Momentum 2.0 Discussion

[bytemaster]

Fast validation times are a requirement.   It is easy to make a memory hard pow that takes too long to verify. 


Sent from my iPhone using Tapatalk

How about launching a slow validation coin, and then create a second currency/genesis bitshare block that maps 1:1 to "SlowCoin" when they are all mined?

That would allow a memory hard pow, right?

This is interesting, and possible because the value of the coin is based almost entirely on the social contract being honored by the backing company.  You could make arbitrary changes like this and it wouldn't be that big a deal so long as the social contract doesn't change to the downside.

This may work for the ProtoCoin but part of the purpose of the ProtoCoin is to test features of the final chain. 

[Lighthouse]

Fast validation times are a requirement.   It is easy to make a memory hard pow that takes too long to verify. 


Sent from my iPhone using Tapatalk

How about launching a slow validation coin, and then create a second currency/genesis bitshare block that maps 1:1 to "SlowCoin" when they are all mined?

That would allow a memory hard pow, right?

This is interesting, and possible because the value of the coin is based almost entirely on the social contract being honored by the backing company.  You could make arbitrary changes like this and it wouldn't be that big a deal so long as the social contract doesn't change to the downside.

[ThisNinja]

Fast validation times are a requirement.   It is easy to make a memory hard pow that takes too long to verify. 


Sent from my iPhone using Tapatalk

How about launching a slow validation coin, and then create a second currency/genesis bitshare block that maps 1:1 to "SlowCoin" when they are all mined?

That would allow a memory hard pow, right?

[barwizi]

true... if only human brains were stackable like servers.

[bytemaster]

Who ever generated the captchas would know the answers.


Sent from my iPhone using Tapatalk

From where i am standing

1) We all seem against VPS etc
2) There is a GPU miner in the works, if not two.
3) The original intention was a well dispersed CPU only coin.

People should stop complaining about amazon users, they are basically just outsourcing hash power.

here's the real deal.

Botnets 1,2 and 3 plus GPU farms 1,2,3,4,5,6,7 ~ 15 are about to wreck havok combined. with a single i7 you'll be lucky to get 0.01 PTS per day.
Momentum 2.0 is the beta product and has to deal with these issues yet remain feasible. So the question really is how do we go to being a truly CPU mined algo without complicating the mining process. Keeping it scalable for those with multiple machines? IMO it's all down to tweaking the algo making it VPS harder and GPU NO-NO, for those that don't get what i am saying. Current arguments like registered mining and long term maturity do nothing in the end, as you know once i am registered i can point all my miners to one account or use one machine to pipe all the others. Long term maturity falls on rocks because botnets will just mine, for those six months plus more. with GPU farms that makes it worse, i know quite a few farms, much to my awe ( 700 mh/s + on scrypt). These belong to companies and individuals who can mine 12 hours of any coin and 12 hours of momentum and still be very profitable. We need sharks and whales in the pool to make it interesting and bring in lumpsum investments, so lets make something scalable, convenient  yet not easy to rape.

Thank you barwizi... you hit the nail on the head.   

Why don't botnet operators CPU mine for Bitcoin?  Because the profits are too low.   We need to achieve the same result.   Right now a GPU farm is only 3x more powerful per card than a CPU farm and of course you can rent GPUs on Amazon as well.   3x is a relatively small gain compared to what Litecoin saw (right?) and many people have GPUs.... I think we need to separate concerns:

1) Giving away the coin like 'free AOL CDs' 
2) Securing the network with hash power.

When we try to 'mix' these two you have things that don't work.   

[barwizi]

Who ever generated the captchas would know the answers.


Sent from my iPhone using Tapatalk

From where i am standing

1) We all seem against VPS etc
2) There is a GPU miner in the works, if not two.
3) The original intention was a well dispersed CPU only coin.

People should stop complaining about amazon users, they are basically just outsourcing hash power.

here's the real deal.

Botnets 1,2 and 3 plus GPU farms 1,2,3,4,5,6,7 ~ 15 are about to wreck havok combined. with a single i7 you'll be lucky to get 0.01 PTS per day.
Momentum 2.0 is the beta product and has to deal with these issues yet remain feasible. So the question really is how do we go to being a truly CPU mined algo without complicating the mining process. Keeping it scalable for those with multiple machines? IMO it's all down to tweaking the algo making it VPS harder and GPU NO-NO, for those that don't get what i am saying. Current arguments like registered mining and long term maturity do nothing in the end, as you know once i am registered i can point all my miners to one account or use one machine to pipe all the others. Long term maturity falls on rocks because botnets will just mine, for those six months plus more. with GPU farms that makes it worse, i know quite a few farms, much to my awe ( 700 mh/s + on scrypt). These belong to companies and individuals who can mine 12 hours of any coin and 12 hours of momentum and still be very profitable. We need sharks and whales in the pool to make it interesting and bring in lumpsum investments, so lets make something scalable, convenient  yet not easy to rape.

[bytemaster]

Who ever generated the captchas would know the answers.


Sent from my iPhone using Tapatalk

[Lighthouse]

What would happen is some miner with ambitions to grab the entire market would pay Amazon's Mechanical Turk to have all of the private keys revealed and cached for when they are required.    It would be a lot of work, but once done would grant them all of the blocks.   Amazon would probably pay $0.05 per solution which means for $1 million dollars you would be able to 'own' $10 million dollars worth of blocks.     Perhaps you could scale it to billions of items...

Why not make it an insurmountable task by changing it periodically? 

[bytemaster]

What would happen is some miner with ambitions to grab the entire market would pay Amazon's Mechanical Turk to have all of the private keys revealed and cached for when they are required.    It would be a lot of work, but once done would grant them all of the blocks.   Amazon would probably pay $0.05 per solution which means for $1 million dollars you would be able to 'own' $10 million dollars worth of blocks.     Perhaps you could scale it to billions of items...

[liberman]

There is no point in debating the human-interrupt designs because they are all based upon secrets (the digital value represented by the turing test) and a DAC cannot keep secrets.   This system can only work with a central authority keeping the secrets and signing validation receipts which are then included in block headers.

Conclusion, human in the loop systems are fundamentally centralized and would also be farmed out to Amazon's Mechanical Turk.

Just thinking about this:

Imagine you put an enormous database of captchas, something like 20 million ones in a public database accesible via p2p, and not tied to any particular server.
Every captcha has a public key, and a private key, but the private key is in the drawing.
So for every region that is going to be mined, a mathematical formula drives the miner to the captcha corresponding to the public key derived from the region signature. Then the miner must manually introduce the private key drawn in the captcha before he can proceed to mine.
The only way to automate the introduction of the private key in the captcha is by manually have a team of humans that spend a lot of time in manually introducing the captchas they are going to mine.
You can optimize the requirement of the introduction of the captcha to be something like once every hour for the average computer. This algo would mean that mining is actually a process in which a human must be involved, so only people really interested would mine.

[luckybit]

I think a core question is, do we want mining to be human scalable or capital scalable.

Current coins including protoshares are capital scalable, you can mine as much as you can spend money.   By requiring individual registration, or human-required interrupts like a rotating captcha you make it non-economical to run 3,000 servers if each server has an average 5 interrupts randomly per day.  Doesn't matter if it's a cloud miner, doesn't matter if it's botnet, the human element does not scale like the hardware does. 

To a certain extent this makes the guy with 40 computers in his garage the hardest working and best paid, requiring constant attention (5 interrupts x 40 divided by 24 hours in the day means an interrupt every 8.3 minutes).

I actually don't hate the idea of random interrupts, it incentivizes people run it on the computer they are actively working on because otherwise they'll be periodically checking at best or waiting for an email alert.  One imagines a new suite of tools will emerge to try and automate this, and that becomes the arms race improving captchas.

Perhaps moving icons could be used as a human interrupt. It could be decentralized and use a random seed which could move a fixed set of icons into a random order which the human being would simply have to put into the correct order or previous order.  Not saying a botnet couldn't eventually beat that too but it would take time and effort for botnet operators to beat and it wont be a situation where the botnets mine all the shares up at launch.

You can make his life easier by giving people points for solving captchas which they can use to skip future captcha challenges. If you solve a challenge or a particularly tough puzzle then you earn a lot of points and you don't have to do any more puzzle solving for a while because you have enough credit. Basically the human being mines by solving the puzzle or captcha to take back his time freedom.

A botnet could not do this.

There is no point in debating the human-interrupt designs because they are all based upon secrets (the digital value represented by the turing test) and a DAC cannot keep secrets.   This system can only work with a central authority keeping the secrets and signing validation receipts which are then included in block headers.

Conclusion, human in the loop systems are fundamentally centralized and would also be farmed out to Amazon's Mechanical Turk.

I would propose that this is a preferable outcome, and much less scalable than EC2 or botnets.  Worst case scenario existing mechanical-turk esque markets are bid up and people working there make a little more along with the market provider.

Unless you make periodic human attention the limiting factor, you make scaling your mining operation about how much money you can bring to bear.

It will always be about how much money you can bring to bear.  Even with people in the loop.  Out sourcing to Amazon's Mechanical Turk is one way, but another way is hiring people on craigslist. 

So long as you are going to be centralized, there is no need for mining.  I could just collect email addresses and payout faucets and then sign a block every 5 minutes.  It would be far less annoying.

That is the main problem with the captcha solution. It does promote centralization and i have not discovered a decentralized way of doing it which is also transparent.

There may be some way of solving the problem but all of the ideas I have are based on the human attention / interaction leveling the playing field. I don't think we should give up, we should continue to research ways to provide incentives to human beings not to cheat with botnets.

[bytemaster]

There is no point in debating the human-interrupt designs because they are all based upon secrets (the digital value represented by the turing test) and a DAC cannot keep secrets.   This system can only work with a central authority keeping the secrets and signing validation receipts which are then included in block headers.

Conclusion, human in the loop systems are fundamentally centralized and would also be farmed out to Amazon's Mechanical Turk.

I would propose that this is a preferable outcome, and much less scalable than EC2 or botnets.  Worst case scenario existing mechanical-turk esque markets are bid up and people working there make a little more along with the market provider.

Unless you make periodic human attention the limiting factor, you make scaling your mining operation about how much money you can bring to bear.

It will always be about how much money you can bring to bear.  Even with people in the loop.  Out sourcing to Amazon's Mechanical Turk is one way, but another way is hiring people on craigslist. 

So long as you are going to be centralized, there is no need for mining.  I could just collect email addresses and payout faucets and then sign a block every 5 minutes.  It would be far less annoying. 

[Lighthouse]

There is no point in debating the human-interrupt designs because they are all based upon secrets (the digital value represented by the turing test) and a DAC cannot keep secrets.   This system can only work with a central authority keeping the secrets and signing validation receipts which are then included in block headers.

Conclusion, human in the loop systems are fundamentally centralized and would also be farmed out to Amazon's Mechanical Turk.

I would propose that this is a preferable outcome, and much less scalable than EC2 or botnets.  Worst case scenario existing mechanical-turk esque markets are bid up and people working there make a little more along with the market provider.

Unless you make periodic human attention the limiting factor, you make scaling your mining operation about how much money you can bring to bear. 

[Lighthouse]

I think a core question is, do we want mining to be human scalable or capital scalable.

Current coins including protoshares are capital scalable, you can mine as much as you can spend money.   By requiring individual registration, or human-required interrupts like a rotating captcha you make it non-economical to run 3,000 servers if each server has an average 5 interrupts randomly per day.  Doesn't matter if it's a cloud miner, doesn't matter if it's botnet, the human element does not scale like the hardware does. 

To a certain extent this makes the guy with 40 computers in his garage the hardest working and best paid, requiring constant attention (5 interrupts x 40 divided by 24 hours in the day means an interrupt every 8.3 minutes).

I actually don't hate the idea of random interrupts, it incentivizes people run it on the computer they are actively working on because otherwise they'll be periodically checking at best or waiting for an email alert.  One imagines a new suite of tools will emerge to try and automate this, and that becomes the arms race improving captchas.

mining is meant to be fire and forget, that is why systen stabilty is one of the big five,

Fire and forget is the exact element that enables botnets and cloud miners, I understand that everything to this point has been fire and forget but I'm proposing that feature is the problem.

[bytemaster]

There is no point in debating the human-interrupt designs because they are all based upon secrets (the digital value represented by the turing test) and a DAC cannot keep secrets.   This system can only work with a central authority keeping the secrets and signing validation receipts which are then included in block headers.

Conclusion, human in the loop systems are fundamentally centralized and would also be farmed out to Amazon's Mechanical Turk.