[Cryptography] Tezos

[iang]

Hi James,

On 21/08/2017 06:55, James A. Donald wrote:

> Tezos is a proof of stake crypto currency

My understanding is that Tezos is a DPOS or delegated proof of stake 
system, having adopted the mechanism invented by Dan Larimer for 
Bitshares/Steemit.

In short, DPOS separates out the problem of PoS into two layers:

  * a selected set of block producers (aka witnesses), and
  * an on-chain voting mechanism to vote in/out the producers 
proportional to stake.

This separation solves the "nothing at stake" problem.  I've written an 
introductory high level description to DPOS here:

https://steemit.com/eos/@iang/seeking-consensus-on-consensus-dpos-or-delegated-proof-of-stake-and-the-two-generals-problem

And here's a more practical take of the workings of the Steemit DPOS 
system, with numbers:

https://steemit.com/witness-category/@someguy123/seriously-what-is-a-witness-why-should-i-care-how-do-i-become-one-answer

> Reading the white paper, I see no explanation of how to solve the hard 
> problem of proving to all stake holders that fifty percent plus one of 
> the stake holders have chosen to support one outcome rather than 
> another when the number of transactions and accounts becomes very large.

Assuming that the DPOS method has been captured faithfully within Tezos, 
the elections by the 50.1% are conducted through an on-chain smart 
contract mechanism and are therefore fully visible and auditable.

> Tezos starts off as a centralized system, as all systems must in that 
> all proof of stake systems start off with the stakes held primarily by 
> those who wrote the software and those who funded them, but for 
> initial investors to cash out, must become a decentralized system, and 
> I am not seeing any plan to become a decentralized system.

I don't follow Tezos so am unclear how they intend to start off, but it 
is a challenging problem.  Let me describe how EOS intends to start off 
for some illumination on the issue (note, below).

As established precedent in 2009, the goal is to launch the chain 
without there being an operator.  In EOS.io's design, there will be a 
release of the final software, and that will be the end of the 
responsibility of the bootstrapping software authors.

Then, the community (quite large) that surrounds EOS will compete to get 
it up and going.  Many chains will start.  The goal is to get one true 
chains.  To achieve this we have a few things in the favour of quick 
consensus:

  - Substantial network effects of the one true chain being worth a lot 
more to everyone than multiple small chains.

  - Substantial pressure on the producers to agree quickly on the chain 
so that they can start making money, and not trying to produce for all 
wannabe chains.

  - Finally, the software is constructed to include a threshold of 15% 
of the value in order to trigger advanced functions like payments.  The 
software will include the list of EOS token holders and their value, and 
it will wait until 15% of that value votes on a confirmation to trigger 
advancment to the next phase.

It will be a lot of fun to watch.

> Of course that is par for the course for a lot of internet companies 
> these days.  Fund me today, and I will figure out how to make money 
> tomorrow.

:-) It also happens to be the fundamental problem of software.  To write 
it we need to fund the devs.  And once we have funding, we can take the 
risk of seeing if the ideas work out.  Hopefully that will lead to 
making money.

The difference with the blockchain process is that first we make the 
money.  Then we build it.  Then we make the money.  ;-)


iang


Disclosure:  I'm related to the EOS project and not to the Tezos 
project.  They are approximately similar projects at a helicopter 
level.  Both have sold tokens for something north of $200mm, both use 
DPOS, and both are proposing smart contracting blockchains.