: is it possible to design a protocol that does not succumb to economy of scale-based secondary centralisation? Is it possible to design a protocol that does not lend itself to economy of scale? Maybe it is possible to design such, but design is not that important.

You _must_ implement it in the real world(tm).

Don't I need a protocol first? :)

...

Some of your threats include: (...) Won't be surprised if bitcoin is illegal in parts of the real world(tm).

Thank you, Captain Obvious. Would you be interested in this rock? -- Pozdrawiam, Michał "rysiek" Woźniak Zmieniam klucz GPG :: http://rys.io/pl/147 GPG Key Transition :: http://rys.io/en/147

3130

Age (days ago)

3134

Last active (days ago)

List overview

Download

12 comments

7 participants

participants (7)

agave
Georgi Guninski
grarpamp
juan
rysiek
Shawn K. Quinn
Zenaan Harkness