I would be slightly (perhaps more than slightly) concerned about the requirement of trust but delays should be completely acceptable.
And rightly so. After all, I know of a way to prevent 100% attacks, both contemporary and historical. Its called VISA. |
|
|
|
Hello Paul, IMO, the 0% attack is more dangerous than just the not responding attack, because at first glance there should be no reason to expect that sending yourself a transaction could have consequences, so why would someone refuse being paid for doing so?
Could you point me to some link or more info on 0% attack? (Not sure if it was part of the thread, if it is, I'd reread the thread). Regards, Shunsai It's described in the thread - basically it applies to bonded stake systems, where the stake is still allowed to be transferred after it becomes 'bonded'. If you bribe 100% of bonded stake to send a transaction to themselves at the same time, the network becomes forever stalled as there is no bonded stake left in the system with which to produce blocks. |
|
|
|
|
If someone has the ability to fix your car, do you pay them just for having that ability, or do you pay them to actually fix your car?
|
|
|
|
Isn't it also just an opportunity cost in PoW as well? If a miner devotes half of his equipment to an alternate fork, he may lose additional earnings but his mining equipment remains intact.
This is a common misconception. Although the miners equipment stays intact, they always lose cost of mining to electricity, and, on average this loss is equal to the block reward. |
|
|
|
Time is not a resource, people got lot of free time and if they could find an incentive tempting enough, they would trade it easily and cheaply.
I'm afraid you are utterly incorrect. The only reason PoW exists in the first place is to provide an unforgeable proxy for elapsed time. |
|
|
|
Well, not exactly true.
Proof of Work, implies consuming a lot of resources. To manage a 50%+1 attack an adversary has trade-offs to make (comparing incentives and costs) as of this proposal , Humans just use their free talent, zero or negligible costs are involved, so voting is free and we are left with only incentives that rule in favor of evil behavior, mostly.
Humans expend the most valuable resource on the planet: time. This is ultimately more valuable than anything else in the universe. The puzzles are created such than no man/machine can perform them faster than a human, therefore this represents the ultimate PoW. |
|
|
|
As of similarities between this idea and PoS, I can confirm that I think there is a similarity between the two as they in share a same vulnerability to zero cost attack.
Anyway, if you can show me how this proposed algorithm could mitigate the attack I described in my reply, I would be fine with dropping my objection.
Any attack you can level at this proposal you can level at any PoW blockchain since all miners are controlled by humans - they are vulnerable in exactly the same way, there is nothing special about whether a human or a machine under the control of a human solves a PoW. |
|
|
|
I am against this proposal as a framework for blockchain technology. But I do appreciate, op's dedication and the motivations behind this, kudos!
As I've briefly discussed this subject before (in the thread which has inspired @ir.hn for starting this topic), the problem of one-human-one-vote approach to blockchains, besides the slavery threat, is its political nature in the extent that can't be applied to socioeconomic games directly.
IOW, you can not fetch/change the ledger's state by means of a nothing at stake, voting system, because voters can easily commit to a zero cost attack and confirm illegal double spend transactions.
You really, really, don't get this idea at all do you? Its not PoS, it's PoW. |
|
|
|
Challenge: give me one NP complete problem for which we can randomly generate instances of any desired difficulty and which humans can solve better than computers.
Additional challenge: how to compensate for the time zone dependence of human hashing power? You're suggesting timezones exist in which no human would be mining... or at least the man power rate would vary too much depending on timezone? I'm not totally sure about that. For example, in traditional gold mining, crews will run 24/7 with a day shift and a night shift. It's fair to say the same thing would happen here, as its a similar, for profit enterprise. |
|
|
|
Let's remove this "problem" you've identified lol, only the longest chain wins. We then have a "new" problem; how to resolve chain forks when 2 blocks are found before either block has 100% acceptance, and so different parts of the network accept either block. This problem was solved in Bitcoin back in 2009 or 2010, and now you're saying that the solution is causing the problem. Those who merited your post should ask for the merit points to be returned.
I don't think you've read this thread correctly. No one is suggesting we remove the cumulative difficulty rule; the OP was suggesting that hashes with a lower numerical value than target are on average harder to mine than hashes at the target value, therefore there might be some merit in using this to order blocks. @tromp rightly pointed out that this would cause wild reorgs when a miner gets lucky with a low target, thereby increasing double spend risk. |
|
|
|
|
For anyone wondering what he's asking, here's some clarification:
He's talking about what would happen if you changed the LCR from sorting branches by cumulative difficulty, to sorting by lowest achieved hash vs target value. Sometimes miners mine blocks with a hash quite a lot lower than the target value, and on average these numerically lower hashes are harder to mine, so does it increase the security of chain to sort by them?
|
|
|
|
1. I understand what you mean, but i doubt some "randomness" is enough.
2. Then people could submit block with more efficient answer and could redo the block with less efficient answer.
3. True, but surely the block generation time invertal would be more sparse than other consensus method.
1. It may not be so simple, I agree. 2. I think this is analogous to how bitcoin works? Miners can generate a 'better' hash of the same block (i.e. hash with lower numerical value). The winner is the one who's block gets built upon. 3. Perhaps. Depends on the solution complexity, though, so with difficulty adjustment, it ought to self regulate. |
|
|
|
This is truly interesting idea since there's no "effective" algorithm to solve NP-complete based problem, but there are 3 major problem which are :
1. How would the nodes/protocol system generate the problem without rely on 3rd/trusted party?
2. How to verify the given answer is the best/most efficient answer, since AFAIK there's no way to verify it besides try all possible combination?
3. Since PoT rely on human's ability/speed to solve the, that would mean the block generation time would be unreliable and lead to long waiting time.
1. Some randomness is basically all that's required 2. "Although any given solution to an NP-complete problem can be verified quickly (in polynomial time), there is no known efficient way to locate a solution in the first place"* 3. Block generation time would have a variance in a similar way to the bitcoin block time. On average it would even out. *) https://en.wikipedia.org/wiki/NP-completeness |
|
|
|
Regardless, I will read the full paper, but we can at least conclude for now that humans can complete the task with near optimal solutions quickly. This is good because humans would be able to guess right answers better than machines presumably in some cases, and therefore a human would have a good chance at winning the block in this sort of proof of work system.
I look forward to reading your conclusion; this line of thinking shows promise. edit: this paper claims to present an algorithm for computing the optimum route order O(n^3.322) https://arxiv.org/ftp/cs/papers/0702/0702133.pdf |
|
|
|
Right from the wiki article on traveling salesman problem: Human performance on TSP The TSP, in particular the Euclidean variant of the problem, has attracted the attention of researchers in cognitive psychology. It has been observed that humans are able to produce near-optimal solutions quickly, in a close-to-linear fashion, with performance that ranges from 1% less efficient for graphs with 10-20 nodes, and 11% more efficient for graphs with 120 nodes.[47][48] The apparent ease with which humans accurately generate near-optimal solutions to the problem has led researchers to hypothesize that humans use one or more heuristics, with the two most popular theories arguably being the convex-hull hypothesis and the crossing-avoidance heuristic.[49][50][51] However, additional evidence suggests that human performance is quite varied, and individual differences as well as graph geometry appear to impact performance in the task.[52][53][54] Nevertheless, results suggest that computer performance on the TSP may be improved by understanding and emulating the methods used by humans for these problems, and have also led to new insights into the mechanisms of human thought.[55] The first issue of the Journal of Problem Solving was devoted to the topic of human performance on TSP,[56] and a 2011 review listed dozens of papers on the subject.[55] I was assuming the highlighted sentence implied efficiency versus computers, but I see how that assumption could have been wrong. Here is the full paper, I don't have time to look through it now but here it is: https://link.springer.com/content/pdf/10.3758%2FBF03213088.pdfI think you're confusing the meaning of 'efficiency' here. They're talking about solution efficiency, rather than solution speed. |
|
|
|
That's true, I failed to connect those dot's initially but another requirement to NP-complete would be that it is able to benefit from heuristics. Scientific studies have been done with traveling salesman and humans with no practice have an 11% advantage over machines. Just imagine how it will be when people practice it a lot.
I don't believe Conway's game of life is classified as NP-complete but even so predicting what Conway's game of life will do next would probably be NP-complete and benefited by heuristics so yes that would fit into this definition.
Got a reference for the first claim? I can only find a reference for human performance being within 11% of the optimal solution rather than solution time being faster than optimum machine performance. |
|
|
|
|
Hi Shunsai,
I'm sure you are aware of this, but I just wanted to point out that the punishment incurred by abusers of the signing processes is still only opportunity cost. Really, to be at parity with PoW and to be fully rid of NaS this has to be converted into realised cost somehow.
The trouble is, it is very hard to see all the attack angles when you have NaS looming in the background, so, although you've gone to great lengths to narrow them all down and implement counter measures you can never really be sure the gremlin of NaS has truly been banished.
Cheers, Paul.
|
|
|
|
I like that it got you thinking! The trouble is, we need puzzles that are solvable faster by humans than machines. Just because something is NP complete doesn't mean a human can solve it quicker than a machine, it just means it's tough to solve for. edit: that's not to say this idea itself doesn't have merit. Indeed the finest example of this idea I've seen proposed so far is PoW using conway's game of life: https://bitcointalk.org/index.php?topic=2977765.0 |
|
|
|
For instance I have recently suggested an alternative variant which I believe can totally fix the problem of mining high variance. It was very enlightening for me tho, actually I'm very excited right now to see such a relationship exists between my proposal and a crucial scaling proposal like sharding. You might want to take a look at that whitepaper I posted above. It takes your idea to the n'th degree. |
|
|
|
I meant that AFAICT in all the transaction DAGs (e.g. Byteball, Hashgraph, and MaidSafe’s PARSEC) every node validate every transaction. That is not true scalability as I have defined up-thread in my reply to @HeRetiK because the systemic validation complexity is O(n2) (assuming every node n communicates n transactions). Another characterization of the validation complexity is that for transaction volume to increase then every node’s validation resources must equivalently increase, which means we can’t scale by adding more nodes. For example (the last time I checked) Byteball centralizes to 12 witnesses which do this validation (and provide the transaction confirmations finality) and then the light nodes which are issuing the transactions should trust those witnesses. Yes, every node validates every transaction, that's right. Like I say, though, for true decentralisation the bigger problem with DAGs is that they cannot support lite nodes. |
|
|
|
|
Show Posts
