Bitcoin Forum

This came up in a different thread, and I wanted to talk about it, but not polute that thread, so, here I ask:

Does anyone think there will be another round in the bitcoin mining arms-race after ASICs have saturated the market?

I can’t see it. I think after ASICs are done flooding the market the difficulty will be so high there will be no way to justify purchasing mining hardware. Only people who’ve already made the capital expense will be mining - and some of those will not recoup their expenses - at least not for a long time.

I think this will happen on the first generation of ASICs from the current vendors.

Anyone see it working out differently?

No one can accurately predict the future.

The assumption that it ends with ASICs is based on several assumptions.
#1 Silicon will always be dominant.  I think that one day quantum computers will blow silicon away.  There may be other tech we have yet to consider that lays between here and there or after there.  Any or all of which would blow an asic away.  There could also be algorithmic short cuts that could be implemented in current silicon.
#2 The keysign algo is mathematically intractable.  

We've seen lately a lot of ingenious attacks against previously hardened crypto.  My guess is that the algo we use for key signing will eventually fall.  While we could always switch, the fact that the algo fell would undermine confidence in the currency and render it worthless.  

In fact if someone just took a large enough set of public keys and ran a brute force random attack against them on a large enough botnet, they could fairly easily show "control" of enough bitcoins to undermine the entire economy.  

There are ALOT of bitcoins tied to keys that have long since been lost, these haven't been moved at all in years.  I know I had a hardrive with some 1,500 bitcoins years ago that died but back then they might have been worth a total of $1.50

I would guess that the next mining innovation beyond what we see now, will be bots built to do ummm let's calling it private key mining. :)

Crypto experts claim that these algorithms, like SHA256, don't just fail all of a sudden. They first show the signs of weaknesses as mathematicians find a new ways of attacking them. There is always big time gap between showing the weakness and first demonstrated practical exploit, and those exploits usually are applicable in special cases. I don't see why would change from SHA256 in, say, 10 years from now, with some much, much more secure algo undermine the confidence in the BTC. It would be natural evolution, and even if then would be SHS256 realistically exploitable, accumulated proof of work would be barrier to any attack. And the story will probably repeat in few decades after that, and so on.

It's trivial to do. People have developed programs that just generate absolutely huge numbers of bitcoin addresses using your GPU.

Why don't you build a powerful network and just run these for a few months, checking each created address to see if it has any bitcoins in it, and let me know how many bitcoins you steal.

This makes no sense. You can generate the addresses on any GPU or asic you want, and chance that you generate any address that has ever been in the block-chain is close to 0.

List of addresses that have any BTC is even smaller, something around 150MB. If checking generated address against those addresses would have any chance of success bitcoin would be dead long time ago.

..because with this, you would make many million times less bitcoins than with simple mining.
I can't reproduce the exact numbers nor have the thread available, but with comparing the mining difficulty with the 256 bit private key, you, well, get some pretty large numbers ;-)
In fact with such large numbers all that stuff is simply not imaginable. Like "more cpus needed than grains of sand on a beach" and the like.

I'm sure someone will produce real numbers here!

Ente

Actually, I think quantum computers becoming viable will end the effectiveness of many crypto systems - including bitcoin.

I think there's enough inertia in the current migration to ASICs to drive difficulty past 100M this year. So much of the cost of hardware is not ASIC, so I don't think Moore's law will help for a while.

I guess it depends on the value of the currency.

Am I wrong, but doesn't mining end around 2040?

It will not be the end with the first generation asics.  However, it probably would be if BTC maintained its price, but as many people point out it is growing exponentially. Therefore, while growth in difficulty will outpace the increase in BTC value, mining will still be profitable beyond the first big difficulty increase.

It only ends for new BTC, but it will continue for transaction fees.

WTB dead drive for 'research purposes' ;D

ASIC is a general term. GPU were once ASICS if you look back at Apple's Quickdraw accelerator card.

The current crop of BTC mining ASIC's are crude inefficient, there is at least an order of magnitude hash rate to be gained with better smaller designs. I mean lets face it most of the chips out there are 110nm fab clocking at 300MHz. That's  90's tech. not 21st century.

2140, so still a long way off.

I think that the current (first) gen ASICs are just the beginning. There is still a huge room for improvement. Liken them to a Commodore 64 compared to a Core i7!

Asics aren't the end by a long shot. Soon we will have super asics and then quantum asics and then super quantum asics and then...

Yes, because quantum computers haven't been heralded as the god send for the last 2 decades have they... You think they're going to make their way into bitcoin miner's hands over research, space and defense? Once they actually exist?

And cold fusion.

... kick ass asics.  ;D

Well, if they do come up with a working, functionally complete quantum computer SHA256 calculations become trivial. I don't expect we'll have one any less than 20 years from now. They've only just barely got the first hybrid quantum chips working. I'd love to be wrong though, there are a lot of things we could do with quantum computers that current digital computers suck at.

ASICs will probably get better, especially if fabbed at smaller processes but quantum computers change the game enough that an SHA or scrypt-based Bitcoin would likely not survive. Quantum computers would change the field of cryptography significantly. Because these algorithms are all based on the sums of large primes Shor's algorithm could be used to solve them vast orders of magnitude faster than current computers. Fast enough that one system could conceivably calculate out any and all possible answers fairly quickly. That's just simple COMPSCI 101 stuff.

This is one of the reasons that Bitcoin will not last indefinitely, I think it has the possibility to last for quite a while but if they ever build a fully functional quantum computer system the potential to bring down the network is there and eventually someone will.

To date, I think they can formulate about the first 10 digits, effectively with a standard computer, of any sha2-cypher. Beyond that, you have to just use brute-force, and "hope" for a solution, or a collision.

But as for ASICs... all they will do is squash out anyone with lower hash-rates, just as the GPU miners did to CPU miners. There is still only x-number of coins made per day, so the market will go from having several thousands of miners, down to about a few hundred, and then down to about three. The last three will be the ones who actually manufacture the ASICs... and I doubt that will be the ASIC producers we see here today. (BFL, AVALON, ASIC-Miner)... but it could be them... in the end, fighting over the same devalued coins they caused to devalue. (Using our money to make them what they are today. While they leave us behind.)

Funny thing is... once they all have 50THs each... that will make 49THs more than is required to "process" the few actual transactions that are trying to get in to the list, to be processed... (See, that will be three individual "processing gateways/servers", because there will be no more of the thousands of "GPU servers", running to collect the orders. Thus, even if the price does not fall, few will actually be able to get an actual transaction done. Unless they scatter the machines all across the world, and buy lots of internet connections, and then they will be in massive debit.)

Once the price hits $1000 per BTC... All the GPU's will have moved-on, and all the mini-ASICs below 1.5THs will be gone too. I am guessing, by income-tax time... next year. (It was already up to about 1/4 of that price, and people paid it...)

Then they will be forced to do a pump-and-dump of BTC into every other still-thriving alt-coin on the market. (Where all the GPUs will be mining... and they will try to rape us again... It will not work. We will just make a new system, that avoids all this horrible loss, and has a little more stability. Then they can come back and use the machines to HELP us instead of RAPE us.)

ASIC = AIDS
:P (Just like GPU was AIDS to CPU's)

But AMD didn't turn around and build mining rigs for themselves, and then stop selling us video cards... They made us more video cards! ASIC manufactures are building 1000 for them, and 1 for us... thus, selling us back 0.1% of the market, that we helped them build, with our cash.

asics cost about $0.10 each to make after your R&D costs are realized.  The more centralized bitcoin mining becomes, the higher the price will go and the more incentive there will be for people to mine (even small amounts would be worth it if the price was sufficiently high)

Sounds like your Compsci 101 failed you. Shor's Algo will not break hashing algo's like SHA, Grover's will. Shors will break ECDSA (bitcoin signing) by solving the ECDLP. I don't expect someone in Compsci 101 to know all of that though.

It is going to take some time to get to 28nm. 

..that's what we thought about ASICs just a year ago too ;-)

Ente

Anyone stealing Bitcoins with quantum computers (or any other technology) is doing a 'Mutually Assured Destruction' of Bitcoin.  Once word is out that Bitcoins can be stolen via brute force attack on the blockchain, it's value as a secured medium of transaction is gone and the price plummets near instantly.

The game will be over for everyone.  Only some entity who wants to destroy Bitcoin would do this.

I hope this can never happen.

I'm confident quantum computers will come and they will, of course, be light-years beyond amazing for code-cracking and Bitcoin mining. But having said that, we are still a very long, long way from anything approaching a quantum comp that you can pick up from a local hardware store. It may not even be economically feasible to do this in ... what 40 years?

I haven't been keepin' up with my tech news reading as much these days, but a working reliable early expensive quantum-based computer available even for researchers to order and purchase is at least 10 years away, I'd imagine. So basically so far in the future of BTC that its not worth considering (due to the rate of change). Early working and reliable q. comp's will solely be the property of the NSA and other similar power entities' code-breaking organizations for the first stretch.

Since ASICS are by their nature specialized circuits for mining, they are the end-all for silicon-based, traditional computing Bitcoin mining. Of course we will see Moore's law progression in efficiency for ASICS as tech develops, but I can't see any disruptive change coming anytime reasonable soon that will be anything like the shift from CPUs to GPUs, and now GPUs to ASICS.  If anyone else has any ideas on any potential disruptive tech that could do bring this kind of change besides quantum computing, please share, I'd love to hear about it.

There are other types of computers (mostly theoretical atm) that can give and check all possible answers in an equation.  Surely one will come out within 10 years.  It was mentioned they would just be for scientists and military, but if someone sees the profit to make one it will happen. 

I think the answer here is that there will always be encryption of some sort.  With greater computing power comes greater ability to encrypt.

Uhm. Yes, that's quantum computing.

---

Don't forget we can easily replace the algorithm for mining at any time, as soon as enough miners agree on it. Not all algos are solveable by quantum computing.

Ente