GHOST can provide higher transaction throughput by minimized block interval, but community people seldom mention it. What's wrong with GHOST?

  • 1
    Side note if the bitcoin community wouldn't even allow for 8MB blocks or segwit2x I doubt they would even think about GHOST. Nov 13 '17 at 16:53

GHOST is outdated, the more relevant protocol based on recent research by Aviv Zohar et al is called "SPECTRE" - https://medium.com/@avivzohar/the-spectre-protocol-7dbbebb707b5, https://eprint.iacr.org/2016/1159.pdf.

It has many advantages over the traditional longest-chain rule of Bitcoin, but:

  1. It requires a hard fork, something which has never been done intentionally in Bitcoin.

  2. It is a huge departure from Bitcoin's protocol, and as such it will require extensive research and testing before it is even considered for inclusion in Bitcoin.

There are alts under development that are trying to implement SPECTRE, if it works well for them it can be a first step towards consideration in Bitcoin. You don't risk untested ideas on a $100B production system.

  • Thank you! This 2 points are so called "bitcoin conservative" right?
    – Anderson
    Nov 14 '17 at 2:31
  • What alts are trying to implement SPECTRE?
    – Anderson
    Nov 14 '17 at 2:31
  • @Meni, are there any trade-offs?
    – rny
    Nov 14 '17 at 2:58
  • 2
    @AndersonC: You could say that, they are both related to the rigor required to make changes in Bitcoin. I'm not sure about the specific alts, but I'm sure we'll all know when they're ready. Nov 14 '17 at 10:00
  • @renlord - SPECTRE is more computationally intensive than longest chain. Furthermore, the main advantage of SPECTRE is that it allows shorter time between blocks without losing hashpower, but this has other tradeoffs, notably more resource requirements from SPV clients. Other than that, AFAIR SPECTRE is strictly superior, though I may be missing something. Nov 14 '17 at 10:02

One thing to note also is that even ethereum is not using GHOST and I dont see any coin using some sort of ghost or phantom protocol at this point, so yeah its hard to expect bitcoin will jump to it before it is tested elsewhere.



We are using a modified version of Ghost at Hycon. The original plan was to move forward with Spectre, but as yet there are some unsolved issues, particularly around selfish mining and efficient resolution of superposition states at the tips of the DAG. Ghost is useful when the block time is shorter and orphan events are more common. In the case of bitcoin, these days orphan events are rare, so unless it was decided to reduce the block time it is unnecessary.

  • 1
    ghost also massively amplifies selfish mining... my understanding is that this is why ethereum dropped it. Impressive that there are other altcoins more recklessly yolo than ethereum...
    – G. Maxwell
    Feb 12 '19 at 2:36
  • 1
    Our implementation includes the difficulty from uncle blocks as part of our dynamic difficulty adjustment, negating selfish mining strategies as laid out here. To quote the article "(a difficulty algorithm) based on the chain weight (weight/time), then mining uncles would not increase the rate of the best chain, and uncle mining would be prevented." Feb 13 '19 at 8:21

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.