[LTC] Changing the litecoin Proof of Work function to avoid ASIC mining?

[rph]

The miners can vote on whether they wish to change a protocol rule, as was the case with BIP16 (P2SH) in Bitcoin.

But do you hold the vote before or after the ASIC is deployed in significant quantity?
In effect, whoever decides the timing, decides the outcome of the vote.

-rph

[1715414799]

1715414799

[1715414799]

1715414799

[1715414799]

1715414799

[1715414799]

1715414799

[1715414799]

1715414799

[iddo]

The miners can vote on whether they wish to change a protocol rule, as was the case with BIP16 (P2SH) in Bitcoin.

But do you hold the vote before or after the ASIC is deployed in significant quantity?

Before, maybe non-mining users can also vote via some sort of a fee, though this can be messy...

BTW, I'll believe that ASICs that use 128.5kbytes (or less with TMTO) per scrypt() invocation exist only after I see any, the production cost of such an ASIC should be much higher than SHA256 ASIC.
Maybe some Bitcoin ASIC manufacturer like Avalon can estimate for us how costly it is to manufacture scrypt ASIC ?

[Tomatocage]

I'm all for ASICs as long as we can keep the cost of them prohibitively expensive so that only us early adopters can afford to stay on top. I'm so sick of all these newbs just now getting into LTC. Price them at 500-1000 LTC each and we'll be good.

*cough*

[bill.joy]

I'd suggest not to change it. Since BTC's success has proven that ASICs are good, LTC can follow the same trajectory: to take off further.

BTC is exploring on the moon where no one has been before. Being number one means BTC has to innovate in order to continue its success. LTC has been a successful follower so far. Being number two takes less market share but also less risk.  In almost any market, there's always room for a number two, a follower, who doesn't need to be too innovative.

[OneMINER]

It seems like a question of philosophy and degrees to me. Scrypt is GPU and ASIC resistant. Should it be ASIC hostile?

Some of the things talked about when litecoin was just a glimmer in coblee's eye were low initial cost mining, mining on consumer hardware, spreading coins out with a wide mining base, discouraging centralization, being open and transparent.

ASICs worry me. It's a fact that a scrypt ASIC would have to be much more expensive and more complex than their bitcoin mining counterparts. That could work against the majority of miners in the end. All PoW mining eventually comes down to hash/watt. If scrypt ASICs are going to be more expensive AND more efficient at hashing then that could stratify the mining economy. Potentially limiting mining to those with tens of thousands of dollars to spend in a pessimistic scenario. Going further, what if the ASIC producers refuse to sell? Self mining while protecting their ASIC designs in the courts.

There's obvious trust issues there. The real innovation of crypto coins is their trust-less nature. What gmaxwell said about litecoin differentiating itself resonates with me.

What would be the downside of small periodic changes to the PoW? It's pretty much limited to a software problem. I don't think any scrypt ASICs have started mining yet. That would have shown up on the network hashrate. So there's no miners to hurt at this time. Miners and users should have to do more frequent software updates. There would have to be steady software development.

The miner side of that could be a problem with GPU support for cgminer being dropped. On the other hand, a few months ago there was a very successful funding drive for litecoin development. Some will be unaware of changes or refuse to update their software. On the other hand, encouraging users to stay on the latest versions could give huge security and efficiency benefits to the net.

Changing the PoW also sends a message. Litecoin is ASIC HOSTILE, forget about it, don't even try. Without a doubt there will be other coins that would not take this path and scrypt ASICs would be produced eventually (maybe not without a doubt, I think it's likely though). It also says that litecoin is a work in progress, subject to change. I can see that in both a positive and a negative light. Shouldn't we keep trying to make the more perfect coin? It's not a tyranny, any software developers are free to make their bids and users are free to vote.

I'm surprised to say that I'm in favor of the proposal. I'm not fully versed in all aspects of this choice but it seems to be the lesser of two evils and would be in line with litecoin's principals and core philosophy, as I understand them.

EDIT: I don't think this is any kind of problem in the near future. I'm trying to imagine the network YEARS from now. I'd rather not see a couple of data center type mining farms being the only ones hashing for litecoin.

[bill.joy]

Potentially limiting mining to those with tens of thousands of dollars to spend in a pessimistic scenario. Going further, what if the ASIC producers refuse to sell? Self mining while protecting their ASIC designs in the courts.

There won't be only one ASIC designer. Don't worry about it. ASICs will drive the price of LTC much higher. The reason is simple: the initial few ASIC adopters will be very profitable since there rigs are faster, thus driving the price of ASICs high. As more people use ASICs, mining will become much less profitable. Miniers will then consider choosing between the following:
a) spending 2000 dollars to buy ASICs to mine just 2 LTCs in 3 moths
b) spending 2000 dollars to buy 4 LTCs

LTC price thus goes much higher. This is what we've seen in BTCs history.

By losing some miners, you will see more buyers of LTC. There need to be just enough miners in order to secure the network, not too many of them. But a successful currency needs more investors. LTC initially need to be GPU/ASIC proof in order to have enough miners to prove it is secure. But now I think there are more than enoughh miners.

I'd rather not see a couple of data center type mining farms being the only ones hashing for litecoin.

If this would be a problem then BTC would encounter it first. LTC will then still have enough time to avoid it. Being a follower takes much less risk.

[gmaxwell]

A significant advantage of existing cryptocurrencies is that the key attributes were made fully public up front, held constant,

Arguably the attribute made public was, technically, that litecoin was best mined on CPUs. If I found a bug that let me make additional litecoin out of thin air, would it be immoral for litecoin to fix it? You couldn't really make the same argument to introduce such a bug.

In any case— just a thought.

LTC price thus goes much higher. This is what we've seen in BTCs history.

This is the first time I've heard someone suggest that. I thought it was generally accepted that Bitcoin's price in march was primary related to the increased press Bitcoin received from the cyprus default and bank funds confiscation. 

[eroxors]

It certainly should be ASIC-hostile. ASICs allow for more mining centralization than card-miners. It's another thing that allows Litecoin to be distinct from Bitcoin.

Then the question remains, will other scrypt coins follow suit? or will there be a "split" between asic-minable scrypt coins (feather coin/bbq) and litecoin?

[bill.joy]

I suspect that gmaxwell started this discussion in order to let LTC fail. Changing hash algo is very dangerous and is the easiest way for a hard fork. Enough altcoins have dies because of hard fork.

I own both LTC and BTC. I sincerely hope that the BTC and LTC communities will work together. There will be alot more difficulties ahead along the way, for any currency to be finally successful.

[OneMINER]

Potentially limiting mining to those with tens of thousands of dollars to spend in a pessimistic scenario. Going further, what if the ASIC producers refuse to sell? Self mining while protecting their ASIC designs in the courts.

There won't be only one ASIC designer. Don't worry about it. ASICs will drive the price of LTC much higher. The reason is simple: the initial few ASIC adopters will be very profitable since there rigs are faster, thus driving the price of ASICs high. As more people use ASICs, mining will become much less profitable. Miniers will then consider choosing between the following:
a) spending 2000 dollars to buy ASICs to mine just 2 LTCs in 3 moths
b) spending 2000 dollars to buy 4 LTCs

LTC price thus goes much higher. This is what we've seen in BTCs history.

By losing some miners, you will see more buyers of LTC. There need to be just enough miners in order to secure the network, not too many of them. But a successful currency needs more investors. LTC initially need to be GPU/ASIC proof in order to have enough miners to prove it is secure. But now I think there are more than enoughh miners.

I'd rather not see a couple of data center type mining farms being the only ones hashing for litecoin.

If this would be a problem then BTC would encounter it first. LTC will then still have enough time to avoid it. Being a follower takes much less risk.

I disagree, I don't think there's a hard link between coin prices and difficulty going up. Your statement contradicts itself a bit there too. If prices were to go up because coins are harder to mine, then those mined coins would be more valuable. Making mining more attractive.

I think being reactionary and changing the PoW after the release of ASICs when centralization has become a problem would be a disaster. That just sounds like pure turmoil to me.

LTC was created to be as fair as possible. Attacking high priced custom hardware in the hands of a few in favor of consumer grade GPGPU mining available to everyone makes sense to me. The bar for entry is very low, easy for many to step over. If you have a PC you can start mining reasonably effectively with about $150. Litecoin will seem very different on the whole if you need to spend enough to buy a car to get in the door.

EDIT: Slight change to first comment

[laowai80]

It would be a really bad precedent if technical specifications of a coin are that arbitrarily changed. Other coins would start changing their specifications left and right, and it would end in a total chaos. I suggest we consider initial specifications of a coin as a written contract that absolutely cannot be changed. It'll be easier and safer for everyone that way. When rules are defined once and forever - this is an HONEST coin. Redefining rules is highly suspicious and screams of FRAUD. Just my modest 20 millilitecoins.

[iddo]

I suspect that gmaxwell started this discussion in order to let LTC fail. Changing hash algo is very dangerous and is the easiest way for a hard fork. Enough altcoins have dies because of hard fork.

It indeed can be dangerous if the network topology is truly decentralized, but not so much when there are no more that several dozens mining pools. We already have a precedent, the transition can be done safely in the same way that the BIP16 transition was done, i.e. first a majority of the miners vote that they are willing to accept the change in the future, and only then after there's a significant majority they switch to a client that actually implements this change.

[vnezapno]

There wont be really serious investments into litecoin until energy-efficient devices will come out.

[kramble]

My concern would be the perceived "fairness" of any change. The majority of the current miners are using GPU, but a change of algorithm would affect the relative profitability of GPU vs CPU mining (in extremis making GPU mining impossible) and I expect it would change the balance between different GPU cards. Whatever is chosen then someone will feel maligned (and certainly the very few FPGA miners we have currently in the wild).

There is a need to ensure that miners for the new algorithm are publicly available and promoted well in advance of the change, or we risk a scandal where a few lucky individuals gain a huge advantage by (for example) having a more efficient GPU mining implementation than is publicly available.

Finally the algorithm needs to be provably secure. I'm no crypto expert, but from what I've seen of the SHA3 candidates, these were designed for ease of hardware (ASIC) implementation, so this would rather limit the field, perhaps just to tweaking the parameters of the existing scrypt algorithm, which would probably not be sufficient to rule out ASIC (or FPGA) using external memory (its memory bandwidth hungry, not memory per-se, so just increasing the number of rounds, or the scratchpad size would not necessarily prohibit an ASIC solution). EDIT: And as for ECDSA, I believe fpgaminer already has an implementation in FPGA for vanity mining, so that algorithm is certainly not ASIC/FPGA proof.

Anyway, I'm not convinced ASIC will offer a huge gain over GPU, perhaps a little more energy efficient but the NRE cost will make it expensive compared with GPU.

So I voted no (perhaps to be expected as one of those evil FPGA developers )

[psychok9]

I hope ASIC mining will be push out from LTC...

[scyth3]

Yes.

ASICs will centralize the network. We can see it happening with BTC.



 

[earonesty]

It would be trivial to change the algorithm... as long as most of the miners switch... it's effectively changed.  And, what's more, you can make it perfectly ASIC resistant:  have a public forum vote on the next algorithm.   Select from a few alternatives.   Vote on the best one, and whichever wins, schedule a time for the switchover.   Old/new blocks can be distinguished with a version stamp.   No integrity has to be lost.   It would make for a great story and litecoin could remain "lite" forever.   Other coins can jockey for CPU position (my fav CPU is primecoin).

[uyo]

No, because I want LTC to strengthen (higher hashrate leads to stronger network).

[rangedriver]

[fatman01923]

http://www.prometeogallery.com/wp-content/uploads/2011/01/Invitation_NO_GLOBALTOUR-800x575.jpg

This.  My vote is no, let litecoin develop on its own even if ASICS will come out for it to mine them.