gave us valuable insights on the limitations of this technology and on how to improve the current model.

Can you elaborate on these details and/or create another post to explain what limitations you couldn't work around on the Steem mainnet?