5 ms·

Awesome deep dive into the space. The two positions I object to are: > In some ways, Bitcoin could be considered the first decentralized application since it r

by hudon 9y ago

Awesome deep dive into the space. The two positions I object to are: > In some ways, Bitcoin could be considered the first decentralized application since it runs on blockchain technology, is fully open-source, and runs without a central authority. Decentralization is not a new thing that only came into existence with Bitcoin... Torrents are decentralized, the email protocol is decentralized, Tor is decentralized, etc. Decentralized protocols and applications have existed before the blockchain and what the blockchain adds as a decentralized application is decentralized transaction validation. Basically decentralizing the validation of what can and cannot get written to a database. If you make your "writes" application state, then you can generalize the validators to be not only validating database-writes, but program execution steps as well (Ethereum). > By no means do I believe that decentralized applications have no benefits. In fact, I foresee a future where applications are 10x more secure, 10x cheaper, 10x more efficient, or 10x more on some dimension than the current ones. Decentralization is always costlier. Netflix is faster and more reliable than torrents, Facebook is more reliable, faster and easier to use than Diaspora. Fiat and credit cards are faster and easier to use than Bitcoin. AWS is faster, cheaper and more reliable than StorJ, and on and on. When you decentralize something, you must add coordination and communication costs such that your overall costs will always be at least as high as the centralized alternative. So the claim that the decentralized applications will magically be faster and cheaper definitely needs research and a [citation needed]. If we realize that the decentralized application will always be costlier than its centralized alternative, we must ask ourselves: why do we need decentralization? Then, we will really know what blockchains are for!

mtgx 9y ago
One of the main characteristics of decentralization, and one that must not be underestimated, is censorship-resistance. How much worse, as in less creative and less free, would the internet be if not just China, but every government knew in the 80's or 90's what they know now about the internet? That decentralization has let a "trillion flowers/websites bloom", and from that extreme experimentation (think of it like life's evolution over billions of years, but condensed in two decades) a lot of great things have resulted, and many more things will come.
snarf21 9y ago
Well said. I hate when every new technology is always a panacea when in reality it is about tradeoffs. It may be much better at one thing but have other costs/issues. The blockchain is interested because it creates distributed trust that enables anonymity not cause it is cheaper.
JohnJamesRambo 9y ago
What are the pros of decentralization? Are they enough to outweigh the cons?
- hudon 9y ago
  The biggest pro of decentralization is censorship resistance. You can compute, store and transfer data and disregard authority or oppressors. This is why torrents worked. They allowed people to share media even though it wasn't legal. Then people found a way to make a more performant centralized alternative and it is so much easier to use that most people now are willing to pay for it, even though free torrents still exist. Email and git are also prone to centralizing forces. They are both decentralized protocols, but everyone ends up using Github and Gmail anyway, and not actually using them in a decentralized way. If you are sending "sensitive" emails or are sharing code for a dark market, email and git will be useful if you steer clear from centralized hubs. Bitcoin is the same thing. Some people don't want you to send money to Wikileaks, so Bitcoin is the first time we can now send money online and disregard who the authorities tell us we can or cannot pay. With Ethereum, decentralizing computation is an open question. Do we not already have censorship resistance in computation? Is anyone preventing us from running code? Is Tor not sufficient to run shady decentralized apps? I don't know.
  - DennisP 9y ago
    We have fully decentralized computation, but only when we don't need consensus. When we do need consensus on our computations, what we've always done before is run the computation on a server (or centrally-managed cluster) that everybody agrees to use. Bitcoin gave us decentralized consensus for simple ledger updates, and Ethereum does it for more general computation and statefulness.
- panabee 9y ago
  innovation is one benefit. as an aside, this is an understated benefit of jeff bezos' API edict. by forcing APIs throughout the company, bezos made it radically simpler for teams to experiment and innovate without permission.
clamprecht 9y ago
Vitalik (of ethereum) has a pretty good blog post talking about the different types of decentralization: architectural, political, and logical.[0] For blockchains, he says: "Blockchains are politically decentralized (no one controls them) and architecturally decentralized (no infrastructural central point of failure) but they are logically centralized (there is one commonly agreed state and the system behaves like a single computer)" [0] https://medium.com/@VitalikButerin/the-meaning-of-decentralization-a0c92b76a274 https://medium.com/@VitalikButerin/the-meaning-of-decentrali...
- kbody 9y ago
  So ironic to read "decentralized (no one controls them)" coming from the leader of ETH that made the decision to fork ETH and revert transactions related to the DAO hack.
  - Lerc 9y ago
    But with decentralized control it's not a decision, it's a proposal. A lot of people feel like Bitcoin etc. are Stick-it-to-the-man endeavors. In the long term I think they will be disappointed. I think the real benefit of this sort of decentralization would be if multiple governments took an active part in its operation. America could propose something but it wouldn't take effect if they can't convince the mining majority of China,Russia,France,Australia,Nigeria and India. No one entity can unilaterally make a decision for their own benefit.
    - 52-6F-62 9y ago
      Well put. I've seen the 'DAO-hard-fork' line swung around pretty wildly over the recent popularity surge the past few months -- but you nailed it. That difference is the whole point. It could be argued that Vitalik has enough political clout in the community now to do nearly whatever he wants, but that only goes so far as whatever he's proposing is agreeable upon a consensus. If the miners disagree, then it's a no go because they make it work. If the larger userbase disagrees, then the miners will be bent to disagree because there won't be any money in it if everyone sells off. It's shortsighted to think that because there are influential parties the whole system is broken. (hint: functioning democracies work this way. As far as most people can tell so far, it's a good thing) I also agree with the rest of your positions here. To wit: while governments are hesitant and slow to get directly involved, they _are_ doing active research[0][1] into how/why/when it will be valuable to use existing blockchains or for what reason they could employ new ones. So are many banks. The distributed ledger seems to hold inherent value for everybody. [0] Being from Canada, I know more about the goings here than elsewhere. The article has a rather negative tone, but the point being they are experimenting and recognize the need to keep up technologically, and the need to improve upon current systems. They are speaking more directly to virtual currencies here. -- https://www.theglobeandmail.com/report-on-business/bank-of-canada-says-wont-use-blockchain-for-interbank-payment-system/article35112169/ https://www.theglobeandmail.com/report-on-business/bank-of-c... [1] Fintech and Financial Services: Initial Considerations - IMF -- http://www.imf.org/~/media/Files/Publications/SDN/2017/sdn1705.ashx http://www.imf.org/~/media/Files/Publications/SDN/2017/sdn17...
abhinavkulkarni 9y ago
> Decentralization is always costlier. Netflix is faster Because Netflix puts their content in the internet exchanges (IXs) (Netflix Open Connect), bypassing CDNs in some cases, thus getting as close to the users as they can given the current internet piping infrastructure [1]. Sure, decentralization (being operated by an aggressive management that can strike background deals with IXs) makes that possible, but we all ought to wonder if it is a good thing in the long run. Netflix, YouTube, etc. have found ways of getting their data faster to the users and the little guy running his podcast stands no chance. [1] https://arxiv.org/pdf/1606.05519.pdf https://arxiv.org/pdf/1606.05519.pdf
CyberDildonics 9y ago
> If we realize that the decentralized application will always be costlier than its centralized alternative, we must ask ourselves: why do we need decentralization? Then, we will really know what blockchains are for! Are you really asking why money should be decentralized? Also do you realize that bitcoin takes about 30 minutes for an irreversible transaction, while credit cards take 30 days before you can spend the money that someone gave you.
- Nursie 9y ago
  For many people (most?) irreversable is a bug, not a feature.
  - CyberDildonics 9y ago
    That depends on which side of the transaction you are on and what you are doing. You are thinking about it purely from a consumer standpoint, where credit cards are built on top of another currency.
    - Nursie 9y ago
      Indeed I am. In the case of a consumer spending money, a method of electronic spending that's not unilaterally reversible is most definitely a bug. For a lot of honest vendors that's also true - they want a situation in which people don't feel constrained from spending because of the high level of risk involved. Reconciliation between finance houses? Not so much, sure.
Matt3o12_ 9y ago
> Decentralization is always costlier. Netflix is faster and more reliable than torrent Is it more expensive, though? In a lot of cases it is certainly more expensive (bitcoin, Facebook, etc) but Netflix should be a counter example. In theory, a decentralized Netflix should be a lot cheaper because there are very little bandwidth costs. In practice, though, this is impossible because of all the DRM regulations. There is just no way one could design a decentralized network without getting into legal troubles. Furthermore, adoption would be another problem because a decentralized Netflix would mean I have to upload content and download it. In practice, this would put me a long time into jail if I did that in my country (sometimes even more then murderer and rapists). We have only seen one such successful "decentralized" application so far: popcorn time. It proves that it can be more reliable and easier to use but in reality the legal implications are just too impractical. Furthermore, decentralizing other video platforms such as YouTube could be a lot harder. Netflix has a small amount of viewing content but a huge amount of viewers. YouTube has a lot of content and less viewers per minute of content then Netflix has. This means any user would have to store much more then they would on Netflix. There is just no way to save content for a long time. On YouTube I recently found a video I did when I was 13 about a tutorial for a video game. If I stored it on a hard drive, I would have lost it for sure (because I never really cared about it much) and there is no way it would have been stored for some many years by just 3000 viewers who have ever seen that. Just seeing this video after all those years is kind of amazing. I know that there is no guarantee that this video will stay on YouTube forever but the chances are a lot higher (and I don't see it disappearing as long as YouTube stays popular and their main business model is make small video makers big)
- hudon 9y ago
  > [popcorn time] proves that it can be more reliable and easier I don't think this is true... My friend used popcorn time and I can tell you that it is not easier or more reliable. The movie comes in somewhat random chunks so most of the time you are stuck having to download a large portion of the movie before being able to view it. The content discovery is lacking too. You have to rely on people hosting old movies and most of the time only blockbusters or recent movies are available. If you do get lucky on more obscure films, downloading it becomes an issue because of lack of seeders, and seeders come and go. It is close to Netflix quality, but still not quite, which moreso is evidence for my "will always be at least as costly" point. Pick a not-super-recent and average-popularity film, say V for Vendetta, and see how Netflix and Popcorn Time compare. You can also think about the tech: one ensures it's hosting high quality content on reliable servers that have the highest possible connectivity to its users. The other relies on users' good-will and connectivity, and both of those come and go.
Geee 9y ago
You forgot that blockchain enables capitalism and free markets on decentralized applications. Voluntarist models such as BitTorrent don't quite work, as we know (BitTorrent only works for popular files). This is a huge paradigm shift, and it will enable revival of all kinds of P2P applications, and actually make them viable and able to compete with centralized solutions. There's Sia for decentralized storage (StorJ is not decentralized) and Mysterium for decentrarized VPN, etc.
- hudon 9y ago
  Sia is going to have a hard time defending the content that their system will end up hosting... But ethics aside, if the tech works, why can't the nodes be paid in Bitcoin? If payments are the only piece missing in making this work, as you say, why does their computation need to run on the blockchain?
  - Geee 9y ago
    Sia could probably run on Bitcoin or Ethereum, but that's not relevant to the question. Blockchain smart contracts enable file contracts and other features that make Sia work. The smart contracts determine how hosts are paid, what happens if they lose files etc. Everything just works without any central authority or payment processor because of the blockchain. https://support.sia.tech/knowledge_base/topics/why-do-you-have-your-own-coin-and-blockchain https://support.sia.tech/knowledge_base/topics/why-do-you-ha... To the first question: "Sia" doesn't have to defend anything, because Sia is software, and not a legal entity. The developers have no control, authority or ownership of the network, and thus are not legally responsible for the contents. Individual hosts are responsible for the content that they host, and will have to respond to legal queries and remove illegal content.
wslh 9y ago
I see the article full of text just as a SEO tactic. There is no real substance there: > Problem #1: Lack of developer friendliness As we described above, Ethereum solves this problem by design through its expressive programming language and strong developer tooling. The author doesn't have direct experience with Solidity, where the tooling is buggy and there are many subtle ambiguities and security issues that are not yet addressed. Any ICO, smart contract developer, or security auditor knows this and this is why ERC20 tokens that are only a few lines of code and very repetitive can hide terrible issues. You can check for example: "Smart contract short address attack mitigation failure" https://blog.coinfabrik.com/smart-contract-short-address-attack-mitigation-failure/ https://blog.coinfabrik.com/smart-contract-short-address-att...

by hudon 9y ago

mtgx 9y ago

One of the main characteristics of decentralization, and one that must not be underestimated, is censorship-resistance. How much worse, as in less creative and less free, would the internet be if not just China, but every government knew in the 80's or 90's what they know now about the internet? That decentralization has let a "trillion flowers/websites bloom", and from that extreme experimentation (think of it like life's evolution over billions of years, but condensed in two decades) a lot of great things have resulted, and many more things will come.

snarf21 9y ago

Well said. I hate when every new technology is always a panacea when in reality it is about tradeoffs. It may be much better at one thing but have other costs/issues. The blockchain is interested because it creates distributed trust that enables anonymity not cause it is cheaper.

JohnJamesRambo 9y ago

What are the pros of decentralization? Are they enough to outweigh the cons?

hudon 9y ago
The biggest pro of decentralization is censorship resistance. You can compute, store and transfer data and disregard authority or oppressors. This is why torrents worked. They allowed people to share media even though it wasn't legal. Then people found a way to make a more performant centralized alternative and it is so much easier to use that most people now are willing to pay for it, even though free torrents still exist. Email and git are also prone to centralizing forces. They are both decentralized protocols, but everyone ends up using Github and Gmail anyway, and not actually using them in a decentralized way. If you are sending "sensitive" emails or are sharing code for a dark market, email and git will be useful if you steer clear from centralized hubs. Bitcoin is the same thing. Some people don't want you to send money to Wikileaks, so Bitcoin is the first time we can now send money online and disregard who the authorities tell us we can or cannot pay. With Ethereum, decentralizing computation is an open question. Do we not already have censorship resistance in computation? Is anyone preventing us from running code? Is Tor not sufficient to run shady decentralized apps? I don't know.
- DennisP 9y ago
  We have fully decentralized computation, but only when we don't need consensus. When we do need consensus on our computations, what we've always done before is run the computation on a server (or centrally-managed cluster) that everybody agrees to use. Bitcoin gave us decentralized consensus for simple ledger updates, and Ethereum does it for more general computation and statefulness.
panabee 9y ago
innovation is one benefit. as an aside, this is an understated benefit of jeff bezos' API edict. by forcing APIs throughout the company, bezos made it radically simpler for teams to experiment and innovate without permission.

clamprecht 9y ago

Vitalik (of ethereum) has a pretty good blog post talking about the different types of decentralization: architectural, political, and logical.[0] For blockchains, he says: "Blockchains are politically decentralized (no one controls them) and architecturally decentralized (no infrastructural central point of failure) but they are logically centralized (there is one commonly agreed state and the system behaves like a single computer)" [0] https://medium.com/@VitalikButerin/the-meaning-of-decentralization-a0c92b76a274 https://medium.com/@VitalikButerin/the-meaning-of-decentrali...

kbody 9y ago
So ironic to read "decentralized (no one controls them)" coming from the leader of ETH that made the decision to fork ETH and revert transactions related to the DAO hack.
- Lerc 9y ago
  But with decentralized control it's not a decision, it's a proposal. A lot of people feel like Bitcoin etc. are Stick-it-to-the-man endeavors. In the long term I think they will be disappointed. I think the real benefit of this sort of decentralization would be if multiple governments took an active part in its operation. America could propose something but it wouldn't take effect if they can't convince the mining majority of China,Russia,France,Australia,Nigeria and India. No one entity can unilaterally make a decision for their own benefit.
  - 52-6F-62 9y ago
    Well put. I've seen the 'DAO-hard-fork' line swung around pretty wildly over the recent popularity surge the past few months -- but you nailed it. That difference is the whole point. It could be argued that Vitalik has enough political clout in the community now to do nearly whatever he wants, but that only goes so far as whatever he's proposing is agreeable upon a consensus. If the miners disagree, then it's a no go because they make it work. If the larger userbase disagrees, then the miners will be bent to disagree because there won't be any money in it if everyone sells off. It's shortsighted to think that because there are influential parties the whole system is broken. (hint: functioning democracies work this way. As far as most people can tell so far, it's a good thing) I also agree with the rest of your positions here. To wit: while governments are hesitant and slow to get directly involved, they _are_ doing active research[0][1] into how/why/when it will be valuable to use existing blockchains or for what reason they could employ new ones. So are many banks. The distributed ledger seems to hold inherent value for everybody. [0] Being from Canada, I know more about the goings here than elsewhere. The article has a rather negative tone, but the point being they are experimenting and recognize the need to keep up technologically, and the need to improve upon current systems. They are speaking more directly to virtual currencies here. -- https://www.theglobeandmail.com/report-on-business/bank-of-canada-says-wont-use-blockchain-for-interbank-payment-system/article35112169/ https://www.theglobeandmail.com/report-on-business/bank-of-c... [1] Fintech and Financial Services: Initial Considerations - IMF -- http://www.imf.org/~/media/Files/Publications/SDN/2017/sdn1705.ashx http://www.imf.org/~/media/Files/Publications/SDN/2017/sdn17...

abhinavkulkarni 9y ago

> Decentralization is always costlier. Netflix is faster Because Netflix puts their content in the internet exchanges (IXs) (Netflix Open Connect), bypassing CDNs in some cases, thus getting as close to the users as they can given the current internet piping infrastructure [1]. Sure, decentralization (being operated by an aggressive management that can strike background deals with IXs) makes that possible, but we all ought to wonder if it is a good thing in the long run. Netflix, YouTube, etc. have found ways of getting their data faster to the users and the little guy running his podcast stands no chance. [1] https://arxiv.org/pdf/1606.05519.pdf https://arxiv.org/pdf/1606.05519.pdf

CyberDildonics 9y ago

> If we realize that the decentralized application will always be costlier than its centralized alternative, we must ask ourselves: why do we need decentralization? Then, we will really know what blockchains are for! Are you really asking why money should be decentralized? Also do you realize that bitcoin takes about 30 minutes for an irreversible transaction, while credit cards take 30 days before you can spend the money that someone gave you.

Nursie 9y ago
For many people (most?) irreversable is a bug, not a feature.
- CyberDildonics 9y ago
  That depends on which side of the transaction you are on and what you are doing. You are thinking about it purely from a consumer standpoint, where credit cards are built on top of another currency.
  - Nursie 9y ago
    Indeed I am. In the case of a consumer spending money, a method of electronic spending that's not unilaterally reversible is most definitely a bug. For a lot of honest vendors that's also true - they want a situation in which people don't feel constrained from spending because of the high level of risk involved. Reconciliation between finance houses? Not so much, sure.

Matt3o12_ 9y ago

> Decentralization is always costlier. Netflix is faster and more reliable than torrent Is it more expensive, though? In a lot of cases it is certainly more expensive (bitcoin, Facebook, etc) but Netflix should be a counter example. In theory, a decentralized Netflix should be a lot cheaper because there are very little bandwidth costs. In practice, though, this is impossible because of all the DRM regulations. There is just no way one could design a decentralized network without getting into legal troubles. Furthermore, adoption would be another problem because a decentralized Netflix would mean I have to upload content and download it. In practice, this would put me a long time into jail if I did that in my country (sometimes even more then murderer and rapists). We have only seen one such successful "decentralized" application so far: popcorn time. It proves that it can be more reliable and easier to use but in reality the legal implications are just too impractical. Furthermore, decentralizing other video platforms such as YouTube could be a lot harder. Netflix has a small amount of viewing content but a huge amount of viewers. YouTube has a lot of content and less viewers per minute of content then Netflix has. This means any user would have to store much more then they would on Netflix. There is just no way to save content for a long time. On YouTube I recently found a video I did when I was 13 about a tutorial for a video game. If I stored it on a hard drive, I would have lost it for sure (because I never really cared about it much) and there is no way it would have been stored for some many years by just 3000 viewers who have ever seen that. Just seeing this video after all those years is kind of amazing. I know that there is no guarantee that this video will stay on YouTube forever but the chances are a lot higher (and I don't see it disappearing as long as YouTube stays popular and their main business model is make small video makers big)

hudon 9y ago
> [popcorn time] proves that it can be more reliable and easier I don't think this is true... My friend used popcorn time and I can tell you that it is not easier or more reliable. The movie comes in somewhat random chunks so most of the time you are stuck having to download a large portion of the movie before being able to view it. The content discovery is lacking too. You have to rely on people hosting old movies and most of the time only blockbusters or recent movies are available. If you do get lucky on more obscure films, downloading it becomes an issue because of lack of seeders, and seeders come and go. It is close to Netflix quality, but still not quite, which moreso is evidence for my "will always be at least as costly" point. Pick a not-super-recent and average-popularity film, say V for Vendetta, and see how Netflix and Popcorn Time compare. You can also think about the tech: one ensures it's hosting high quality content on reliable servers that have the highest possible connectivity to its users. The other relies on users' good-will and connectivity, and both of those come and go.

Geee 9y ago

You forgot that blockchain enables capitalism and free markets on decentralized applications. Voluntarist models such as BitTorrent don't quite work, as we know (BitTorrent only works for popular files). This is a huge paradigm shift, and it will enable revival of all kinds of P2P applications, and actually make them viable and able to compete with centralized solutions. There's Sia for decentralized storage (StorJ is not decentralized) and Mysterium for decentrarized VPN, etc.

hudon 9y ago
Sia is going to have a hard time defending the content that their system will end up hosting... But ethics aside, if the tech works, why can't the nodes be paid in Bitcoin? If payments are the only piece missing in making this work, as you say, why does their computation need to run on the blockchain?
- Geee 9y ago
  Sia could probably run on Bitcoin or Ethereum, but that's not relevant to the question. Blockchain smart contracts enable file contracts and other features that make Sia work. The smart contracts determine how hosts are paid, what happens if they lose files etc. Everything just works without any central authority or payment processor because of the blockchain. https://support.sia.tech/knowledge_base/topics/why-do-you-have-your-own-coin-and-blockchain https://support.sia.tech/knowledge_base/topics/why-do-you-ha... To the first question: "Sia" doesn't have to defend anything, because Sia is software, and not a legal entity. The developers have no control, authority or ownership of the network, and thus are not legally responsible for the contents. Individual hosts are responsible for the content that they host, and will have to respond to legal queries and remove illegal content.

wslh 9y ago

I see the article full of text just as a SEO tactic. There is no real substance there: > Problem #1: Lack of developer friendliness As we described above, Ethereum solves this problem by design through its expressive programming language and strong developer tooling. The author doesn't have direct experience with Solidity, where the tooling is buggy and there are many subtle ambiguities and security issues that are not yet addressed. Any ICO, smart contract developer, or security auditor knows this and this is why ERC20 tokens that are only a few lines of code and very repetitive can hide terrible issues. You can check for example: "Smart contract short address attack mitigation failure" https://blog.coinfabrik.com/smart-contract-short-address-attack-mitigation-failure/ https://blog.coinfabrik.com/smart-contract-short-address-att...