Skip to content

Latest commit

 

History

History
47 lines (44 loc) · 2.77 KB

README.md

File metadata and controls

47 lines (44 loc) · 2.77 KB
Raw

VALO

A slowly improving Python-based crypto-currency.

##Features

  • Currently VALO uses Proof of Stake with a premine. Proof of work is implemented and tested as well.
  • A transition to eigentrust or similar would move the stake from being money to trust and reputation.
  • The balance system is based on accounts instead of UTXO's, allowing for significantly smaller transaction sizes and therefore chain.
  • Transactions need a rate limit of some sort (for example a fee).
  • Transactions could be compressed by giving rewards for the usage of 5 byte usernames instead of 32 byte public keys.
  • The LWMA difficulty algorithm makes sure that the chain won't be attacked using hash attacks.
  • The hashing algorithm used is the blake2b implementation from pynacl. Switching to blake3, where possible, will improve hashing speeds a lot.
  • A flexible database backend exists, currently written using REDIS.
  • A transition to RocksDB or QuestDB would improve the query speeds and scalability drastically
  • The custom networking stack is tailor-made for this application. This allows for higher performance, even though the limitations of python apply.
  • Currently the networking uses jsonpickle for dictionary objects. Writing custom binary encodings for the two datatypes will further improve the performance.
  • Everything is currently written in Python, allowing for easy contribution.
  • A rewrite into Nim/Rust is targeted to take place once the basic functionality of the python-baseline stands and cross-client communication can be tested.
  • The ultimate goal is distributing jobs across a trustless kubernetes cluster, essentially using the kubernetes protocol as a smart contract layer.
  • Voting for foreign nodes should be allowed, to enable offline users to use their stake to secure the network. Both voter and delegate receive rewards for their actions.
  • Pruned history by removing the entire history if full consensus is achieved.
  • Node tiering system to split the task itself across groups (similar to "model parallel" in machine learning)
  • Sharding to split the invocations across participants (similar to "data parallel" in machine learning)

Contributing

There are two kinds of contributions possible to this project, as donations are not accepted.
The first and less technical way is to get creative and submit ideas for future improvements in the form of an issue or an email. Duplicates will be closed and linked to a main thread, there is no action required on your part. Anything is possible.
The second would be to look through current to-do's and start implementing one on a fork.