4 ms·

from a guy that many years ago said that the more code a consensus algorithm has - the worse it is. and we've only seen more and more code, turning the knobs an

from a guy that many years ago said that the more code a consensus algorithm has - the worse it is. and we've only seen more and more code, turning the knobs and coming up with workarounds, year after year.

serverholic 5y ago
Does this actually have anything to do with consensus? From what I understand the core consensus algorithm has stayed the same for awhile.
- keymone 5y ago
  seems like it does, because it changes block validation function.
rfd4sgmk8u 5y ago
Agreed. And it just gets worse. The latest generation of 'VC chains' are more complex again. One popular one even failed to produce blocks for 17 hours. Even when >50% of the miners were kicked out of Chinam Bitcoin didn't fail. It took a few weeks of slower blocks until the consensus agreed on a lesser difficulty. Complexity is the enemy of security. These systems just grow more complex and fragile as they progress.
- vmception 5y ago
  They just grow faster and get patched quickly. Bitcoin has had brief and temporary service failures a variety of times early on. A brief issue during a chain's first year of operation really isn't as damning as you think it is. There is a lot of immediate commerce right now which is awkward to disrupt, but just patch and move on.
  - keymone 5y ago
    exactly what i need from the foundation of a financial system, to grow quickly and patch bugs on the fly! what could possibly go wrong.
    - vmception 5y ago
      remember to ask: are you making a separate higher standard than the one you use with the system you actually respect? useful to mention because I notice people often aren't aware of how the system they respect works and succumbs worse to the same criticisms
      - keymone 5y ago
        nope, same standard as i have for bitcoin. consensus algorithm barely changed in 12 years. the only known consensus-breaking bug that caused a chain rollback was found long before bitcoin had any significant financial relevance in the world. developers have extreme propensity for slow, gradual, cautious changes, especially when it comes to consensus algorithm, realizing that "moving fast and breaking things" is not a great basis for stable system and a horrible precedent for future development/governance.
        vmception 5y ago
        Well the person above was originally talking about Solana And I thought I was responding to a no-coiner, but it seems you are conflating everything not-bitcoin together Blocks stopped briefly and it was inconvenient because there is a bit of commerce since the adoption curve is way faster, we can just say “thank you bitcoin for paving the way” and move on. That circumstance is just not different enough.
        marcoslozada 5y ago
        Things have to be done quite well on this area, little error margin
everfree 5y ago
Ethereum's rollup-centric roadmap explicitly tries to push code out of the consensus layer into the smart contract layer, in contrast with other chains that try to make the chain's entire feature set monolithic in a single protocol spec. This calldata change is a very small consensus code change in comparison to the bigger picture here. And yes, they have been turning the knobs year after year, but turning knobs is a lot different than adding more and more code.

serverholic 5y ago

Does this actually have anything to do with consensus? From what I understand the core consensus algorithm has stayed the same for awhile.

keymone 5y ago
seems like it does, because it changes block validation function.

rfd4sgmk8u 5y ago

Agreed. And it just gets worse. The latest generation of 'VC chains' are more complex again. One popular one even failed to produce blocks for 17 hours. Even when >50% of the miners were kicked out of Chinam Bitcoin didn't fail. It took a few weeks of slower blocks until the consensus agreed on a lesser difficulty. Complexity is the enemy of security. These systems just grow more complex and fragile as they progress.

vmception 5y ago
They just grow faster and get patched quickly. Bitcoin has had brief and temporary service failures a variety of times early on. A brief issue during a chain's first year of operation really isn't as damning as you think it is. There is a lot of immediate commerce right now which is awkward to disrupt, but just patch and move on.
- keymone 5y ago
  exactly what i need from the foundation of a financial system, to grow quickly and patch bugs on the fly! what could possibly go wrong.
  - vmception 5y ago
    remember to ask: are you making a separate higher standard than the one you use with the system you actually respect? useful to mention because I notice people often aren't aware of how the system they respect works and succumbs worse to the same criticisms
    - keymone 5y ago
      nope, same standard as i have for bitcoin. consensus algorithm barely changed in 12 years. the only known consensus-breaking bug that caused a chain rollback was found long before bitcoin had any significant financial relevance in the world. developers have extreme propensity for slow, gradual, cautious changes, especially when it comes to consensus algorithm, realizing that "moving fast and breaking things" is not a great basis for stable system and a horrible precedent for future development/governance.
      - vmception 5y ago
        Well the person above was originally talking about Solana And I thought I was responding to a no-coiner, but it seems you are conflating everything not-bitcoin together Blocks stopped briefly and it was inconvenient because there is a bit of commerce since the adoption curve is way faster, we can just say “thank you bitcoin for paving the way” and move on. That circumstance is just not different enough.
        marcoslozada 5y ago
        Things have to be done quite well on this area, little error margin

everfree 5y ago

Ethereum's rollup-centric roadmap explicitly tries to push code out of the consensus layer into the smart contract layer, in contrast with other chains that try to make the chain's entire feature set monolithic in a single protocol spec. This calldata change is a very small consensus code change in comparison to the bigger picture here. And yes, they have been turning the knobs year after year, but turning knobs is a lot different than adding more and more code.