Topic: Which Proof of Stake System is the Most Viable - page 17. (Read 25752 times)

I like that idea. Does it mean transactions run a risk of picking the wrong recent block? As I understand it transactions can only be included in block-chains that contain the block they nominated, so if they nominated a block that gets orphaned, they will never be accepted into the main block-chain and the transaction itself becomes orphaned too.

This feels like it could have potential for double-spends worse than in Bitcoin. In Bitcoin, when the block-chain forks, both forks can contain the same transactions so most transactions are unaffected by a fork. In TaPoS, as I understand it, a fork means a lot of transactions will be reverted, and will have to be re-broadcast with a new nominated block. If the sender does nothing, the receiver will never get paid, so double-spending becomes the default result of a fork rather than something that only happens if a sender is actively criminal.

To avoid nominating a wrong block, a transaction would like to nominate a really old block, ideally the genesis block, so it can be included in either/both sides of the fork. Presumably the protocol disallows that by enforcing "recent". If so, does that mean that transactions expire? If "recent" means within 5 blocks, a transaction that fails to be included for 6 blocks can never be accepted. Transactions seem to be in a bit of a bind. They want to nominate a young block to minimise their risk of expiring, and an old block to minimise their risk of being discarded by a fork.

Is there anything to stop the same transaction (ie, same inputs/outputs) being issued multiple times, each time nominating a different block, to minimise the chances of it being orphaned? At most one of the versions will be accepted into the longest chain, so it would seem to be free. If so, isn't that the rational thing for a client to do (supposing it genuinely wants its transactions to be executed), and doesn't it make the "Transactions as Proof of Stake" thing pointless? Every transaction will vote for every block, just in case that block wins.

Is there a thread where these issues are discussed? I'm probably mis-understanding or re-inventing wheels.

here are some other relevant links:

first suggestion: https://bitcointalksearch.org/topic/proof-of-stake-instead-of-proof-of-work-27787
Wiki Page: https://en.bitcoin.it/wiki/Proof_of_Stake
Proof of Activity: https://bitcointalksearch.org/topic/proof-of-activity-proposal-102355

i messed up the link to the other half of the bitshares solution which is transactions as proof of stake. i think the tapos paper is probably the best of all the ones that i posted. it seems to solve all the problems that peercoin has and at a much lower cost given that peercoin is a hybrid that still issues coins to miners. the network is secured by coin days destroyed as with peercoin but instead of requiring individuals to volunteer there stake to the network and expose their private key, tapos simply uses the coin days destroyed from transactions. this shifts proof of stake from being a manual process to an automatic process. additionally, each transaction includes the hash of a recent block so as to make attacking the network by way of a secret chain unfeasible. since with pos systems the coin days destroys determine confirmation, someone with a secret chain could not leverage the coin days destroyed of the public chain. tapos uses proof of work mining only to produce blocks, not for network security. the difficulty of mining a block is inversely correlated to the coin days destroyed in that block, thereby preventing denial of service. although bitshares gets rid of the mining aspect from the original tapos paper with the delegate model, i do believe that tapos is still the underlying consensus mechanism.

Thanks clout.  Very good topic and
thanks for providing the links so
we can compare the 3 systems.
 
Here's just my 2 cents on the
matter:

I think the most important things
are how are new blocks formed,
and how is consensus achieved.
(Minting is a distant second,
and I didn't delve into that).

I'll give my summary of each
system and my conclusions to follow.

1. Peercoin:

- Instead of using the longest
chain method, they use the
chain with the longest "coin days"
of the transactions in the blocks
in the chain.

An interesting proposal, seems to work
in theory because someone
with a large wallet who sends
themselves coins goes back to
0 coin days, and thus cannot
be easily manipulated.

However, in practice is currently
relying on checkpointing... and
the whitepaper admits it
may not even work as well
as Bitcoin to prevent a 51%
attack.

2. Nxt

This uses a semi-random
deterministic protocol to
decide what node forges
the next block.

While this sounds quite neutral
and fair, the downside is this:

.

with 60-second blocks, 1440 minutes = 24 hours to fully confirm.  Not great.
If this could be improved, it would be excellent.

3. Bitshares:

Uses a "delegated proof of stake" protocol,
which is similar to the Ripple consensus algorithm
(basically a club of trusted nodes) except you
use your shares in the coin to vote for what
nodes you trust ("delegation").

The positive side of this system is faster
confirmation, but its more centralized.

They make the point that Bitcoin isn't really
that distributed due to mining pools, and
other proof of stake systems aren't really
that distributed due to people not running
full nodes...

Conclusion:

(these are just my initial OPINIONS,
which always subject to revision):

Bronze Medal:  Bitshares.

I like the Bitshares solution the least out
of all 3.  It seems far too centralized
and dependent on people rather than math,
although it still seems like it could work.


Silver Medal:  Nxt

Currently, it takes way too long to fully
confirm transactions,  but has some powerful
principles behind it, very well documented,
with smart people...and probably further
research could be done to improve confirmation
time.

Gold Medal:  Peercoin

I like this solution the most because
it is the simplest.  Simple solutions
are usually the best, which is why
proof-of-work and longest chain has
stood the test of time.

It would be nice to see if Peercoin
would work without checkpointing.

-----

There you go.  Will be interesting
to see what other folks think.

you have no idea that in the grand scheme of things it doesnt really matter.. be honest what will the differences actually achieve, dont think of the narrow minded problem, think global and long term

Peercoin POS is interesting as it still allows people who want more coins to contribute more hast that cost money to get coins

answer: none

if a certain mining script was any better then the other due to efficiency, then giv it enough time and greedy miners will pile on so many rigs to get the biggest slice of the reward pie. and others will do the same to compete until its no longer efficient.

no matter what th 'puzzle' is to get a reward, people will buy as much equipment as they can to compete. so changing bitcoin with an aim of efficiency is only going to last a few weeks before the competition meets or exceeds the old mining protocols.

so the big picture is that it wont change a thing

There has been a lot of discussion lately on whether or not proof of stake is a feasible substitute for proof of work. I have come to believe that it is a necessary progression of the crypto currency space toward consensus systems that do not involve arbitrary work and artificially valuable resources. Proof of stake, as I see it, is a step in that direction. Of all the proof of stake coins/projects, I found three thoroughly documented approaches to solving issues associated with distributed consensus - PeerCoin, Nxt, and Bitshares. Each project has its on merits as well as shortcomings. I put it to the community to vote and discuss on which approach has the best promise.

The relevant white papers are as follows:

• Peercoin - http://www.peercoin.net/assets/paper/peercoin-paper.pdf

Abstract:

• Nxt - http://wiki.nxtcrypto.org/wiki/Whitepaper:Nxt

Abstract:

• Bitshares  (DPoS - http://bitshares.org/delegated-proof-of-stake/)

Dpos Abstract:
[/list]