Bitcoin Forum

Bitcoin => Development & Technical Discussion => Topic started by: baritus on May 18, 2013, 04:08:20 PM



Title: Why Scrypt and alternative hashing algorithms are pointless.
Post by: baritus on May 18, 2013, 04:08:20 PM
The bitcoin network hashing power is and has been growing exponentially. GPU's have played a large part and are now becoming more and more obsolete as ASIC miners are shipped.

Litecoin, and other scrypt based cryptos, think that stopping ASIC miners joining the network will protect the value of the currency for miners. This idea is flawed. Here are my reasons:
1. As more ASIC miners join the BTC network, GPU miners will just move into alternative currencies
2. There are already enough GPU's to ensure steady displacement
3. The rate of ASIC entrants into BTC and the rate GPU miners leave the network will always be highly correlated.

Overall, there's enough GPU mining power in BTC to distribute the ASIC effect across all currencies, making a change of hash algorithm a useless concept.

Just some thoughts.. :)


Title: Re: Why Scrypt and alternative hashing algorithms are pointless.
Post by: abbyd on May 18, 2013, 04:19:59 PM
The bitcoin network hashing power is and has been growing exponentially. GPU's have played a large part and are now becoming more and more obsolete as ASIC miners are shipped.

Litecoin, and other scrypt based cryptos, think that stopping ASIC miners joining the network will protect the value of the currency for miners. This idea is flawed. Here are my reasons:
1. As more ASIC miners join the BTC network, GPU miners will just move into alternative currencies
2. There are already enough GPU's to ensure steady displacement
3. The rate of ASIC entrants into BTC and the rate GPU miners leave the network will always be highly correlated.

Overall, there's enough GPU mining power in BTC to distribute the ASIC effect across all currencies, making a change of hash algorithm a useless concept.

Just some thoughts.. :)

Uhhmm, no?  First off, bitcoin ASICs are really just coming on the market now - there won't be Scrypt/Litecoin ASICs for at least 6 months. Logically, as BTC GPU mining becomes obselete, the hashing power will move to Litecoin (it already has in a major way). Adding hashing power (thus increasing difficulty) has historically driven currency values UP. Just when miners think they won't make a profit anymore, the value of the coin goes up and gives them a boost...

Or maybe I'm not understanding your statements?


Title: Re: Why Scrypt and alternative hashing algorithms are pointless.
Post by: baritus on May 18, 2013, 04:35:29 PM
The assumption that price increases with difficulty has been proven false by LTC, BTC, TRC, NMC...+ a long list.


Title: Re: Why Scrypt and alternative hashing algorithms are pointless.
Post by: abbyd on May 19, 2013, 06:02:06 AM
The assumption that price increases with difficulty has been proven false by LTC, BTC, TRC, NMC...+ a long list.

ORLY? http://bitcoin.stackexchange.com/questions/419/is-there-empirical-data-about-a-relationship-between-bitcoin-price-and-difficult

Not an expert on this, but I see a correlation (not saying causation). Isn't it only logical to assume that more people mine when coin is more valuable?
I don't have a diff/price chart for LTC yet, but the recent skyrocket in price correlates quite nicely with the 10-fold increase in difficulty.


Title: Re: Why Scrypt and alternative hashing algorithms are pointless.
Post by: justusranvier on July 11, 2013, 10:46:38 PM
I can think of one good reason to change the hashing algorithm, and it has nothing to do with ASICs.

Homomorphic hashing algorithms (http://blog.notdot.net/2012/08/Damn-Cool-Algorithms-Homomorphic-Hashing) are apparently a thing. If Bitcoin used a homomorphic hashing algorithm then you could do some cool optimizations with regards to block propagation at high transaction rates.

If it's going to take 10 minutes on average to find valid nonce, wouldn't it be nice if the miner could immediately broadcast the transaction list he's working on so that the rest of the network could work on downloading and verifying that? With a homomorphic hashing algorithm the miner could compute the hash of the transaction list first and broadcast that, then combine that hash with the header to find a valid hash for the entire structure.


Title: Re: Why Scrypt and alternative hashing algorithms are pointless.
Post by: gmaxwell on July 12, 2013, 01:13:36 AM
I can think of one good reason to change the hashing algorithm, and it has nothing to do with ASICs.

Homomorphic hashing algorithms (http://blog.notdot.net/2012/08/Damn-Cool-Algorithms-Homomorphic-Hashing) are apparently a thing. If Bitcoin used a homomorphic hashing algorithm then you could do some cool optimizations with regards to block propagation at high transaction rates.

If it's going to take 10 minutes on average to find valid nonce, wouldn't it be nice if the miner could immediately broadcast the transaction list he's working on so that the rest of the network could work on downloading and verifying that? With a homomorphic hashing algorithm the miner could compute the hash of the transaction list first and broadcast that, then combine that hash with the header to find a valid hash for the entire structure.

You don't need homomorphism-wizzamacallit for that, _at all_.


Title: Re: Why Scrypt and alternative hashing algorithms are pointless.
Post by: justusranvier on July 12, 2013, 02:37:27 AM
You don't need homomorphism-wizzamacallit for that, _at all_.
It would be possible to do without homomorphic hashes, but I think there could be some advantages when the transaction lists become extremely large because the would that creates the header and searches for the nonce would never need to parse the entire entire transaction list. That job could be performed by someone else.


Title: Re: Why Scrypt and alternative hashing algorithms are pointless.
Post by: gmaxwell on July 12, 2013, 03:15:14 AM
You don't need homomorphism-wizzamacallit for that, _at all_.
It would be possible to do without homomorphic hashes, but I think there could be some advantages when the transaction lists become extremely large because the would that creates the header and searches for the nonce would never need to parse the entire entire transaction list. That job could be performed by someone else.
You can already do that right now because the transaction "list" is actually a hash tree. You can create a header knowing only a single root hash, or even update the coinbase with log2(n) hashes.


Title: Re: Why Scrypt and alternative hashing algorithms are pointless.
Post by: polarhei on July 12, 2013, 03:55:57 AM
Pointless? I don't think so as each method has its weakness. I don't think the alterative way is pointless. It is just time.


Title: Re: Why Scrypt and alternative hashing algorithms are pointless.
Post by: Jazkal on July 12, 2013, 04:30:02 AM
Because the Bitcoin community is just going to lay down for the ASIC's, the migration of the GPU's to Scrypt just guarantees that the Altcoins will be replacing Bitcoin in the next 1 to 2 years. What is going to happen is that the first gen ASIC market (everything out now, and about to come out) will mainly be held in a couple of big players hands. You'll have some expansion to the masses, but your talking minor diversification. The next gen ASICs which are in development now (which is on a magnitude better than the first gen), won't be sold to the public. They will be held in private hands, the same hands that are developing them. And that will be the end of it, it will be all centralized in two or three hands. That is how Bitcoin dies. I hope I'm wrong, really I do. I've been a Bitcoin fanboy since I got into it in early 2011.

I just hope that the Altcoins will stand up and fight the upcoming Scrypt-ASICs, unlike the Bitcoin community.


Full disclosure: My GPU farm is pointing to Litecoin, for the past month and I do have ASICs on order.


Title: Re: Why Scrypt and alternative hashing algorithms are pointless.
Post by: stdset on July 12, 2013, 06:27:44 AM
Adding hashing power (thus increasing difficulty) has historically driven currency values UP.
Why wind blows?
Because trees sway. It is so obvious! Always, when trees sway, wind blows!


Title: Re: Why Scrypt and alternative hashing algorithms are pointless.
Post by: jl2012 on July 12, 2013, 06:39:57 AM
Because the Bitcoin community is just going to lay down for the ASIC's, the migration of the GPU's to Scrypt just guarantees that the Altcoins will be replacing Bitcoin in the next 1 to 2 years. What is going to happen is that the first gen ASIC market (everything out now, and about to come out) will mainly be held in a couple of big players hands. You'll have some expansion to the masses, but your talking minor diversification. The next gen ASICs which are in development now (which is on a magnitude better than the first gen), won't be sold to the public. They will be held in private hands, the same hands that are developing them. And that will be the end of it, it will be all centralized in two or three hands. That is how Bitcoin dies. I hope I'm wrong, really I do. I've been a Bitcoin fanboy since I got into it in early 2011.

I just hope that the Altcoins will stand up and fight the upcoming Scrypt-ASICs, unlike the Bitcoin community.


Full disclosure: My GPU farm is pointing to Litecoin, for the past month and I do have ASICs on order.

If you think it's a good idea, just create your Bitcoin-Scrypt fork and ask people to follow it


Title: Re: Why Scrypt and alternative hashing algorithms are pointless.
Post by: defaced on July 12, 2013, 01:26:20 PM
The miners migrate, sounds good.


Title: Re: Why Scrypt and alternative hashing algorithms are pointless.
Post by: AlexWaters on July 12, 2013, 07:27:39 PM
Price influences difficulty for certain. There are a bunch of other factors, but for the most part this holds true (currently):

-People want to mine when the price rises.
-People can't afford to mine when the price drops.

The reverse hypothesis (difficulty influences price) is probably way less of a significant correlation. Other than the fact that an acceleration in hashrate growth - like we are seeing now - causes more bitcoins to be released than the anticipated 1block/10min. Those "extra" bitcoins are being put in the hands of people who likely just spent a bunch of money on new mining equipment, and want to sell the coins to offset their initial investment. Thereby increasing sell volume at strike.

You can see how many "extra" coins are being created per day here: http://blockchain.info/stats


As for alternative hashing algos, if it ain't broken...


tl;dr Price affects hashrate, hashrate doesn't affect price.


Title: 99%
Post by: Yurock on July 12, 2013, 11:21:37 PM
ASICs can screw Bitcoin, but we need these things, because otherwise someone will have ASICs anyway and will use them for "99% hashpower" attack.


Title: Re: Why Scrypt and alternative hashing algorithms are pointless.
Post by: Impaler on July 13, 2013, 04:26:21 AM
The long term solution is a 'basket' in which multiple hashes are used on a single chain, thus making ASICs too specialized to be effective and the mining stays in 'general' computing aka CPU and GPU.


Title: Re: Why Scrypt and alternative hashing algorithms are pointless.
Post by: enquirer on July 13, 2013, 08:48:28 AM
tl;dr Price affects hashrate, hashrate doesn't affect price.

And I think it's the other way around: diff increases means more capital invested in mining -> harder to launch 51% attack -> currency more secure -> people believe in it -> price rises

more money invested -> more incentive to invest in development and infrastructure -> price rises
more money invested -> more vocal proponents -> more propaganda -> price rises


Title: ASICs and security
Post by: Yurock on July 13, 2013, 09:50:59 AM
diff increases means more capital invested in mining -> harder to launch 51% attack -> currency more secure
The network can actually become less secure against certain attacks. ASIC is not something many people have or can easily buy. Total hashrate greatly increases, but at the same time, the number of miners greatly decreases. If one wants to stop Bitcoin, he will have a relatively easy target – the few miners that control most of the hashpower.


Title: Re: Why Scrypt and alternative hashing algorithms are pointless.
Post by: mmeijeri on July 13, 2013, 09:58:42 AM
Litecoin, and other scrypt based cryptos, think that stopping ASIC miners joining the network will protect the value of the currency for miners. This idea is flawed.

Not only that, the goal is illegitimate. BTC does not exist to serve the goals of miners. One potentially good reason to switch to scrypt, or better yet, to switch to a mix of scrypt and SHA-256 as proposed (https://bitcointalk.org/index.php?topic=180020.msg1879112#msg1879112) by Adam Back, is to make sure that decision making power isn't concentrated too much. It's too early to tell whether it will be necessary, but it's good to be prepared.