Pages:
Author

Topic: [ANN] eMunie (EMU) - NOT a BitCoin fork/clone - call for beta testers - page 20. (Read 78405 times)

sr. member
Activity: 328
Merit: 250


So, to recap:  SHA256 is replaced with running merkel roots.  Whoever is able to fire up the most instances can assign themselves 79% of all new coins, and is able to double spend.

Dishonest hatchers will not lose their reputation in this case.  Honest hatchers will reject the double spends, but the honest ones are only 1% so the clients won't know.  All the clients will see are 99% of the amazon nodes telling them the double spends are valid.

Keep in mind a little something... when network is first started and runs with honest nodes for some time, nodes will build their reputation off the amount of available transactions... as network runs on, transactions will being going first to most trusted nodes before they reach untrusted nodes. If someone comes into the network with a million hatchers... this will be way more hatchers than is needed to meet the transaction volume of the network, so those hatchers will  likely never even be asked to process a single transaction.

As the network matures, it becomes hardened. New hatcher nodes that are not required to fulfill the transaction volume become redundant and would be ignored.

In this way, hatchers meet network demand perfectly. You cannot go firing off a billion-gazillion nodes as hatchers and expect them to be used by the network when there are only a few trickling transactions that need to be verified.

I wonder though... if perhaps you created a 3rd class of client, a "monitor", and allowed these monitors to do extra work in verifying that the network was secure.... like spot check the work of hatchers or something? Perhaps they could gain trust if they find a dishonest hatcher and report it to other hatchers? and perhaps they could also be rewarded with pay-outs if they find dishonest hatchers.

Of course you can fire up a billion hatchers.  If they hardly have to do any verification, there is no downside to firing them up and essentially getting free lottery tickets for more coins.  Just do a difficulty adjustment like bitcoin so that anyone who wants to win monies has to do a lot of work.  You need to just copy the bitcoin code, change the algorithm to the tree search, and make the difficulty adjustment work with the tree search.  Then at least you can beat litecoin.  Just don't botch the launch.  Litecoin completely failed at being a CPU algorithm, and that was its initial draw.  Litecoin only has something like 25 merchants.  If you sign up that many merchants plus launch as a CPU currency and cut out all the GPU farmers and do a somewhat fair launch, you will crush litecoin fast and take its whole market cap.
sr. member
Activity: 384
Merit: 250
It Looks Promising - I'd Like To Be Tester!
member
Activity: 84
Merit: 10


So, to recap:  SHA256 is replaced with running merkel roots.  Whoever is able to fire up the most instances can assign themselves 79% of all new coins, and is able to double spend.

Dishonest hatchers will not lose their reputation in this case.  Honest hatchers will reject the double spends, but the honest ones are only 1% so the clients won't know.  All the clients will see are 99% of the amazon nodes telling them the double spends are valid.

Keep in mind a little something... when network is first started and runs with honest nodes for some time, nodes will build their reputation off the amount of available transactions... as network runs on, transactions will being going first to most trusted nodes before they reach untrusted nodes. If someone comes into the network with a million hatchers... this will be way more hatchers than is needed to meet the transaction volume of the network, so those hatchers will  likely never even be asked to process a single transaction.

As the network matures, it becomes hardened. New hatcher nodes that are not required to fulfill the transaction volume become redundant and would be ignored.

In this way, hatchers meet network demand perfectly. You cannot go firing off a billion-gazillion nodes as hatchers and expect them to be used by the network when there are only a few trickling transactions that need to be verified.

I wonder though... if perhaps you created a 3rd class of client, a "monitor", and allowed these monitors to do extra work in verifying that the network was secure.... like spot check the work of hatchers or something? Perhaps they could gain trust if they find a dishonest hatcher and report it to other hatchers? and perhaps they could also be rewarded with pay-outs if they find dishonest hatchers.
sr. member
Activity: 328
Merit: 250
"or a tree search algorithm as the proof-of-work "

Isn't that exactly the same as a forward traversal of a transaction chain from a genesis transaction??

Which is what I've been explaining as the POW for about 8 hours.

The HD was an example of just one bottleneck in the system.

Is this really the same?  I'm not technical enough to know.  If it is, then just change the bitcoin code to replace SHA256 with tree search.  Just leave the rest of the bitcoin code the same.  It needs to have difficulty adjustments to work properly.  You will probably overtake litecoin at some point if you do launch it like that, because the CPU userbase is WAY higher than the GPU userbase.  Whoever attracts the most users wins.  You'll never pass bitcoin due to the network effect, but you might become #2.
member
Activity: 84
Merit: 10


So, to recap:  SHA256 is replaced with running merkel roots.  Whoever is able to fire up the most instances can assign themselves 79% of all new coins, and is able to double spend.

Dishonest hatchers will not lose their reputation in this case.  Honest hatchers will reject the double spends, but the honest ones are only 1% so the clients won't know.  All the clients will see are 99% of the amazon nodes telling them the double spends are valid.

Keep in mind a little something... when network is first started and runs with honest nodes for some time, nodes will build their reputation off the amount of available transactions... as network runs on, transactions will being going first to most trusted nodes before they reach untrusted nodes. If someone comes into the network with a million hatchers... this will be way more hatchers than is needed to meet the transaction volume of the network, so those hatchers will  likely never even be asked to process a single transaction.

As the network matures, it becomes hardened. New hatcher nodes that are not required to fulfill the transaction volume become redundant and would be ignored.
legendary
Activity: 1050
Merit: 1016
It doesn't need to.

Does BitCoin monitor the number of miners signed in and increase difficulty?  No.
hero member
Activity: 798
Merit: 1000
"or a tree search algorithm as the proof-of-work "


But this doesn't get more difficult as a factor of the number of hatchers.
legendary
Activity: 1050
Merit: 1016
"or a tree search algorithm as the proof-of-work "

Isn't that exactly the same as a forward traversal of a transaction chain from a genesis transaction??

Which is what I've been explaining as the POW for about 8 hours.

The HD was an example of just one bottleneck in the system.
member
Activity: 84
Merit: 10


So, to recap:  SHA256 is replaced with running merkel roots.  Whoever is able to fire up the most instances can assign themselves 79% of all new coins, and is able to double spend.

Dishonest hatchers will not lose their reputation in this case.  Honest hatchers will reject the double spends, but the honest ones are only 1% so the clients won't know.  All the clients will see are 99% of the amazon nodes telling them the double spends are valid.

Keep in mind a little something... when network is first started and runs with honest nodes for some time, nodes will build their reputation off the amount of available transactions... as network runs on, transactions will be going first to most trusted nodes before they reach untrusted nodes. If someone comes into the network with a million hatchers... this will be way more hatchers than is needed to meet the transaction volume of the network, so those hatchers will  likely never even be asked to process a single transaction.
legendary
Activity: 1050
Merit: 1016
They still have to get the data from somewhere and that takes time,  however you cut it.

With my read test, point 10 nodes at the same data source, they will all take longer to do it, and all will take longer than each node having its own data source.
sr. member
Activity: 328
Merit: 250
So to recap:  Whoever is able to fire up the most miners can assign themselves 51% of all new coins, and is able to double spend.   That's BitCoin.  Yet you have no issues there.

Additionally you are under-estimating the work involved with the verification with the HD bottleneck in place, disregarding the actual real world costs in order to produce such a high number of nodes, and claim that in a in a random distribution of n, 1 n is is able to have a much higher result (trust) than all the rest.

This is just pointless circle running.



I just answered this.  Anyone with 51% of bitcoin nodes CANNOT double spend.  An attacker must have 51% of MINING POWER to double spend.  You don't increase your power in the bitcoin network by firing up more trivial full validating node instances.  You have to do the SHA256 proof-of-work.  

Looks like Etlase2 already answered your "HD bottleneck" claim.

The 2nd most important function of proof-of-work (besides double-spend protection) is to get an initial distribution of the coins into a lot of people's hands.  If the only people able to get coins are people running millions of instances, this will be very few individuals, and the coin will fail.  You need to get a wide initial distribution for it to work.  The best current method of doing a distribution is to try to make it so it can only be mined on a CPU, since there are many more people with CPUs than GPUs.  Using a radix sort or a tree search algorithm as the proof-of-work is currently your best bet, since those run faster on CPUs and no one has implemented one yet in an alt-coin.  Another method is the Ripple model of assigning yourself all the coins and then giving them away to anyone with a forum account or a facebook account.
https://bitcointalksearch.org/topic/development-cost-for-a-gpu-resistant-coin-216049
hero member
Activity: 798
Merit: 1000
With this there is all kinds manner of other external influences to think about, memory bandwidth, CPU power, HD bottle necks.  Even an SSD is no match for DRAM, but sure, you can buy up a server or instance with 64GB of RAM and fit it all in, but then that has cost attached to it.

You did not address the fact that a (theoretically) infinite number of nodes can all use the same source of information. Mining is different because the more people that do it, the harder it gets...
legendary
Activity: 1050
Merit: 1016
As an exercise in "work" and "triviality"

Someone set up a DB, any DB, populate it with 1M rows, 5 fields a row, drop 512bytes into each row.   Query every row, read every byte of said row, time it.  Don't do ANY processing of data, just read it.

Now you have to do it 100 times as that's how many transactions you have to process.

Trivial?
legendary
Activity: 1050
Merit: 1016
It's not relying on HDD's it's an example of one.  It's 6am here so perhaps I'm not being as clear in my points.  I'm attempting to point out that straight SHA256 hashing has little else other than number crunching power to bog it down.

With this there is all kinds manner of other external influences to think about, memory bandwidth, CPU power, HD bottle necks.  Even an SSD is no match for DRAM, but sure, you can buy up a server or instance with 64GB of RAM and fit it all in, but then that has cost attached to it.

I'm feeling like I've ran around in circles all night, with a issue being presented, I explain a solution to that problem, then that solution is used to justify a totally separate argument.

hero member
Activity: 798
Merit: 1000
Anything that is relying on an "HD bottleneck" is cruisin' for a bruisin'. You do realize that only 1 HD is required for an infinite number of nodes and they can all get their hash trees from one source, right? RAID array of SSDs perhaps?
legendary
Activity: 1050
Merit: 1016
So to recap:  Whoever is able to fire up the most miners can assign themselves 51% of all new coins, and is able to double spend.   That's BitCoin.  Yet you have no issues there.

Additionally you are under-estimating the work involved with the verification with the HD bottleneck in place, disregarding the actual real world costs in order to produce such a high number of nodes, and claim that in a in a random distribution of n, 1 n is is able to have a much higher result (trust) than all the rest.

This is just pointless circle running.

sr. member
Activity: 328
Merit: 250
I'm sure I cannot think of even 1/10th of the downsides that switching proof-of-work from SHA256 to merkel roots has.  Let me see if I can get a dev in here to comment.
sr. member
Activity: 328
Merit: 250
Sure it takes some CPU, but a even a cell phone could do it.  Why don't you go in the bitcoin dev forum and say "hey guys why don't we switch the proof-of-work system from SHA256 to just calculating the merkel-tree of the blockchain?"  See what they say.  Maybe Gavin will post again and tell you.  Replacing SHA256 with a competition to run the most nodes is idiotic.

You are forgetting a very significant bottleneck, HDD.  It may be trivial in CPU time, but performing an entire sweep of the transaction chain, both forward and backward involves lots of disk access, which increases as the transaction chain grows, and that will slow you down.

Hashing doesn't require HDD or large amounts of memory, performing this work does.


The point is that the attacker has 1m nodes (essentially unlimited number) of nodes with equal trust to the honest nodes.

Now address my real questions:
"Lets say A and B are hatchers.  They both have the same work load capacity, and over time, they both do an equal share of the work.   D rolls in with 1 million amazon instances, which combined are 1 million times more powerful A & B."

Now instead of A and B each getting 40% of new EMU, they get .001%.  D gets 79%.

And?? D has done the majority of the work and had real world costs to pay, so he should get the majority share.

So, to recap:  SHA256 is replaced with running merkel roots.  Whoever is able to fire up the most instances can assign themselves 79% of all new coins, and is able to double spend.

Dishonest hatchers will not lose their reputation in this case.  Honest hatchers will reject the double spends, but the honest ones are only 1% so the clients won't know.  All the clients will see are 99% of the amazon nodes telling them the double spends are valid.
legendary
Activity: 1050
Merit: 1016
Sure it takes some CPU, but a even a cell phone could do it.  Why don't you go in the bitcoin dev forum and say "hey guys why don't we switch the proof-of-work system from SHA256 to just calculating the merkel-tree of the blockchain?"  See what they say.  Maybe Gavin will post again and tell you.  Replacing SHA256 with a competition to run the most nodes is idiotic.

You are forgetting a very significant bottleneck, HDD.  It may be trivial in CPU time, but performing an entire sweep of the transaction chain, both forward and backward involves lots of disk access, which increases as the transaction chain grows, and that will slow you down.

Hashing doesn't require HDD or large amounts of memory, performing this work does.


The point is that the attacker has 1m nodes (essentially unlimited number) of nodes with equal trust to the honest nodes.

Now address my real questions:
"Lets say A and B are hatchers.  They both have the same work load capacity, and over time, they both do an equal share of the work.   D rolls in with 1 million amazon instances, which combined are 1 million times more powerful A & B."

Now instead of A and B each getting 40% of new EMU, they get .001%.  D gets 79%.

And?? D has done the majority of the work and had real world costs to pay, so he should get the majority share.
sr. member
Activity: 328
Merit: 250
There is no "work" being done.  POW has been eliminated from the system.  Validating transactions is a trivial task.

@timeofmind:  Anyone with 51% of bitcoin nodes doing trivial validating work CANNOT double spend.  An attacker must have 51% of MINING POWER to double spend.  This has been eliminated from eMunie for some reason.

"Lets say A and B are hatchers.  They both have the same work load capacity, and over time, they both do an equal share of the work.   D rolls in with 1 million amazon instances, which combined are 1 million times more powerful A & B."

Now instead of A and B each getting 40% of new EMU, they get .001%.  D gets 79%.  Also, the odds that one of D's instances accumulates a higher trust is incredibly high.

Firstly, POW has not been eliminated, and that is an important point. Calculating the merkel-tree of the whole chain of transactions is not trivial. It takes some CPU. If your node pretends to do this work, it will eventually get caught and lose its rep. If you want to actually do this work and build your reputation with millions of nodes, then all the power to you. You deserve your share of the funds for your work done. Keep in mind that there will be heavy competition, as everyone will want to do this work, and they will likely all fire off enough nodes so that they can do as much of the share of the work that is possible. In fact, this system would probably be most secure if you allow every participant to fire off as many nodes as their CPU can handle, and then let them fight over the transactions.
Sure it takes some CPU, but a even a cell phone could do it.  Why don't you go in the bitcoin dev forum and say "hey guys why don't we switch the proof-of-work system from SHA256 to just calculating the merkel-tree of the blockchain?"  See what they say.  Maybe Gavin will post again and tell you.  Replacing SHA256 with a competition to run the most nodes is idiotic.

"the odds that one of D's instances accumulates a higher trust is incredibly high."

I fail to see how on earth you came to that conclusion.

In this scenario, no hatcher would have a significantly higher chance, or resulting trust anywhere near even 1 order of magnitude greater than any other node in the system.

If that's the case then you should be able to flip a coin 1000 times and have a clear majority winner of heads v's tails every time you perform 1000 flips.  But that doesn't happen.

The point is that the attacker has 1m (essentially unlimited number) of nodes with equal trust to the honest nodes.

Also, address this:
"Lets say A and B are hatchers.  They both have the same work load capacity, and over time, they both do an equal share of the work.   D rolls in with 1 million amazon instances, which combined are 1 million times more powerful A & B."

Now instead of A and B each getting 40% of new EMU, they get .001%.  D gets 79%.
Pages:
Jump to: