toychain/docs/notes.md

---
layout: simple
base: '../'
---
# Design Notes


## Initial design choices

- note that serialization with JSON doesn't guarante key's order when using maps, so for CAS storage yaml is prefered
- server-side fileformat : yaml
- client-side fileformat : JSON

## Language choice:
  serverside: perl (or nodejs)
  clienside: javascript

## block chain structure

 minimally sized blocks :
  - segreated signature, timestamp, audit etc.
  - block: {address, previous, pow}
  - use balances for state rather than TXO to avoid the need keep txo states, and link txi to utxo

## Networking layer choise

 initally we thought websocket would be the simple choice for the P2P network,
 however the protocol implementation in perl is a little cumbersome, so we fall back
 to simple tcp sockets, no ssl support for the first version !
- blocking socket

 next version :
- protocol/routing/network layer: websockets (multi nonblocking clients)

## observation ...

- revokable block if sign w/ a DH on txo and ECDSA on txi

-  txroot (merkle root can be global)

-  PoW is to slow down chainupdate to avois spam
-  consensus is on the longest length
-  /!\ maximal work is hackable (as pow is exponential to difficulties)
 
-  sort of tx sequence can be done on node side
-  block chain can store tx out of order (FIFO style)

-  does mining really need to be conpetitive ?
-  if mining is done by all node the recipient of txo will be willing to mine!
-  if a .2% tax on transaction can replace/solve all VAT taxes
   then mining rewards is "artificially" inflated to compensate for a wasteful
   technology

-  nano-trading is stealing (i.e. taking advantage of who has the power to be informed fast)
   on need to put a "lowpass" filter on trading :
   speculation is necessary for long term protection against future uncertainty...
   short term predicition make no sense !

   long unpredictable queue time (mempools) can deter speculator as it would be like "lotery"
   lotery is a "gamified" tax extractor.

-  PoW can be replaced by ZK proof of some Work
-  should it be txo which carries the burden of proof of work ?
-  validator already do a great job making the system "alive" 
-  block reward could be proportional to the number of tx in the block
   i.e. the "real" work is indeed the verif not a hash-puzzle.
-  How to mitigate  DDoS on Mempool ?
   why would we want making block more and more difficult to add when technology advance ???
   (how to make spice of spamming constant through technology ?)

### misc & links

- testing javascript with a [gist:42835214f63750080a8cb73d4be5a4a1](https://hologit-ml.ipns.cf-ipfs.com/tstanchfield/code.git)