Topic: Some thoughts about consensus design (Read 326 times)

Haven't seen a nicer brief on how to explain consensus mechanism design. This can easily be extended to explain consensus through a directed-acyclic-graph (DAG).

Through a virtual Byzantine voting agreement, we discuss amongst peers which transactions they have logged and tally down weights for all transactions. The highest-weighted path out of the DAG directs in only one path, and thus acts as a decentralized source of time that prevents the notion of double-spending.

That works out for DAGs because of the fact that they are directed and acyclic. Going forward, one can then simply modify the process in which virtual Byzantine voting is performed with a plethora of different unique non-double-spendable sources. (whether it be to lock in computational time or virtual currency stakes)

You could try to make something that is almost impossible to be double accepted. And I think the idea of POS that uses Masternodes is more about to prevent the double accept, them to prevent the double spend. And as most of POS will be created together, it makes more difficult to creat to things that are almost equal and that the blockchain will accept.

A 51% attack will keep their fork private until they want to spend their coins.

Yes I think the problem is more related to science in general, the quantaization problem,

"Quantization" is the process of constraining an input from a continuous or otherwise large set of values (such as the real numbers) to a discrete set (such as the integers).

Regards more specifically to our system and latency ( assuming the internet is a millisecond machine, that speed of light in a optic cable[1]  )

[1] The speed at which light propagates through transparent materials, such as glass or air, is less than c; similarly, the speed of electromagnetic waves in wire cables is slower than c. The ratio between c and the speed v at which light travels in a material is called the refractive index n of the material (n = c / v). For example, for visible light the refractive index of glass is typically around 1.5, meaning that light in glass travels at c / 1.5 ≈ 200,000 km/s (124,000 mi/s); the refractive index of air for visible light is about 1.0003, so the speed of light in air is about 299,700 km/s (186,220 mi/s), which is about 90 km/s (56 mi/s) slower than c.   

ps-> assuming we are not colour blind 

'All' you need to prevent history attacks is historical evidence of forks. But for that to be possible, you have to incentivise references to said forks.

Please don't make me repeat what @anonymint told me which I already posted for you, because I want to minimize my posts.
He had pointed out very clearly that TaPoS is not effective as a short-range objectivity and thus by implication it will not prevent short-range double-spends.
It is only effective against a long-range attack.
And the behavior of the chain can be made to respect the TaPoS, but @anonymint is not going to reveal his design today.

There is no incentive structure here to control the behaviour of the chain, so what's left is the incentive to double spend instead. Coin days destroyed can be double spent just as easily as any other transaction.

The concept of time presumes a universal total order, yet for time to exist requires friction which implies the lack of a total order (and unbounded partial orderings) because not all witnesses can observe all events in "exactly now" real-time.
A humorous example about losing car keys points out that the only entity that knows the reality of "now" is the entity observing it.
Each of our realities can't be communicated precisely, only related. No other entity will ever be able to know all the thoughts that passed in your mind as they occurred.

Therefore triangulation of observations doesn't exist and is always an approximation of reality and thus never 100% final.


As stated there at the linked post.
The owners of the UTXO that is intertwined with the history by TaPoS don't want their UTXO to be reverted thus they will agree with the fork (the perspective of reality) which contains those TaPoS.
So the inertia is tying all the stake holders to the history.
Thus the nothing-at-stake is removed without consuming an external resource.
@anonymint helped Dan Larimer invent TaPoS in 2013.
It is probably the one significant invention he created that might help us convert nothing-at-stake to something at stake.
The inertia is that they have something at stake.
These are decentralized checkpoints which are implicitly a consensus and don't rely on any central party to state what the consensus is.

Vitalik argued that TaPoS can be subverted by bribing old UTXO owners to sell their private keys, but this is not only unrealistic but the current UTXO owners will not agree.
The unrealism is because the old UTXO owners and the new ones have a significant overlap.
The stakeholders en masse don't want to destroy their own money.
And this line of thinking will lead you towards the consensus system @anonymint has devised.
But you have to be cautious of the fact the majority are apathetic and can be divided-and-conquered by their own selfishness..

No. Of all nodes there must be at least one who can provide a proof that you burned your coin. That part works. Instead of burning a unit, my idea just gave a it a new deterministic owner. That works in an environment with reduced wallet hash sizes, and thus, participants sharing a unit.

That doesn't work because I can just rewind the prng after I send the first payment and send the same payment again to someone different.

my thoughts about this were most of the time that i would invent a coin where every unit can be spent as a whole and to only one destination (one prng output derived from the unit)

it then works as expected, no double spend, but very hard to use. if you're having not many units to choose the destinations, and the guy you want to pay has not many wallets/units or anything to receive, than you're out of luck. i guess it's a mathematically different money system where the rich can not just pay more but have more freedom to choose, and i think, it's hard to calculate then what a unit is worth

the alternative is to ask everyone you need to ask to pass it around until it's where it is supposed to be. kind of sucks, too

When I think about time-travel in general I am not thinking about how CERN machinery works and what physical theory is on stake ...
I think about protocols like NTP ( for historical reasons I will mention RFC 958 )

--AC

Exactly how are you defining 'inertia' in terms of TaPoS combined with burning?

Agreed.
Costless witnesses have nothing-at-stake.

Albeit an imperfect proxy which can theoretically be broken thus breaking the assumption that probabilistic finality is asymptotic.

We can in the multiverse.
Quantum computing will (if it is achieved) open a portal into the multiverse.

Not necessarily.
For example, transactions can be converted into inertia which becomes probabilistically unlikely to be double-spendable.

Hope these out-of-the-box thoughts help your conceptualization.

If you want to solve the double spend problem, you have to start with something that isn't double spendable. In bitcoin this is CPU cycles, which is a proxy for time. You can't double spend time.

Once you have this, you can use it to build your consensus mechanism. If you start out your design by choosing to use something which is double spendable (a transaction, stake*, a vote) then you're going to be chasing your own tail.

I keep having to remind myself that although financial markets operate under a nash-equilibrium without using PoW, they are built upon a foundation of the notion of an atomic transaction (handled by the centralised exchanges) which cannot be double spent. Therefore chasing some exotic consensus design around their operation will be lost cause.

You have to find something to use which you cannot double spend.


edit: *) except in the special case where that stake is stationary for the entire duration of the blockchain