Topic: Dear Satoshi Nakamoto (Read 3732 times)

Apart from the economics basics, you may want to learn more about how Google is actually built inside

Banking worked in the past and continues to be working so far simply because there was no such thing as decentralized "banking" in a trustless network in the past (in the form of payment channels). Strictly speaking, we are not there yet (you should know that yourself) but for issues not related to its advantages or deficiencies. Personally, if we assume free competition (i.e. no governments with their laws), I'm inclined to think that a decentralized payment system will quickly crowd out the current banking system. In fact, I don't even need to address this issue as such since this is how Internet itself is built. To access hosts on your network you use your nearest cheap router or switch (which is able to process only a few requests per second) and you don't have to walk to the top-level nodes (which otherwise should be able to process millions of such requests per second as well as have millions of ports) just to reach out for your neighbor. In this way, Internet is also scalable without limits conceptually (if we discard delays due to signal processing and the finite speed of light). You just need to add more digits (octets) to an address pool as the network expands

You're right.  That's why banking never worked, and why nobody's using facebook, amazon or google.  Sorry for my verbiage, your argument is crystal clear.

You may want to learn the basics of economics. Top processors are times more expensive than the mainstream ones while their processing power may be only a dozen percent higher


Technically better doesn't mean economically cheaper

In fact, the direct opposite is almost always true. For a better efficiency you have to pay a higher price. But that's not the main problem here. The main problem is that to get a somewhat more efficient system you have to pay a significantly higher price. In short, you just don't know what you are talking about. But even that is inconsequential to the discussion. The scalability of a decentralized payment system is built in, i.e. it scales up together with its expansion (and thus its processing capacity is effectively unlimited). With a centralized system, its processing capacity remains the same, and if the number of transactions increases beyond its limits, you will have to buy new capacity. As simple as it gets. And I hope that will help you shut at last the fountain of empty verbiage you produce

How can that be ?  If it is purely technically more efficient, that's what determines the (open or hidden) cost.


Mmm.  That's hard logic 
This must be the reason why people don't use facebook.

Seriously, your "argument" doesn't hold water.  If you agree on the fact that technically, the centralized option is cheaper, then that's all that matters, because that is the real cost.  The hidden cost for the decentralized users will make them ready to pay a fee of the height of their hidden cost, and that fee can, as per technically lower cost, largely finance the centralized option.  Running an online node at home all the time, or having to pay a fee of $20 a year, almost all people will go for the fee.

As I said higher-up, I remember in the 90-ies, internet being exactly such kind of distributed system, where some people had some hardware at their homes, and connected to others through leased telephone lines, serving as a meshed network of decentralized hubs for the internet.  Others could connect to them.  That quickly got replaced by centralized bigger data centres to which people were connected in a wheel-and-spokes way, and that's now the general structure of the internet.  Simply because one data centre with a few racks of electronics and 3 employees, is way more efficient and cheaper than the decentralized model.

I see what I see, and I think it's no good trying to call white black and vice versa. The fact is that you chose to bring forward a devastatingly deceitful and fraudulent argument, which, as to me, was a deliberate act and which tells it all if you have ears


I've told you already that this is not a technical matter

And I never meant it to be so since this is all about economics. You can take 1,000 independent nodes and 1 unit with the same processing power as theirs combined, and it will be more efficient on purely technical ground but it will be 10-100x more expensive than these 1,000 independent nodes. The latter would likely cost you, say, 1M dollars in total while the former dozen million dollars per unit (and don't forget about redundancy). The numbers are purely random but the relationship holds. And you basically don't have to pay anything for these processing nodes, they are paying for themselves (as I also told you already). In other words, their payment processing power comes at no cost to the system. No need trying to fool anyone here, you may want to try your luck elsewhere

I think that this is probably the best time for satoshi to make an appearance. Bitcoin community is divided by never before right now and the drama is all over the place, although the price is still staying high.

His words left on the forum has been speculated upon, and bitcoin "as satoshi intended it to be" is sort of an ideal right now.

I really don't think that he'll make an appearance though. I think he'll respect the community's decision as well.

I'm not.  Try to build logical arguments instead of trying to make a trial of intentions on my back, and try to understand what I'm writing instead of thinking I'm ill-intentioned.

I'm telling you that a centralized but eventually distributed communication system is always, purely technically, more efficient and more performing than a decentralized one, simply because the centralized distributed one can in principle take on exactly the same TECHNICAL topology as the decentralized one, but it has more possibilities than the decentralized one.  So the set of possible technical configurations of a decentralized system is a SUBSET of the set of possible configurations of a centralized system ; as such, any technical metric on the whole set will never reach a maximum in the subset, that will not be reached in the overall set.

You claimed otherwise, that centralized systems are fundamentally limited in their capacity, while decentralized systems aren't, which is in obvious contradiction to the above, rather simple, observation.

I illustrated two forms of a centralized system: one with a bank office serving a million of customers (A), and one with boxes for every 1000 customers (B).

I compared that to two forms of decentralized systems: one where each node is doing processing (C), and one where only one out of 1000 nodes is doing processing (D).  (considering "decentralized" systems where only one out of a million nodes does the processing, begs the question of decentralization, although we could do the exercise, and the result wouldn't change).

In all cases, the centralized system can largely be financed by the hidden costs of the decentralized system.  The lowest cost system is (A).  It can be financed with the hidden costs of (C) (the 20 dollars per year times a million).
But as you weren't happy, I showed that (B) and (D) are technically equivalent and that (B) can be financed by the fees for (D) - which is the essence of my "theorem" (the subset).   Although most probably, (A) is cheaper than (B), so the bank will most probably go for (A) - which the decentralized system cannot do unless it becomes de-facto centralized.

You knowingly tried to compare what you simply can't compare

And now when you got caught with that (for the sake of your "proof"), i.e. comparing a full node processing payments with a smartphone which is only requesting that processing ("a computer running all or most of the time... and only a smartphone application"), you backpedal your comparison. I'm not much interested in this sort of discussion since it is obvious that you are just weaseling out of your own pitfalls. I may be unfamiliar with some technical details but you can't hope to push through such stuff. Anyway, I won't let you get away with it since it shows that you don't have substance behind your claims if you stick to these tricks (you should understand what it means for the credibility of your claims)

I'm comparing a decentralized system, where your participant needs to run a full node and be online to process LN transactions, to a centralized system where the bank's offices do the processing, and the customer only has to chime in with his smart phone application when he wants to do a payment, if it isn't the terminal at the cashier that is doing it for him, because these are the two systems we are comparing.

My claim is that a centralized (even though maybe distributed) system is always at least as performing than a decentralized system, because in the worst case, the centralized system can take on exactly the same form as the decentralized one ; a hypothesis I didn't even need in the particular case we're considering, but which I will try to take on here now.

Your claim is that a decentralized system "pays for itself".  My claim is that the hidden cost of this decentralized system are largely capable to pay for the visible cost of a few distributed, but centralized nodes that handle all the traffic.

If each of the participants in the decentralized system works as a "payment processor", then the (small) overhead of each of these nodes at the place of each of these network users, is in total much larger than the cost of a few centralized nodes that handle the same load, where the customers don't have to bother at all, and just pay a small fee for the service provided.


Ah, but if you are considering that not every node in the decentralized network is a "payment processor", but only "one out of X", then the argument becomes even easier: replace these decentralized payment processors by centrally controlled, distributed controllers ; it would most probably not be practical, because X would be too small a number: if X is 1000, that means, every 1 out of 1000 participants has a "payment processor node" in the decentralized system then we would have "banking agencies" which handle only 1000 customers.  There are better economies of scale to be had by putting many of these into one single small data centre.  But for sake of argument: each of these decentralized "payment processing nodes" would be replaced by a "box from the bank" that you place in your home, but cannot access.  If you do so, you get a "free bank account", while others have to pay a small fee.  You cannot do anything to the box: it is just connected to your network, is entirely secured, and is remote-administered by the bank's officials.  You are just a bank customer, that accepts to place this box at your home, in return for the favour of a free bank account.  

The network configuration is then identical to the decentralized one, so the more customers, the more network nodes the bank has, in exactly the same way as the decentralized system, but the difference is that the box is entirely controlled remotely by the bank (that's why it is a distributed system and not a decentralized system).

As I said, the bank won't do that, because it is more hassle to put 1000 boxes at 1000 customers, rather than to have one single office that can handle 1000 000 customers directly.  The 1000 boxes at 1000 users is the decentralized system ; one hub with 1000 000 customers is most probably what LN will give us for reasons of economies of scale.

Yes, the bank has to finance a box for every 1000 customers.  However, the decentralized node operator will also finance his computer and network with the fees he takes.  So if the bank asks these same fees to customers (not having a box at their place), it can normally finance without problems the box it has to put at the special customer's place.

But it can apply economies of scale, and use more performing data centres.

You are again severely distorting the facts as they better serve your point

In this case, you compare things which you simply can't compare. You compare a full node running as a payment channel with a smartphone with which you can only make payments, not process them. I guess this is not a valid comparison by any means. First, you should compare a payment processing node with your rack full of "industrial PCs" (and consequently their costs), and, second, I could access a payment node with the same smartphone and make a payment through it just as easily as you would contact your local bank with it to do basically the same stuff. I think you may want to restate your other ideas since I didn't get much of what you tried to say. In the very least, you should address the flaw in your argument that I have just pointed out

[Quote from: satoshi on October 03, 2010, 13:07:28]

https://bitcointalksearch.org/topic/m.23049


So, Satoshi already chimed in on this long ago....it was all already foreseen.

I don't ignore this.  But for the sake of my "proof", there is a difference between the bandwidth and a computer having up and running all or most of the time to "serve on the network", and the bandwidth, and only a smartphone application, to contact your local bank to do payments (like I do now).  I agree that the difference is not big, but we're talking about very tiny costs in any case per customer/user.  My point is that if you charge a fee comparable to the cost induced by having to have a computer running, and the bandwidth consumed, by millions of individual users, you have more than enough money to run a centralized, eventually distributed system that has the same or bigger capacity, handling what these people do.  Hell, most probably, the usual household internet bandwidth and a running PC can handle the *centralized* needs of a few tens or even hundreds of thousands of "bank customers" !   Let's estimate: if a centralized transaction costs, say, 10 KB (big, no !), and a typical customer does 10 transactions a day, that would put him at 100 KB a day of bandwidth.  Now, I have a modest 1 MB/s link, but I could get much more if I wanted to.  That means I can serve 10 customers per second, or 10 000 customers in 15 minutes at maximum, so 100 000 a day seems feasible.  With a single simple PC and a household bandwidth.   So a local bank office, with a very small data centre in its basement, can easily handle several millions of customers.  Centralized banking is really, really simple as compared to decentralized banking, because there's no consensus to be found, there's no trustlessness to be solved, etc....  

Suppose that the "negligible cost" of running your PC and your network bandwidth is $20 a year.   Hell, then the small data centre 10 times bigger than my home PC and some more powerful network connection, serving a million customers, could reap in $20 million per year.  With $20 million per year, you can power such a system with personnel and everything !

Imagine, a banking fee of $20 a year ?  Most banks are way more expensive, so the "network cost" is really, really nothing.

Can you imagine people finding $20 a year too expensive for their banking affairs, and wanting to do everything with their own PC ?   Hell, people spent of the order of $100 for a Windows licence, while they could have linux for free.


No, as you can see, a tiny data centre (well, simply a rack with a few industrial PC !) can easily handle a million customers or more without any hassle in a centralized network, at a cost that is probably even smaller than the hidden cost of running a decentralized system yourself.


The point is that the cost per customer in the trusted network is smaller or at worst comparable to the self-inflicted cost of the trustless system.  

You don't build trustless systems because they are economically more performant, they aren't.  They are way more expensive.  However, you build trustless systems because for political reasons, you are willing to bear the extra cost and hassle to avoid the power of centralisation.  But it comes with a price.

I hope you not only spoke on the lips only. Show idea that you think not to fail. Post your profile are here, i will remember you, we as hunter is readily worked if it is the best, remember that. I want to a real job. Not only rhetoric a moment

The banking density would come at a cost, right?


Indeed, their computer equipment just like network bandwidth would cost them something

But you seem to deliberately ignore to see my point here. If they are using Bitcoin (let's assume Bitcoin will be the first to introduce decentralized payment channels), they would have to bear these costs anyway, while turning their hardware into payment processing nodes would cost them nothing apart from what they have to pay anyway just due to simple use of Bitcoin. Why do you choose to completely ignore this?


Your so-called proof lacks one crucial point which instantaneously renders it null and void. Your distributed centralized system requires trusted network, and to expand it would require you to pay tremendous amounts of money. This is the point which you choose to ignore as well. As soon as you take into account this factor, your system becomes economically losing since you can add only trusted nodes to it (read you have to pay for them) while in a trustless network it doesn't matter (since every node pays for itself if it wants to use Bitcoin). Before it was not possible because there was no decentralized currency circulating in a trustless network. Modern fiat currencies are already decentralized but only within a banking system (where commercial banks are trustees of a central bank)

Maybe it was agreed at the time, but it does not mean that people will accept it now, things change depending on the market, never keep something too long.

And what does it change ?  If the banking density in those countries attains the banking density in western countries, why should there be a bigger problem for them to have credit cards than for us ?  Really, centralized systems have absolutely no problem with the amount of customers.  Facebook and amazon need to handle more data and processing than banking for the same number of customers, and they don't have problems.


This is where your reasoning fails, because of course there are expenses by all these customers of a decentralized network: the use of their computing equipment and network bandwidth.  You can think that it is tiny, but then, with economies of scale, putting them on more centralized entities is even cheaper.  1 million people running software on 1 million computers at their homes, and using the bandwidth of their internet connections, is in total much more expensive, than a few distributed data centers serving all of them.  This is why the internet is not a P2P network, but has a spoke structure.  The internet is not, as it was in the beginning, a network of nodes linked together with point-to-point lines, making up a mesh: it has a hierarchical structure with many customers connected to centralized even though distributed ISP.  You don't connect with a wire to your neighbour, who connects to his neighbour, etc... as it was in the beginning.  You use lines to a centrally served entity.

In other words, the total cost for a large network is less per subscriber in a centralized, distributed architecture, than the total cost of a true P2P network.  As such, the centralized distributed architecture can ask a small fee to its participants, which is lower than the self-induced costs they would have to run a true P2P system. 

My "proof" is very simple: if ever it were possible that a true P2P architecture were more performing than a distributed, centralized system, the distributed, centralized system can take on exactly the same architecture as the P2P system, and hence will be just as performing.

I'm just talking here about the technical performance, not about any political viewpoint.  Centralized, distributed systems are, on the purely technical side, always superior to decentralized P2P networks.  However, decentralized P2P networks have political advantages, but that is their *sole* reason of existence: the lack of central authority, the lack of a single point of political failure.

The problem is that the technical advantages, by economies of scale, are such, that centralized, distributed systems tend to win almost always over true P2P decentralized systems, unless the cost and difficulty of the P2P system is accepted for its political advantage.  My point is that this is not the case in payment systems.  The political desire to be "independent" with money and payment, apart from some useful idiots like me and maybe you, is non-existent with the mainstream public.  It is even not there for communication. 

Everybody can set up a web page at home.  True P2P "social networking".  Nobody does.  Everybody uses facebook.  Very few people set up a Tor node.

I also avoid using cash as much as possible

Though I have some cash reserves in case of emergency, naturally. But I understand that if I'm extensively using payment cards, that doesn't mean that everyone around me uses them as well (though some certainly do). It would be an obvious case of selection bias (I hope you know what I mean). Anyway, we have 1.4B Chinese and almost as many Indians. Do you really think that even one tenth of them uses payment cards? I doubt it


I guess you first have to explain why anyone would ever want that

If you are going to say that banks and payment card companies are going to pay for these nodes, then you have to explain how that would be conceptually different from just buying more processing equipment. As you can see, this still comes down to expenses, while in case of an expanding decentralized payment network, the capacity expands on its own and at no cost at that. Further, what you say is just pointless as such since what you suggest means losing control over payments and you no longer remain "centralized". Basically, you just confirm the infinite scalability of a payment mesh or grid built around independent payment channels in a trustless network (banks are trustees of MasterCard and friends, which is a key difference here). As I said before, you have to stick to your guns, i.e. not change your point halfway

Yep that one post is the I told you so.
On the other hand consensus at that time probably didn't seem like a big issue but alas it was.

There's maybe a cultural difference then.  Where I live, almost everything is done with credit cards.  At the supermarket, at the gas station, even at the baker's, grocery shops.  Even buying coffee, although that is maybe 50/50 done with cash.  I rarely have more than 10-20 Euros of cash in my wallet, unless I want to do something without taxes or so.  I essentially pay everything with credit cards/bank cards (no difference, same piece of plastic), down to amounts of about 1 or 2 Euro.
I'm doing hundreds of transactions every month only using credit/bank cards, and so are most of the citizens around me.  Essentially only elder people, children, and drugs traffickers pay mostly in cash.


Why do I have a feeling that you are trying to sidestep the issue?