You are viewing a single comment's thread from:

RE: In Defense of Consortium Blockchains

in #eos7 years ago (edited)

You and Anonymint have been debating these issues since before Bitshares. I remember reading on Bitcointalk.

My opinion on Proof of Work is it favors TOTAL Proof of Work. Eventually botnets run by AI will do all the mining, probably by stealing CPU cycles. If government can come up with Stuxnet they can figure out a botnet to own Bitcoin. I think it is actually only a matter of time before humans are no longer profitable to be involved in mining because human labor has a cost while botnets don't require any human labor.

Proof of work in my opinion WILL be ungovernable once the block reward halving, difficulty increases, etc, push all the humans out of the mining roles. But hey maybe that was the idea all along? To be ungovernable by any human and let the AI control the whole of POW.

If Bitcoin is supposed to have mining for 50+ more years there is no way in hell human beings will be running mining farms by then. The whole industry will be automated, so we have to consider what that might mean right now. Personally I favor hybrid Proof of Work+Proof of Stake networks, as Total Proof of Work is not ideal.

I know these posts may make me unpopular or look like a lunatic but no one else is willing to post anything about the AI angle to POW. Demand for unlimited efficiency increases due to years of block reward having etc, put market forces on a certain evolutionary trajectory.

Reference

  1. https://steemit.com/crypto-news/@dana-edwards/will-an-artificial-intelligence-ultimately-become-the-global-mining-cartel-ai-as-mining-monarch
Sort:  

@dana-edwards this is the same way I see POW too. Hard to see it going away anytime soon but at the very least POW/POS hybrids are more ideal and sustainable in time. The other day I wrote an article that briefly talks about the Evolution of the Blockchain, referring to hybrids and POS being the new norm. https://steemit.com/cryptocurrency/@hedge-x/the-evolution-of-blockchain-has-just-begun-the-path-to-trillion-dollar-protocol-currencies

there is always the dark side of the moon present, but I do agree, bitcoin is far from decentralised, I do favor dan's statement here when you compare PoS and PoW, but you do make another important point, I'm not sure how AI will run a minenet the whole talk on AI will be unfolding in the years to come, all we have is fiction for now, I do think people should seriously consider and reconsider their ideals and goals for the future. I don't see AI and automation as that great of a advancement, it only makes people less valuable, bunches up a lot of resources for infrastructure and development that later totally blow out any humans out of the water, I don't really see things getting better since the internet was around and even before that, sure I do see tradeoffs but a lot of culture and value is lost in the process, so if we are going to be having a network that runs itself from AI first off someone has to release it and second it has to be worthwhile.

I don't see a way at the moment it will benefit anyone since either it has to serve someone who can be targeted or it won't be controlled and people do fear that, loosing control xD...

Botnets exist today and mine today, this is not fiction. That said, if we assume AI will become more advanced like we do with CPUs becoming more efficient then yes eventually there is a tipping point. Bitcoin will not be mined by human beings in 2100 as there is no in my opinion it will take AI that long to figure it out. AI does not have to be AGI, but can be narrow insect level intelligent AI, and that alone would be enough for a botnet to take over mining. The thing about Bitcoin, it could evolve into a fully autonomous agent while POS or DPOS will always be semi-autonomous until the stakeholders are bots.

Wow! Glad to meet you. I'm one of those people who think PoW+PoS as the best model and I have a history of not shutting up about Dash. Taking a hint of Delayed PoW of Komodo, I've thought of having PoW as a Service. If you need to attack Komodo which use the Equihash mining algorithm, you need to break Bitcoin's SHA-256 hashpower too which makes KMD much more secure than BTC.

I know that PoW is inefficient.But are we and should we really be done with PoW? Digital didn't make analog go away. It gave a new meaning to analog tech. Robots never destroyed any jobs. We just had to rethink and re-allocate human labor. It's time we look at PoW with a different eye.

The defining feature of PoW is that it is a brute force system. Software or computer networks can be meddled with. In 2010 92 billion Bitcoins were brought to existence because of a software bug. It was fixed of course. I'm not a programmer. So correct me if I'm wrong. DPoS is still vulnerable to a Ethereum like software hacks. There is no brute force layer an attacker has to go through. This concerns me a bit. Again; not a programmer. Correct me if I'm wrong.

To hack komodo, one needs to brute force Bitcoin first. What if other blockchain projects adopted a similar system? The networks will obviously be extra secure. This is especially important to newer kids on the blog.

This leads me to propose PoW as Service model. It won't really be a given service; but more of an adopted one. When more and more projects start depending on a specific PoW coin for an extra layer of security, then that PoW blockchain itself will have more value just as Ethereum increase in value when more and more ICOs stat using ETH. This can turn into a positive feedback loop. We could end up with one popular PoW coin per mining algorithm.

IMHO the single best candidate for the job is Dash. I'm not a professional cryptographer. I know that SHA-256 is part of the SHA-2 cryptographic hash function initially designed by the NSA which is a branch of world's greatest agressor and privacy invader that is called "the land of the free". It does worry me at times. I know X11 is relatively very efficient and it use 11 hashes. SHA-1 has already been broken. SHA-256 won't be that easy. But breaking 11 should be much harder even if they aren't as good as SHA-256 on their own. It is also worth mentioning the next iterations of this algorithm which are X13, X14, X15, and X17.

Now what if various blockchain projects start implementing dPOW (delayed proof of work) for added security. Personally I see a far better future for Dash compared to Bitcoin. The Hashrate has been growing crazy fast for dash.. The overall community has great harmony and it's a project with excellent fundamentals even though X11 isn't much time tested. Maybe that'd work in favor and give the algorithm some extra years before it's broken. I don't have a crystal ball (a major drawback of my life). I can only take informed actions.

And I think using existing PoW mining as an extra layer of security leading to better valuation for PoW coins would be an excellent thing for the blockchain community. Don't just dump the old; Recycle them!

Hopefully I added something valuable to the discussion :-)
Happy steeming!
@vimukthi

I think POW guarantees security based on economic assumptions. I think those economic assumptions as you say, only protect against the obvious brute force attack. I think true security requires looking at things statistically and no I do not think POW is more secure than hybrid POW+POS, not is it more efficient for stakeholders. What I mean is POS provides for more evovability in governance than POW.

I could go in on this topic but I would rather let Dan communicate on the technical side of things because he is one of the main innovators in POS. My opinion is that while I like DPOS, I have my own model of how things can or should be, based on what some people might think of as "sci fi" if I tried to describe it today. All I can say is what looks like sci fi if described right now will not look like sci fi in 5 years, and all the topics I discuss are in development, even if we do not see the fruits for up to 5 years.

The botnets, the agent based ai, the post I wrote on autonomous negotiators, these will be in the crypto-space probably in the next year or two. AI is coming and will merge with blockchain and this in my opinion is unavoidable because they have so much synergy and will make trillions in profit.

I'm very interested to know about your "sci fi" model.
AI is certainly begging to be connected with the blockchain. But the development must be kept open source. There could be private implementations but closed source AI could be worse than a FEDcoin.

@dana-edwards AI lacks entropy and thus will never autonomously dominate the universe. Please understand that humans necessarily make irrational decisions (e.g. based on emotions and hormones and incomplete local information reasoning), because otherwise the species would not be resilient.

PoW can never be dominated by botnets because infectable computers do not have ASICs that reach even 1/1000th of the computing power the latest ASICs which are always distributed first to TPTB (i.e. the Zionists) because they control the latest technology (e.g. 14nm currently) chip fabs.

I don't know what physics you refer to but nothing in the universe as I understand it "lacks entropy". Intelligence in the human, in the non-human animal, or in the AI, is physically equivalent as we currently understand physics at least. That being said my point is that mining does not actually require human emotions or any high order thinking. Mining by way of Proof of Work is a 'set it and forget it' situation.

On top of it being 'set it and forget it', it also promotes automation by design. Block reward halving happen every 4 years and this promotes increasing efficiency indefinitely. The logical way efficiency is increased is by cost reduction of the business operations. Eventually the inefficiency will be the human beings and the botnets will be the most efficient mining businesses until eventually a fully autonomous botnet is released.

Tell me why this will not happen? It is happening already. People have been finding hidden miners on websites all over the place. It is starting to become ubiquitous and even Intel is mentioning releasing a chip with a miner built in. Once the ASIC is built into the Intel CPUs, and the AI chipset built into the GPU, well how long before what I predict to happen becomes reality?

Assuming all computers will not have ASICs in 5-10 years is ridiculous. They all very well could have ASICs if it is profitable for Intel. I doubt it will even take 5-10 years, we might begin seeing these chips next year.

The one point you used to refute my hypothesis is that the current chipset is 14nm? But you do realize due to physics we will reach a maximum efficiency for chips pretty soon. When this happens the only way to scale is more chips not more efficient chips.

Reference

  1. https://www.marketwatch.com/story/nvidia-more-bullish-on-cryptocurrency-than-amd-2017-08-10

I don't know what physics you refer to but nothing in the universe as I understand it "lacks entropy".

A vacuum is one example.

On top of it being 'set it and forget it', it also promotes automation by design.

The engineering and research quest to produce smaller and smaller transactors is not capable of being automated. It requires the random occurrence of ingenuity. I am not going to write a treatise on this here.

The logical way efficiency is increased is by cost reduction of the business operations.

For mining it is mostly all about the electrical efficiency of the ASIC chips.

Once the ASIC is built into the Intel CPUs

These can never be even orders-of-magnitude as efficient as the dedicated, state-of-the-art ASIC chips.

But you do realize due to physics we will reach a maximum efficiency for chips pretty soon.

Malthusians never defeat human (nature's) ingenuity.

Paradigms will change as necessary, e.g. quantum computing, etc.. The ingenuity race will not cease.

That is where you are wrong. Chip design can and is already automated. Algorithm generation is also already automated. Even program synthesis (programming) is automated.

If we have a prediction market are you telling me you are willing to bet against AI being able to automate the mining industry within the next 20 years?

And my point about chips is there is a physical limit beyond which no chip can become more efficient. There will be a final ASIC design and once that design is found there will be no need to continue to innovate there because physics cannot be improved beyond the optimal design. This will happen for sure in our lifetime. So once the final design is produced it becomes a matter of how many chips you have as the "new currency" of POW rather than how efficient of a chipset you have. Even blogging on Steemit is not future proof because algorithms can blog too and soon AI will be able to do it better than people so why would I think I can blog for much longer? I see the writing on the wall.

References

Van Berkel, S., Turi, D., Pruteanu, A., & Dulman, S. (2012, July). Automatic discovery of algorithms for multi-agent systems. In Proceedings of the 14th annual conference companion on Genetic and evolutionary computation (pp. 337-344). ACM.

  1. https://en.wikipedia.org/wiki/Electronic_design_automation
  2. https://en.wikipedia.org/wiki/Program_synthesis
  3. http://futurecontent.co/automated-content-can-algorithms-write-your-content/
  4. https://en.wikipedia.org/wiki/Genetic_algorithm

Chip design can and is already automated. Algorithm generation is also already automated.

You're citing automation of existing technology, not the new innovations that have been required along the way to keep Moore's law advancing, such as Intel's 3D transactions, etc..

I have already stated the facts and provided a link to where I have explained it in much greater detail. You are free to believe anything you want to. It will still not make it correct.

Please go interview someone knowledgeable at Intel to become more aware of the reality. For example just reading a story about how difficult it was for Intel to release the latest generation CPUs and the human ingenuity it required to sort out various issues with the smaller transistors. In short, you're oversimplifying your assumption about the capabilities of AI.

As well, you do not comprehend the generative essence which is the entropy argument (which insures the capabilities of AI will never replace/dominate the necessary imperfection and randomness of nature). And I have no desire to try to explain it further to laymen (beyond what I already linked to), who have some fantasy about what they saw on Star Trek.

Even blogging on Steemit is not future proof because algorithms can blog too and soon AI will be able to do it better than people so why would I think I can blog for much longer? I see the writing on the wall.

Any intellectual activity that requires the randomness of ingenuity will remain the domain of humans. Remember for example humans can leverage AI against AI. AI is a replicated low entropy tool for humans. Every human is biologically unique. That random imperfection can't be captured by AI, because AI is top-down replicated phenomenon.

To the extent that AI ever does become randomly imperfect and unique like humans (i.e. if it truly becomes alive), then it will lose it's illogically postulated ability to form a universal total order (i.e. imperfections require partial orders). Humans can compute just as fast as AI can, we use computers to do it. I already explained why the notion of AI autonomously directing itself faster than humans directing AI is low entropy and thus not dominant.

In short, Kurzweil is selling snake oil.

Hey man, I know you are smart and everything but if you want people to listen to you stop talking to them like they are idiots.

If you swallow your pride and check your ego you will get much more acomplished. I like the subject matter you discuss but the way you present it is not pleasant to read and comes off more like you have a personal grudge than knowledge to share.

Perhaps the forcefulness of my factual exposition is due to trying to explain it nicely dozens of times already in the past and realizing that humans are necessarily hard-headed.

The necessity of irrationality was alluded to in my exposition.

Also take in account that @dana-edwards is declaring that I am wrong, and she/he clearly did not even take the time to go read all the linked explanation I provided my first polite post before he/she responded declaring that I am wrong.

I understand very well her/his fantasy and misunderstanding. I have explained this point numerous times over and over and over again. The disease of the propaganda about the AI Singularity is at the same level of junk (false) science as human caused global warming, feminism, oestrogenizing vegetarianism & fat-free diets, and other absolute bullshit propaganda that people believe.

Btw, I have no animosity towards @dana-edwards. I am friendly with everyone, but w.r.t. to factual debate, I have the responsibility to present my knowledge in a way that the other side actually takes notice of the facts. Unfortunately the nice and soft approach was not working with her/him. Yet I think my last response is not mean-spirited. Just trying to be forceful enough with the facts to penetrate the cranium and emotional centers of the brain and reach the pre-frontal cortex where the logic circuitry is.

Thanks for your feedback. I hope my explanation satisfies you to some degree.