Thursday, December 24, 2020

Decentralized Internet

 

  • Dove into the fascinating world of decentralized internet building blocks developed by Blockstack.  These are the things I learnt
    • Blockstack is trying to build decentralized building blocks for the internet. The basic blocks like TCP, DNS are already decentralized but the cloud apps built on top of them are not. 
    • It is building a decentralized identity system
    • It is building a decentralized database (which only encrypts and can save it anywhere)
    • It is building a blockchain system for security and privacy (used for identity and smart contracts)
    • What this enables is that there is no one who owns your data but you yourself. Then there will be no difference between your data and cryptocurrency. They are all encrypted by the same keys
    • What that means in real life is
      • it will become easier for new startups to emerge once enough users own their own data as they can easily share the data with new companies without permission of google or fb
      • It also means that things will not be free (paid by ads) though that is also possible and it will be a more direct transaction than the indirect transaction that is today
      • Users should also get used to paying for their username, database to store things (Google is already moving in this direction anyway, but this has the benefit of being privacy preserving). Given the purchasing power differences between different currencies, if the price has to be same everywhere, this might be out of reach for most of the world. 
      • What this also means is that things will be a little slower (data has to be encrypted and decrypted) compared to the centralized applications which has huge implications for the deployment of these apps. If this problem is solved then it means a totally different set of companies will come up.
      • Having things centralized also had the benefit of faster iterations and machine learning. But if this approach of DApps becomes popular, then it directly improves the development of privacy preserving machine learning approaches where models are trained on device.