Topic: Fundamental flaw in consensus algorithms? - page 2. (Read 2568 times)

The only easy answer to that is a state sponsored coin where all mining is linked to a social security number and each number is capped at x reward per day.  The government would just manufacture a few million numbers for themselves and siphon off all the transaction fees until they accumulated all the wealth though.  There's no easy answer in getting out of the finite resource linked to the blockchain.  Combining proof of capacity and proof of stake to cancel out each other's weakness sounds promising to me.

So now discussion of consensus
algos other than PoW are consigned to altcoin discussion?

I'm well aware that this section serves as
the equivalent of the sewers where disagreeable
posters get dropped (by pushing a red button that
opens a trapdoor below them - whoosh), but this
is getting ridiculous.

BTW welcome back AnonyMint.

Please note that Gregory Maxwell moved this thread from the Bitcoin Technical & Discussion thread. I explained more about this fraud at the following linked post:

https://bitcointalksearch.org/topic/m.12231625

Note I have added a poll to this thread.

The poll would have been offtopic before, but now it is unfortunately ontopic.

The technical thread was unfortunately politicized by the aforementioned person.

I take it you are already familiar with Proof of Ticket?

https://bitcointalksearch.org/topic/alternative-to-proof-of-work-proof-of-ticket-1016517

He auctions off the block reward to the highest bidder - but you still have the problem I mentioned.

Please do not expect me to reveal my secrets before they are published.

People gamble actively, not passively. This is why this doesn't work for rational miners - they are always profit motivated. Now, if you were to auction off the block reward, then you would be in a situation very similar to POW. The only trouble is, voting will rely on transaction which themselves require consensus.

Maybe try reading johoe's lottery statement again. Is playing Lotto profitable? Do people do it?

I do not believe Lotto is sufficient, because the payoff it so small. But there are more strategies within that generative essence concept that I think do work. Please do not expect me to reveal my secrets before they are published.

Not sure what point you were linking to there?

Isn't this statement just obviously true, though for any rational miner?

This is my long standing (since 2013 when I as AnonyMint was debating with the author of Decrits) objection to proof-of-stake (a.k.a. proof-of-share was my misnomer at the time) that the entropy of the system is bounded and thus can be gamed. I think Vitalik refuted Andrew's other arguments to some extent (arguing both PoS and PoW have some subjectivity), yet this entropy argument remains true afaics. Thus I think it should be emphasized.

Note that if PoS does indeed have this inherent flaw, it is likely those who are aggregating stake will try to hide this and thus PoS coins may have no metric of how centralized they are. Of course a similar argument can be made against PoW in that we don't know which pools are really controlled by the same entity behind the curtain, but the distinction is the point below...


Agreed proof-of-work as it is formulated by Satoshi is resistant to the above self-reinforcing collapse into centralization (a.k.a. entropy collapse) if the maximum hashrate that can collude for attacking the network is less than some figure between 25 - 51% depending on network propagation vulnerabilities.

I'd appreciate anyone pointing out any errors in this post.

I don't believe that is true:

https://bitcointalksearch.org/topic/m.12219576

But idea still needs to be proven in a real design.

Proof of burn has very similar characteristics to POW, it will centralise in the same way - those with the most burn capacity will win the competition for block production.

If you make mining non profitable, though, there will be no incentive to mine.

Section 4.4 of  https://download.wpsoftware.net/bitcoin/pos.pdf

Seperately, selfish mining is basically orthorgonal. And as you note, it only works when a miner has a very high proportion of the hashrate. A system where any participant has anywhere near one quarter of the authority isn't all that decenteralized.

This is just an idea which I would like to get feedback on.

I am thinking the fundamental flaw in any consensus algorithm that relies on some asset, is that the economic game theory is such that the asset will concentrate towards a winner take all over time.

For example in proof-of-stake or proof-of-importance (reputation+stake) the miner with sufficient stake and/or reputation can keep the minority chain sufficiently paced but more often behind the other chain such that most vote on the other chain, but ultimately he uses his power to vote the minority chain to the permanent lead thus orphaning all the rewards of the lesser stakes who were fooled into voting for what they thought was statistically the more likely longer chain. There are other variants of this strategy.

In proof-of-work selfish mining works in general when the adversary has at least 33% of the hashrate, and 25% under some network configurations.

One potential solution I see to this problem is to make mining an unprofitable asset burning operation.

But if this is accomplished by burning coins, eventually end up with 0 money supply.

Or I did I think show it might be possible to eliminate the selfish mining strategy in proof-of-work but at the cost of penalizing the honest party to a double-spend, by paying all chains proportionally. I am not clear how this could work for non-proof-of-work.