Source of Entropy

[strophy]

Perceived Problem

In order to keep the network secure, the selected consensus mechanism must retain a source of entropy. Historically, the proof of work hashes provided this security to the Dash network. One of the main criticisms of a proof-of-stake model is around security concerns related to the source of entropy.

Proposed Solution

• Utilize BLS-based proof-of-stake

Key Questions for Discussion

• To what extent is proof-of-work entropy still important within the system? What would it take to eliminate these dependencies?
• What form of proof-of-stake makes sense for Dash?

YouTube link: Potential Components of a Solution
Lire en français

 

[qwizzie]

• To what extent is proof-of-work entropy still important within the system? What would it take to eliminate these dependencies?

If we want to utilize BLS-based proof-of-stake as a solution, we need to have a secure and always available backup (in case BLS signatures fail).
I would like to know what kind of backup we can combine with BLS-based proof-of-stake, so far i just heard about possibly using a reduced proof-of-work as backup,
but maybe there are other backup solutions that can offer the same kind of security that proof-of-work offers ?

• What form of proof-of-stake makes sense for Dash?

Preferably a proof-of-stake that is :

* accessible to most Dash users (low barrier to entry)
* secure
* trust less
* does not take years to implement

 

[AjM]

I agree with @qwizzie

 

[toknormal]

It would be useful to first define what's meant by "entropy" in this context. Is it supposed to mean "randomness" ?

Entropy (which is a thermodynamic term) is a well understood physical property. It can be measured, quantified and predicted. But sometimes it gets used in comp-sci contexts to mean "synthetic randomness" which is a completely different thing.

So when the original post says

the selected consensus mechanism must retain a source of entropy

should that be interpretated as

"the selected consensus mechanism must retain a source of synthetic randomness" ?

 

[qwizzie]

It would be useful to first define what's meant by "entropy" in this context. Is it supposed to mean "randomness" ?

I think I read or heard somewhere that both proof of work and BLS consists of random hashes that are extremely difficult to manipulate. (I think Ryan said this in his Open House speech).
That is why BLS could possibly replace proof of work (but apparently still needs a backup system).

So keyword for entropy seems to be randomness.

Update : i found a reference to randomness in Ryan's Open House speech
Link :

Timestamp --> 1:01:29

 

[Ryan Taylor]

If we don't go to PoS, this threat becomes much less relevant toward developing a solution. That said, PoW is not a perfect source of entropy or "randomness". Why? Because technically, a miner could throw away successful hashes in search of one that is more advantageous (such as one that ensures his/her masternode is in an LLMQ). However, throwing away successfully mined blocks is expensive... the attacker has something at stake in order to manipulate the hashes. So while possible, it doesn't scale well and is impractical.

BLS signatures eliminate even that possibility, so they might be preferred over PoW hashes even if we keep PoW mining. However, this is not likely to be a high priority given the already high effectiveness of PoW mining.