Author

Topic: Let's talk about proof of capacity (Read 1207 times)

member
Activity: 80
Merit: 10
February 21, 2017, 10:36:44 AM
#4


Unlike hashing (basically, math), I don't see how input output or free space can't be faked. How do you know that your hard drive has free space? Because your firmware says it's a 1TB drive and you only have 500GB available. If you alter the firmware to show a different number of cylinders/sectors then you can magically create a 10TB drive or a 100TB drive.

The same can be done for input/output responses from storage drivers. I can create a fake compressed drive stored in memory and do 1M IOPS and there wouldn't be an easy way of detecting it.

In short, I'm not saying that it's not possible, but I think it's much harder to prove that a transaction really took place on a storage space or networking device than is simple hashing math, aka hashcash.

Yeah you could fake the space but that wont help you with mining burst coins / poc coins Tongue

you need the right nonce's from the drive, which you have to write down/calc one time (this can take a long time with a couble of TB ) and then never again, so you cant put 11 tb of nonce's on a 500 gb hdd Cheesy

even if it would show 11 tb you could only write 500gb of nonce's

you just idle around until theres a new block and then your pc will search the plots for a fitting nonce for the current block

You cant fake it you could just get a way better cpu which would calculate the nonce's in realtime and is not reading them from the drive... maybe you should look more into the topic of POC Wink

€dit: sorry for picking up that old thread Tongue
legendary
Activity: 1260
Merit: 1000
August 13, 2015, 11:09:36 PM
#3
Unlike hashing (basically, math), I don't see how input output or free space can't be faked. How do you know that your hard drive has free space? Because your firmware says it's a 1TB drive and you only have 500GB available. If you alter the firmware to show a different number of cylinders/sectors then you can magically create a 10TB drive or a 100TB drive.

The same can be done for input/output responses from storage drivers. I can create a fake compressed drive stored in memory and do 1M IOPS and there wouldn't be an easy way of detecting it.

In short, I'm not saying that it's not possible, but I think it's much harder to prove that a transaction really took place on a storage space or networking device than is simple hashing math, aka hashcash.

Heh, don't get me wrong, I'm only just beginning to look into the possibilities of PoC.  As I've already alluded to after looking at numerous options, when you try to fill it's security holes, you often end up recreating proof of work, or recreating a memory hardened algorithm.  I feel like this is one of the few logical directions with potential though.  

We've watched things like PoS evolve where coin age had to be dropped due to being a security hole.  PoW has a more uneventful evolution.  I think people are eventually going to come to the conclusion these consensus methods standing on their own aren't good enough and have to be combined with something else to fill their weaknesses.  Proof of burn is also a very underutilized tool that can be used to combat weaknesses in consensus mechanisms.

At first, I thought it was a giant weakness of cryptocurrency that coins lost per year, "zombie coins", could be extremely high, even estimated all the way up to 5% by some.  A currency isn't supposed to be inflationary or deflationary, it's supposed to be a stationary unit of equilibrium for trade.  If you can accurately model variables like this, this is not a detriment to your system, it's a security benefit, because you can do things like add a perpetual non-zero block reward to increase security, or add 2% annual stake to increase security without even causing inflation.  In this manner, people who die with encrypted wallets are having their wealth transfered to miners to secure the network rather than the entire population by not spending their coins.

Bitcoin does not utilize this variable for security because there was no way for Satoshi to model the data at the time.  Once the variable is known, the question is not should it be utilized, the question is, is the network most secure by adding this variable to PoW as a non-zero block reward, or does network security benefit the most by issuing this variable with proof of stake combined with another consensus mechanism?  Since we know PoS most likely can't stand on it's own, the next question is, what other consensus mechanism can be combined with PoS in order for it to function and possibly negate weaknesses of PoW as well?  

This post kind of jumped around in multiple trains of thought, but the general summary is, either PoW with a non-zero block reward will prove to be best, since it subsidizes transaction fees and security without a negative hit to economics, or something like PoC + PoS can be combined to defeat it.
legendary
Activity: 1442
Merit: 1001
August 13, 2015, 11:34:00 AM
#2
Many think PoW is the final solution to cryptocurrency.  The Adam Back idea was a good invention, but it's a very primitive, brute force solution.  We're basically all sitting around acting as safe crackers blowing tons of capital and energy in the process.  The reason for this expenditure of resources is because the blockchain has to be linked to a finite, real world resource in order to function, otherwise you have the term coined nothing at stake attack.

Linking a finite, real world resource into virtual space is a difficult task, and not many variables are eligible for the trip.  Things like processing power at an arbitrary computation, memory size, network bandiwdth, etc, can make the journey.  Deadlocking the entire internet bandwidth in order to mine coins doesn't sound like a very good idea though.  Memory size is one of the few logical variables that can be an alternative.

I saw that gremlin, Buterin, try to dismiss PoC entirely saying that it's energy saving implementation doesn't work because while sitting idle, the network bandwidth could be leased to other parties, creating an obvious nothing at stake attack.  If you didn't let the storage idle at all, then you're basically re-creating proof of work in a different way.  Although, it seems to me like that alternative proof of work method might use less energy than arbitary computational proof of work.

The point I'm trying to make is that PoC seems like one of the few obvious ways forward in getting rid of PoW.  It doesn't seem like either PoC or PoS can stand on their own, but combining the two together could probably get rid of both's shortcomings.  You also have other rarely talked about solutions that could be combined with PoC to try and fix it's problems such as proof of burn.  Coins already do exist that have attempted to combine PoW, PoS, and proof of burn together.  I believe Slimcoin was the first to do so, but I think the coin might be in dysfunctional limbo now.

Coins created with that many different systems working together requires the creator to fine tune all of it's variables perfectly in order for it to function, whereas the Satoshi method has much fewer arbitrary variables going on except for things like coin emission schedule.  This is the main reason Bitcoin became popular in the first place, because it's a primitive system based on simple concepts.  Whatever replaces PoW and/or Bitcoin most likely can't and won't be a simplistic system.

Unlike hashing (basically, math), I don't see how input output or free space can't be faked. How do you know that your hard drive has free space? Because your firmware says it's a 1TB drive and you only have 500GB available. If you alter the firmware to show a different number of cylinders/sectors then you can magically create a 10TB drive or a 100TB drive.

The same can be done for input/output responses from storage drivers. I can create a fake compressed drive stored in memory and do 1M IOPS and there wouldn't be an easy way of detecting it.

In short, I'm not saying that it's not possible, but I think it's much harder to prove that a transaction really took place on a storage space or networking device than is simple hashing math, aka hashcash.
legendary
Activity: 1260
Merit: 1000
August 13, 2015, 06:31:58 AM
#1
Many think PoW is the final solution to cryptocurrency.  The Adam Back idea was a good invention, but it's a very primitive, brute force solution.  We're basically all sitting around acting as safe crackers blowing tons of capital and energy in the process.  The reason for this expenditure of resources is because the blockchain has to be linked to a finite, real world resource in order to function, otherwise you have the term coined nothing at stake attack.

Linking a finite, real world resource into virtual space is a difficult task, and not many variables are eligible for the trip.  Things like processing power at an arbitrary computation, memory size, network bandiwdth, etc, can make the journey.  Deadlocking the entire internet bandwidth in order to mine coins doesn't sound like a very good idea though.  Memory size is one of the few logical variables that can be an alternative.

I saw that gremlin, Buterin, try to dismiss PoC entirely saying that it's energy saving implementation doesn't work because while sitting idle, the network bandwidth could be leased to other parties, creating an obvious nothing at stake attack.  If you didn't let the storage idle at all, then you're basically re-creating proof of work in a different way.  Although, it seems to me like that alternative proof of work method might use less energy than arbitary computational proof of work.

The point I'm trying to make is that PoC seems like one of the few obvious ways forward in getting rid of PoW.  It doesn't seem like either PoC or PoS can stand on their own, but combining the two together could probably get rid of both's shortcomings.  You also have other rarely talked about solutions that could be combined with PoC to try and fix it's problems such as proof of burn.  Coins already do exist that have attempted to combine PoW, PoS, and proof of burn together.  I believe Slimcoin was the first to do so, but I think the coin might be in dysfunctional limbo now.

Coins created with that many different systems working together requires the creator to fine tune all of it's variables perfectly in order for it to function, whereas the Satoshi method has much fewer arbitrary variables going on except for things like coin emission schedule.  This is the main reason Bitcoin became popular in the first place, because it's a primitive system based on simple concepts.  Whatever replaces PoW and/or Bitcoin most likely can't and won't be a simplistic system.
Jump to: