狠狠撸

狠狠撸Share a Scribd company logo

Decentralization for public blockchains

?
?
Igor Artamonov
Igor Artamonov

Decentralization for public blockchains, Why it’s critical for blockchain adoption Keynote talk at Financial Cryptography and Data Security 2019 conference, February 18–22, 2019, St. Kitts Following articles: - https://medium.com/@splix/on-decentralization-part-i-49562b817cd2 - https://medium.com/@splix/on-decentralization-part-ii-fb01c0ac0bb9

1 of 63
Download to read offline
Decentralization for public blockchains Igor Artamonov FC’19 / Financial Cryptography and Data Security 2019 February 18–22, 2019 St. Kitts Why it’s critical for blockchain adoption
About Author Was a lead of Ethereum Classic development since the fork and made Ethereum Classic happen Currently doing tech consulting and working on Emerald Wallet
Ethereum Classic Development ? I was one of the ?rst few people who joined Ethereum Classic, before The DAO fork of ETH ? I believe in blockchain principles such as immutability and decentralization ? Found a ballot fork and transaction reversal in interest of core team as unacceptable act of ruining blockchain ideas ? To show my position I started to contribute to the original code, launched a block explorer, eventually build a team of software engineers to work full time on Ethereum Classic core projects
Since December 2018 I’m not working on Ethereum Classic protocol or core projects, i.e. not engaged in Ethereum Classic All I say is my personal opinion
Blockchain
A globally distributed Database or Computer without a trusted 3rd party
Is it effective?
? Mastercard/Visa can process 5,000-50,000 operations per second ? Bitcoin is limited to 3-7 operations per second ? Bitcoin is less e?ective 1,500 times ? Same comparison of Ethereum to AWS gives 1,000,000 times less ? Or, in other words, it cost x1,000,000 times more for the same processing
Why are we paying this price?
We are paying for avoiding a central point For decentralization of our database, computation, payments, agreements, etc
Let’s talk about decentralization
? First of all, blockchain as a physical network seems to be enough decentralized ? But blockchain is more than a peer-to-peer network, it’s a whole infrastructure behind it ? This infrastructure has many ways to a?ect network participants ? Some parts of the infrastructure are centralized or tend to centralization
First check
Which one is centralized, which is decentralized? ? Bitcoin ? Ethereum ? EOS ? Ripple
Decentralized → We will probably all agree that Bitcoin is most decentralized and Ripple is least decentralized, in this set of blockchains
Factors that affect decentralization ? Premine ? Foundation ? O?cial Team ? Leader ? Development tax ? O?cial code ? Whales ? Mining pools ? Nodes concentration Internal External
Premine/Presale Not a big problem, unless it’s big enough to have political weight
Foundation ? Everyone expects the foundation will do development, sponsor research, make decisions, etc. Everything became dependent on foundation ? It concentrates power and decides for everyone what is right path, what is not. ? It decides which fork, in case of community split, is right one. See ETC/ETH split
Leader ? Leader is by de?nition is a power and also a weak point of failure ? Makes even more problem than foundation, because nobody argue with a leader, it’s a disrespect ? Satoshi is the perfect leader btw, he just disappeared at the right time
Core Team/Code Below is the code from Parity Ethereum uno?cial client. At this particular case the code was forced to comply with the code written for an o?cial client, though if wasn’t justi?ed by a speci?cation
Whales ? Just price manipulation is toy wars, but whales can also drastically change development path ? See BCH/BTC/BSV. Groups of whales had disagreement with “core team” and made a fork, with enough capital it became more or less successful
Mining Pools ? 51% attack. ETC is a an example where it happened in practice ? At some point pools will realize that they can make money on providing a service to penalize speci?c transactions ? Pools can also sabotage or spam network, and charge for that as well ? We saw examples when it has political weight in BCH/BTC split
Nodes concentration ? In an ideal world every user will have own full node ? Unfortunately it doesn’t work yet because a full node requires advanced knowledge and is a resource intensive software ? Even businesses depend on central providers like Infura ? Central provider decides which fork is right, which transaction is acceptable, and so on
Every blockchain seems to be centralized
If we’ll come to decentralization metric from 0 to 1, where 1 is fully decentralized blockchains, I believe than average value for the top 10 blockchains will be even less than 0.5, likely less than 0.25
Bitcoin ? Premine - kind of ? Foundation - no ? Team - not very centralized, but people think it has some issues with centralization ? Leader - no ? Development tax - no ? Core code - yes ? Whales - yes, not a big control, but we saw them in BCH/BTC split ? Mining pools - not so much centralized (4 pools ~ 50%) ? Nodes centralization - no
Ethereum ? Premine - yes, 70% of current supply ? Foundation - yes ? Team - formally no, but in fact one group ? Leader - yes ? Development tax - no ? Core code - yes, though they are trying to change it ? Whales - yes ? Mining pools - more or less centralized (3 pools > 50%) ? Nodes centralization - yes
Maybe centralization is good?
Jan 1 - unnatural spike Jan 15 - bug was found, network downgraded, it’s a coordination to avoid failure Network preparations for Ethereum Constantinople hard fork.
Jan 1st: - Parity (uno?cial client) had natural growth - Geth (o?cial client) jumped It seems that an external force pushed upgrade rate, and most interesting is that it has a?ected only o?cial nodes. Thousand of nodes in one day
Maybe something like this forced most of the nodes to upgrade in that day: (That was a fragment from exchanges chat, when ETH found the hack of The DAO contract)
Centralized solution is usually more e?ective. Most of the current blockchains are centralized and are ?ne with that. So what is the problem with centralization?
Central Point is a Point of Failure But what kind of failure?
? Any central point can be used to get some advantage, it’s a power, especially in a context of money ? Control of a public blockchain is a power which governments, big corporations and criminals want to control ? Humans are weak, they are especially exposed if they are part of that central point
Decentralization for public blockchains
Decentralization for public blockchains
? Most people think it’s impossible to force any changes, “because Open Source” ? Unfortunately not every problem easy to notice. Otherwise we wouldn’t have software bugs ? Some backdoors can intentionally planted in a code and pass all veri?cations, only authors would know how to use them. ? There’re many examples
NSA BULLRUN ? Information about the program's existence was leaked in 2013 by Edward Snowden ? NSA has been actively working on inserting vulnerabilities into commercial encryption systems. ? One of planted vulnerabilities was a backdoor added to random number generator Dual_EC_DRBG
Juniper Backdoor ? Juniper replaces secure ANSI X9.31 to less secure Dual_EC ? And changed other parts of software at the same time, like [seems to be intentionally] added some bugs in di?erent places ? Altogether it allowed to decrypt and listen to tra?c
BEA-1 ? Backdoored Encryption Algorithm, version 1. Paper “Proposal for a Backdoored AES-like Block Cipher”, Arnaud Bannier and Eric Filiol, 2017 ? Compliant with FIPS-140 requirements (US NIST standard for crypto) and resist to linear/di?erential attacks ? All looked good, but because of a hidden backdoor, it can be broken on a laptop
Is blockchain affected? ? Many people already don’t trust ZCash because of “Trusted Setup”, people think that someone has a “master key”, and it’s hard to prove opposite ? Power was already misused, we had transaction reversal, arti?cial ine?ciency in the code to keep control over community, economic changes forced by power ? Many blockchain projects has violated a lot of laws, SEC rules, gambling laws, and so on. A government may prosecute founders or force them to cooperate, we wouldn’t know if latter happened already
Decentralization for public blockchains
F15 Categorization
Decentralization for public blockchains
? “Every disagreement can be solved by using pure power” ? If a problem is big enough, the ultimate power is ?ghter jets, who has more of them is right ? Fighter jet can stop anything. Except blockchain, you can’t stop it. Or can you?
The criteria: “How many F15 you need to force changes”
Let’s check blockchains again
Decentralization for public blockchains
? F15 are used to symbolize amount of external power that can a?ect blockchain infrastructure and decision in some way ? It’s a subjective categorization, not real ?ghter jets Most protected from an enemy power Least protected from enemy power …
? Bitcoin ? has a stable protocol, less depends on core progress currently ? suspicious community and opposing to changes, ask tough questions ? too many di?erent forces/groups makes is ine?ective to target one ? Monero ? future still depends on core dev team ? though it’s pseudonymous and distributed ? ?nancing comes from di?erent sources ? hard to attack (but much easier than bitcoin)
? Grin ? interesting new project, right ?rst steps ? vulnerable only because it’s young and small ? Dogecoin ? there is literally no one in charge! much unstoppable
? Ethereum ? known leaders and ?nancing ? code needs a lot of changes before maturity ? many central points/many vulnerabilities ? Ethereum Classic ? unfortunately now is just one coordinated group which controls everything, from code to ?nances, media, community, etc ? ZCash ? control is too concentrated as well, tech is sophisticated for a broad community to be involved
For the most of other blockchains you don’t need any power at all, you just call CEO and make an agreement. Zero F15.
“But we’re using blockchain just for some basic and legal stuff, who cares?”
Internet was launched and designed to survive nuclear war. Literally. I want to remind that
? Internet was made by DARPA - Defense Advanced Research Projects Agency or US Department of Defense ? TCP/IP is also known as “DoD Four-Layer Model”, where DoD is Department of Defense
? Memorandum on Distributed Communications by Paul Baran from RAND Corp ? Research a network that can survive an enemy attack ? Proposed packet-switching, and other things that became a basis of modern internet
Decentralization for public blockchains
? Initial funding for Tor's development has come from the federal government of the United States, initially through the O?ce of Naval Research and DARPA ? “After analyzing documents leaked by Edward Snowden, The Guardian reported that the NSA had repeatedly tried to crack Tor and had failed to break its core security”
? Most of modern internet was build by defense organizations, designed for extreme conditions like surviving a nuclear war ? But is being used to post photos on Facebook ? Such extreme criteria allowed to build a network that can be an universal communication layer for everything
Decentralization for public blockchains
A public blockchain cannot be controlled by a single party (“country”), because there’re few of them who wants power A global public communication network can succeed only if it controlled by no one (i.e. “decentralized” and F15-survivable)
Igor Artamonov igor@artamonov.ru @splix Thank you!

More Related Content

Decentralization for public blockchains

  • 1. Decentralization for public blockchains Igor Artamonov FC’19 / Financial Cryptography and Data Security 2019 February 18–22, 2019 St. Kitts Why it’s critical for blockchain adoption
  • 2. About Author Was a lead of Ethereum Classic development since the fork and made Ethereum Classic happen Currently doing tech consulting and working on Emerald Wallet
  • 3. Ethereum Classic Development ? I was one of the ?rst few people who joined Ethereum Classic, before The DAO fork of ETH ? I believe in blockchain principles such as immutability and decentralization ? Found a ballot fork and transaction reversal in interest of core team as unacceptable act of ruining blockchain ideas ? To show my position I started to contribute to the original code, launched a block explorer, eventually build a team of software engineers to work full time on Ethereum Classic core projects
  • 4. Since December 2018 I’m not working on Ethereum Classic protocol or core projects, i.e. not engaged in Ethereum Classic All I say is my personal opinion
  • 6. A globally distributed Database or Computer without a trusted 3rd party
  • 8. ? Mastercard/Visa can process 5,000-50,000 operations per second ? Bitcoin is limited to 3-7 operations per second ? Bitcoin is less e?ective 1,500 times ? Same comparison of Ethereum to AWS gives 1,000,000 times less ? Or, in other words, it cost x1,000,000 times more for the same processing
  • 9. Why are we paying this price?
  • 10. We are paying for avoiding a central point For decentralization of our database, computation, payments, agreements, etc
  • 12. ? First of all, blockchain as a physical network seems to be enough decentralized ? But blockchain is more than a peer-to-peer network, it’s a whole infrastructure behind it ? This infrastructure has many ways to a?ect network participants ? Some parts of the infrastructure are centralized or tend to centralization
  • 14. Which one is centralized, which is decentralized? ? Bitcoin ? Ethereum ? EOS ? Ripple
  • 15. Decentralized → We will probably all agree that Bitcoin is most decentralized and Ripple is least decentralized, in this set of blockchains
  • 16. Factors that affect decentralization ? Premine ? Foundation ? O?cial Team ? Leader ? Development tax ? O?cial code ? Whales ? Mining pools ? Nodes concentration Internal External
  • 17. Premine/Presale Not a big problem, unless it’s big enough to have political weight
  • 18. Foundation ? Everyone expects the foundation will do development, sponsor research, make decisions, etc. Everything became dependent on foundation ? It concentrates power and decides for everyone what is right path, what is not. ? It decides which fork, in case of community split, is right one. See ETC/ETH split
  • 19. Leader ? Leader is by de?nition is a power and also a weak point of failure ? Makes even more problem than foundation, because nobody argue with a leader, it’s a disrespect ? Satoshi is the perfect leader btw, he just disappeared at the right time
  • 20. Core Team/Code Below is the code from Parity Ethereum uno?cial client. At this particular case the code was forced to comply with the code written for an o?cial client, though if wasn’t justi?ed by a speci?cation
  • 21. Whales ? Just price manipulation is toy wars, but whales can also drastically change development path ? See BCH/BTC/BSV. Groups of whales had disagreement with “core team” and made a fork, with enough capital it became more or less successful
  • 22. Mining Pools ? 51% attack. ETC is a an example where it happened in practice ? At some point pools will realize that they can make money on providing a service to penalize speci?c transactions ? Pools can also sabotage or spam network, and charge for that as well ? We saw examples when it has political weight in BCH/BTC split
  • 23. Nodes concentration ? In an ideal world every user will have own full node ? Unfortunately it doesn’t work yet because a full node requires advanced knowledge and is a resource intensive software ? Even businesses depend on central providers like Infura ? Central provider decides which fork is right, which transaction is acceptable, and so on
  • 24. Every blockchain seems to be centralized
  • 25. If we’ll come to decentralization metric from 0 to 1, where 1 is fully decentralized blockchains, I believe than average value for the top 10 blockchains will be even less than 0.5, likely less than 0.25
  • 26. Bitcoin ? Premine - kind of ? Foundation - no ? Team - not very centralized, but people think it has some issues with centralization ? Leader - no ? Development tax - no ? Core code - yes ? Whales - yes, not a big control, but we saw them in BCH/BTC split ? Mining pools - not so much centralized (4 pools ~ 50%) ? Nodes centralization - no
  • 27. Ethereum ? Premine - yes, 70% of current supply ? Foundation - yes ? Team - formally no, but in fact one group ? Leader - yes ? Development tax - no ? Core code - yes, though they are trying to change it ? Whales - yes ? Mining pools - more or less centralized (3 pools > 50%) ? Nodes centralization - yes
  • 29. Jan 1 - unnatural spike Jan 15 - bug was found, network downgraded, it’s a coordination to avoid failure Network preparations for Ethereum Constantinople hard fork.
  • 30. Jan 1st: - Parity (uno?cial client) had natural growth - Geth (o?cial client) jumped It seems that an external force pushed upgrade rate, and most interesting is that it has a?ected only o?cial nodes. Thousand of nodes in one day
  • 31. Maybe something like this forced most of the nodes to upgrade in that day: (That was a fragment from exchanges chat, when ETH found the hack of The DAO contract)
  • 32. Centralized solution is usually more e?ective. Most of the current blockchains are centralized and are ?ne with that. So what is the problem with centralization?
  • 33. Central Point is a Point of Failure But what kind of failure?
  • 34. ? Any central point can be used to get some advantage, it’s a power, especially in a context of money ? Control of a public blockchain is a power which governments, big corporations and criminals want to control ? Humans are weak, they are especially exposed if they are part of that central point
  • 37. ? Most people think it’s impossible to force any changes, “because Open Source” ? Unfortunately not every problem easy to notice. Otherwise we wouldn’t have software bugs ? Some backdoors can intentionally planted in a code and pass all veri?cations, only authors would know how to use them. ? There’re many examples
  • 38. NSA BULLRUN ? Information about the program's existence was leaked in 2013 by Edward Snowden ? NSA has been actively working on inserting vulnerabilities into commercial encryption systems. ? One of planted vulnerabilities was a backdoor added to random number generator Dual_EC_DRBG
  • 39. Juniper Backdoor ? Juniper replaces secure ANSI X9.31 to less secure Dual_EC ? And changed other parts of software at the same time, like [seems to be intentionally] added some bugs in di?erent places ? Altogether it allowed to decrypt and listen to tra?c
  • 40. BEA-1 ? Backdoored Encryption Algorithm, version 1. Paper “Proposal for a Backdoored AES-like Block Cipher”, Arnaud Bannier and Eric Filiol, 2017 ? Compliant with FIPS-140 requirements (US NIST standard for crypto) and resist to linear/di?erential attacks ? All looked good, but because of a hidden backdoor, it can be broken on a laptop
  • 41. Is blockchain affected? ? Many people already don’t trust ZCash because of “Trusted Setup”, people think that someone has a “master key”, and it’s hard to prove opposite ? Power was already misused, we had transaction reversal, arti?cial ine?ciency in the code to keep control over community, economic changes forced by power ? Many blockchain projects has violated a lot of laws, SEC rules, gambling laws, and so on. A government may prosecute founders or force them to cooperate, we wouldn’t know if latter happened already
  • 45. ? “Every disagreement can be solved by using pure power” ? If a problem is big enough, the ultimate power is ?ghter jets, who has more of them is right ? Fighter jet can stop anything. Except blockchain, you can’t stop it. Or can you?
  • 46. The criteria: “How many F15 you need to force changes”
  • 49. ? F15 are used to symbolize amount of external power that can a?ect blockchain infrastructure and decision in some way ? It’s a subjective categorization, not real ?ghter jets Most protected from an enemy power Least protected from enemy power …
  • 50. ? Bitcoin ? has a stable protocol, less depends on core progress currently ? suspicious community and opposing to changes, ask tough questions ? too many di?erent forces/groups makes is ine?ective to target one ? Monero ? future still depends on core dev team ? though it’s pseudonymous and distributed ? ?nancing comes from di?erent sources ? hard to attack (but much easier than bitcoin)
  • 51. ? Grin ? interesting new project, right ?rst steps ? vulnerable only because it’s young and small ? Dogecoin ? there is literally no one in charge! much unstoppable
  • 52. ? Ethereum ? known leaders and ?nancing ? code needs a lot of changes before maturity ? many central points/many vulnerabilities ? Ethereum Classic ? unfortunately now is just one coordinated group which controls everything, from code to ?nances, media, community, etc ? ZCash ? control is too concentrated as well, tech is sophisticated for a broad community to be involved
  • 53. For the most of other blockchains you don’t need any power at all, you just call CEO and make an agreement. Zero F15.
  • 54. “But we’re using blockchain just for some basic and legal stuff, who cares?”
  • 55. Internet was launched and designed to survive nuclear war. Literally. I want to remind that
  • 56. ? Internet was made by DARPA - Defense Advanced Research Projects Agency or US Department of Defense ? TCP/IP is also known as “DoD Four-Layer Model”, where DoD is Department of Defense
  • 57. ? Memorandum on Distributed Communications by Paul Baran from RAND Corp ? Research a network that can survive an enemy attack ? Proposed packet-switching, and other things that became a basis of modern internet
  • 59. ? Initial funding for Tor's development has come from the federal government of the United States, initially through the O?ce of Naval Research and DARPA ? “After analyzing documents leaked by Edward Snowden, The Guardian reported that the NSA had repeatedly tried to crack Tor and had failed to break its core security”
  • 60. ? Most of modern internet was build by defense organizations, designed for extreme conditions like surviving a nuclear war ? But is being used to post photos on Facebook ? Such extreme criteria allowed to build a network that can be an universal communication layer for everything
  • 62. A public blockchain cannot be controlled by a single party (“country”), because there’re few of them who wants power A global public communication network can succeed only if it controlled by no one (i.e. “decentralized” and F15-survivable)