Proof Of Stake

There is a lot of cutting edge computer science being used in blockchain technology and, because there are multiple blockchains and protocols being proposed and tested, there are various experiments going on at the same time. This is all very good in my view. We will eventually figure out which technologies are best for which applications and standards will emerge. I think we are three to five years away from that level of market stability, and possibly farther away than that.

One of the most interesting questions to me is whether we can figure out how to implement a proof of stake consensus mechanism in a large decentralized trustless public blockchain (ie Bitcoin, Ethereum, etc). The current consensus mechanism of choice is called proof of work and the specific proof of work implementation that is being used by most public blockchains is mining. Mining has a number of issues, most notably the massive amount of electricity consumed by miners to run these mining data centers. Other concerns with mining are centralization (a few large miners control most of the Bitcoin mining infrastructure) and exposure to 51% attacks (if one miner controls 51% of the network, they control the entire network).

Proof of stake is an alternative to proof of work and there is a lot of work being done to see if it can be implemented in a large public blockchain. The leaders of the Ethereum project have publicly stated a willingness to try to implement proof of stake on Ethereum and they are working toward that goal.

In a proof of stake system, you use the very coins/tokens that are at the heart of these systems as the “proof” that you can validate a transaction. Instead of using your capital to buy computers and electricity to run them, you just use your capital to acquire the coins/tokens and they allow you to validate transactions. This is all very simplistic, of course, but that is the heart of the idea.

I read the Proof of Stake FAQ on the Ethereum Wiki this morning. It’s an interesting (and dense) discussion of the various issues that crop up in a proof of stake approach and the algorithms and techniques that are being put forward to resolve these issues. If you are interested in blockchain technology and want to understand how all of this stuff works under the hood, you might do the same and give it a read.

I am hopeful that we will see a large public project, like Ethereum, attempt to implement a proof of stake consensus algorithm in the next year or two. Mining works. It has validated blockchain technology and allowed it to be commercialized. But my gut tells me that mining is not the best consensus system out there and that we can do better. And so we should see if we can. That’s what forward progress is all about.

#blockchain

Comments (Archived):

  1. William Mougayar

    And there’s another variance of both, called Delegated Proof of Stake (DPoS) that includes a small number of trusted miners in an otherwise PoS consensus method. Steem, for example has successfully implemented DPoS. We still don’t know if Ethereum’s research & development in PoS will lead to a pure PoS implementation or a variation thereof. Here’s something to chew on, re: DPoShttp://cryptorials.io/gloss…

    1. jason wright

      Lisk is in the final stage of preparation to launch its delegate network.It will be interesting to observe how it shapes up when live, the human condition being what it is…

    2. Christan

      Worked right up to the point someone wrote a GPU miner and didn’t release it. Two people have been dominating the steem mining for months.

      1. Ned Scott

        The point is irrelevant if we are talking about consensus and chain upgrades. (“Easy-to-compete mining” has absolutely nothing to do in an argument about moving to a sensible PoS algorithm. Beyond that, miners make up only 1/21 of the block production queue in Steem.) DPoS consensus has worked very effectively for years if we include Steem & BitShares and consider the many upgrades both chains have seen in their history.DPoS: https://bitshares.org/techn…Upcoming consensus-required upgrade on Steem: https://steemit.com/steem/@…We should be asking, what’s the point of a blockchain? If it’s to 1/ serve its community’s desires 2/ facilitate value transfer & stop double spends, then DPoS is the best solution in use today.

        1. Christan

          While it’s true it doesn’t make a difference to the chain working it makes the entire idea of steem less appealing to at least me. I stopped using steem when this happened. Considering it needs users they should have thought about this.

          1. Ned Scott

            Maybe this makes a point. If ETH or BTC were to switch to PoS, would they lose their “users”?I believe it’s worth recognizing Steem (because of its social network properties) is going for another audience entirely — the non-mining, not-necessarily-technical folk. However, I believe the community has made an overture to the “user-miners” by incorporating the equihash mining algo in the upgrade taking place 12/6/16.https://steemit.com/steem/@

  2. Twain Twain

    Re “mining is not the best consensus system out there,” ever since there have been Republics and democracies, consensus has been achieved by voting mechanisms.In the Senate, in the gladiator arena and indeed on Facebook, it’s a thumbs up or a thumbs down.There are three fundamental problems for Blockchain:(1.) Mining is not as democratic as assumed because of unequal access to the tools of mining. It’s fine if you have the hardware and the software skills.(2.) The assumption that the objective of coins/tokens accumulation drives consensus.Well, lots of people agree that murder is evil and we don’t need financial incentives to agree this.(3.) Behavior is treated as if it is “proof” of social permissioning and its drivers when that may NOT be the case:* https://techcrunch.com/2016…We’ll need to question if externally-focussed objectives (racing to mine Bitcoins) is proof of the internal subjectives involved in consensus and assessment of “fair” value.My instinct is that, over the next decade, there’ll be new forms of maths (and thereby computing) that can model “proof of value” better than what Satoshi proposed.

    1. William Mougayar

      What would that proof of value entail? Anywhere we can read about it?And to your prior points, I think long term, miners would have less influence on blockchain governance. Guaranteeing the validity of transactions is one thing, but having an absolute say in its governance is another thing. That’s why Bitcoin’s current governance leaves a lot to be desired, since it’s in the hands of miners who are really directed by the core developers.

      1. pointsnfigures

        Value is messy. Look at the difference between normative and positive economics http://www.investopedia.com

        1. Twain Twain

          That article says: “For example, the statement, “government should provide basic healthcare to all citizens” is a normative economic statement. There is no way to prove whether government “should” provide healthcare.””Should” isn’t normative when historical data is collected about healthcare provision and their correlations with the reduction of various diseases (cancer, heart disease, diabetes) and predictions can be made.For example, “The government should provide healthcare to all citizens because between 2010 and 2015, there were X% less of diabetes cases.”The problem with definitions of values and fairness is that it’s assumed current mathematics offers precise tools to represent and model them.There are 2 parts to measuring value:(1.) Quanta — price, premiums, rates, ratios, risk probabilities etc.(2.) Qualia — personal objectives and purpose, perceptions, cultural references.It isn’t the case that current mathematics can represent the entirety of both quanta and qualia.@wmoug:disqus — The need for “proof of value(s)” solutions applies not only in Bitcoin-Blockchain but, clearly, in democratic elections.It has to be a multi-disciplinary effort.

      2. Twain Twain

        In recent months there have been discussions about FB and Google “marking their own homework” and thereby that trust model being broken:* http://mashable.com/2016/11…The current issues of miners determining governance is a variation of that.We may find that Bitcoin-Blockchain is looping the same mistakes as the existing incumbent approach.

  3. jason wright

    the place of greatest mass and energy and light in the solar system is the sun, but there’s no life or evolution to be found there.

    1. Twain Twain

      What do you think about the possibility that the energy radiation of the sun catalyzed the reactions of various particles and sparticles and that’s how life and evolution happened? Thanks.

      1. jason wright

        Bitsun is that, and a place not too close and not too far away is where to look.

      2. Lawrence Brass

        I like the theory that states that subtle energy differentials is what starts life at a molecular level. In the linked publication the authors mention a day-night cycle caused by the sun that enables the primordial soup.Movement and flow is what sparks life… and value.http://phys.org/news/2015-0http://scitation.aip.org/co

  4. creative group

    CONTRIBUTORS:As an early adopter of Bitcoin and complete fool that dumped 20,000 Bitcoins in the wind at .0001 (Costly impatience) we realize the Proof of Stake can definitely work if supported but the Stake of interests by those that spent millions for computers , infrastructure, etc to gain 51℅ it will not be easy to gut that interest. (Asia, Asia, Russia)

  5. Tom Labus

    How would this work with banks and other financial institutions?

  6. pointsnfigures

    Proof of Stake also expands the network to entities that might not want to mine. Allows for more brokerage to happen-and allows for brokerage between blockchain networks.

  7. Dana Hoffer

    The record is one aspect. Identity and access are critical. The solution will be at hand when all three, the record, identities of those in the record, and access rights are resolved conclusively. This is not different than the current property system. The merits of what has worked need to be kept in view.

  8. Aidan McGinley

    Peercoin (http://peercoin.net) has been operating a proof of stake system since it’s launch in 2012

    1. fredwilson

      Yes. Do you think Peercoin has been stress tested enough to determine if POS will work?

      1. Benjamin Cordes

        PoS has been in production with Peercoin, Nxt, Bitshares and a few others since 2012/2013. The results are mixed, in terms of market success (all PoS coins trade well below their peaks at 100-300M$ from 2013-2014). PoS has some issues, but its very much worth studying. One advantage is that PoS is a much simpler system (a simple one can be written in 2000 lines), which reduces development effort significantly. PoS has allowed for a range of innovations, from new governance systems to integrated exchange systems. Likely these systems show also how scaling can be potentially solved – the verification of blocks is more deterministic in PoS than in Pow which advantageous. A system that solves scaling will likely find very rapid market adoption, so it would show up in the marketcap rankings quickly. PoS could be seen as an intermediate step to an entire new algorithm. It is however not decentral enough to beat PoW. Stake can be cryptographically proven, but it can’t be proven how it was distributed, and it can’t be ensured that there will not be strong stake monopolies.

        1. Embedded Thought

          I agree with what you’ve said up until “It is however not decentral enough to beat PoW”. I believe that PoW inevitably leads to higher levels of centralization in terms of network security. One benefit of Proof of Work is its method of coin distribution. I believe a hybrid network such as Peercoin will allow for a more decentralized network as time progresses. PoW used for continual distribution and PoS for network security.

  9. Mike Cautillo

    Though who’s to say mining will be resource intensive or geographically concentrated sometime in the future. This will change…it’s just not the top priority currently. The ecosystem elected to dedicate its resources elsewhere for now but I believe this stone will not be left unturned because it is not the most efficient means. I anticipate a decentralization of mining awaits in more ways than one. Fitting isn’t it.

    1. fredwilson

      Great point.

  10. Sebastian Wain

    It is important to understand that the security robustness of Bitcoin is below 51%. With Selfish Mining it goes below 40%, speculated to be at ~33% and much lower in Ethereum.

  11. OWOMAHN

    PoS will be very relevant for blockchain on mobile devices..Gamerholic.com uses a PoW/PoS hybrid for their Gamerholic coin launched two years ago.Surprised PoS is just now getting headlines.On Gamerholic, you get Gamerholic coin for your gaming content.

  12. Jake Baker

    Fred:I’m really grateful when you discuss relatively edge case or cutting edge technical topics like this. The feedback and discussion, though fewer people participate, are almost always incredibly well-informed and interesting.In this case, I don’t have much to add topically other than to soak up knowledge and learn, but I wanted to more visibly express my gratitude for the post and the discussion.In particular, if comments on AVC ever disappeared, I think the loss of this kind of discussion and community would be a real shame. Yes there are sub-reddits and other communities, but there is something about the nature of a relatively consistent group of people commenting over time across many topics that seems highly valuable, unusual, and worth fighting for.Thank you!

    1. fredwilson

      Thank you Jake

  13. Pete Griffiths

    We can do better.I’ve had discussions on this site before about people confusing the idea of distributed truth with the mechanism to validate and the issues with mining. It’s good to see this issue is getting clearer.

    1. Twain Twain

      Spot on, Pete.Truth defined by mathematicians is different from truth defined by artists. What happens when we put Magritte upside down? Is it still a pipe or a dream?https://uploads.disquscdn.c…What was the “truth” during Election 2016?https://uploads.disquscdn.c…How did Descartes and Socrates define validity?https://uploads.disquscdn.c…What are the mechanisms of mining and hashing?https://uploads.disquscdn.c…Clearly, mining (of and in itself) is not an effective process for validating “truth”.

      1. Pete Griffiths

        Love these diagrams!

  14. Chris DeRose

    Proof of stake is a fools pursuit. The proponents have no cognizance of the science that pre-dated this goal, and are trying to reinvent that knowledge with expensive and fruitless ponzi schemes. Here’s a better guide to the subject matter, for the academically inclined: https://download.wpsoftware

    1. Twain Twain

      Pls see my comment on differences between truth, validity and the process of mining (hashing).

      1. Chris DeRose

        If you believe that innovations in mathematics are the bottleneck, no amount of funding in proto-equity pitches are likely to produce this science.The solution to the Byzantine general problem was an economic innovation, not a computer science innovation. Mining is the process of converting registered value to fungible value. It is nearly 100% efficient (depending on how you qualify seniorage costs) This realization has not been absorbed by the PoS community. The universe does not owe us free consensus, nor does this pursuit differ substantially from any run-of-the-mill perpetual motion pursuit.

        1. Twain Twain

          The Byzantine Generals problem of 1982 is a variant of Nash’s bargaining problem of 1950 which required one of the most brilliant mathematicians to construct his theory of distributive justice between two players optimising for their preferred outcome and to write a mathematical proof for it.Economics and computer science are both contingent upon mathematical tools innovation. We invented 1, 2, 3, algebra and “X” and that enabled us to represent and calculate things in ways that Roman numerals V, X, C, M couldn’t.* https://m.youtube.com/watch…There is a difference between mining efficiency and people effectiveness. The first is about rational optimization of objective functions within the Merkle tree. The latter is about modeling the complexity of the irrational (humans) towards consensus.Whether it is Nash or the Byzantine Generals problem, we have no mathematical constructs for free will and the computers have no bases for perceptions, emotions and other variables that inform WHY we agree or disagree on something.Merkle trees have a binary base. Binary is not suited to recording states of human subjectivity adequately.So … it is just about as big a maths problem as possible.

          1. ben

            “mining” is simply not going to cut it guys, move on. It’s non-democratic and environmentally unfriendly 🙂

          2. Twain Twain

            Haha, quite!Knowing what we do about mining, nuclear fission-fusion and solar energy, it seems somewhat silly to think that mining is the best method.

          3. Chris DeRose

            I don’t know what it means to cut something in this context. The underserved users of blockchain don’t care about the environment, nor should they. Externality management is the responsibility of governments, and if government didn’t want bitcoin to produce environmental impacts they could switch the bitcoin network off in a heartbeat – by making drugs and gambling illegal. As for non-democratic – that’s what you want. No blockchain users care about that either.

          4. Chris DeRose

            Well, if you feel that way, you can understand why I’m not particularly impressed by crowdfund ponzi schemes to fund such goals. Could you crowdfund an einstein?

          5. Twain Twain

            I completely hear you, Chris. The internal motivations of the Einsteins is different from the objectives of investors.It’s not only challenging to crowd-fund the Einsteins, it would also be almost impossible to mentor, set milestones and enforce limitations on what and how Einstein should think.What do you propose instead of PoS and PoW?

          6. Chris DeRose

            mysql-chain. As you can tell by the efficiency and immutability of discuss, it’s a very efficient alternative to Blockchain.

  15. David Siegel

    My main concern is the Matthew Effect: That those nodes with more cash will then get more cash rewards, and those with less cash will get fewer cash awards. This could end with only a few (or one?) nodes owning everything, as it is in your best interest to get big in this ecosystem. The value of your stake increases nonlinearly as your proportion of the total pie grows. Smaller stakeholders could possibly get a better deal by selling to larger stakeholders. Will be interesting to watch.

  16. Preston Byrne

    Proof-of-stake is already widely used in private/permissioned blockchain implementations, where PoW is inefficient and isn’t friendly if you want to do high transaction throughput. I would definitely recommend checking out Tendermint, a proof-of-stake protocol we use as a consensus engine for our blockchain client. It’s worked great for us.

  17. ZekeV

    PoS might run a consistent blockchain for periods of time, but ultimately does not offer the same guarantees of security or integrity as PoW. Aside from the subtle CS arguments which I admit hurt my head a little, I guess my biggest concern is that a large stakeholder cannot be unseated involuntarily.A PoS based coin is dependent on the nature of its largest stakeholder(s) to remain free from censorship, and that is particularly the case where the largest stakeholder is a known entity such as a developer foundation that can be subjected to pressure. In a PoW based coin, it is also possible for a large miner or group to misbehave, but the large miner can be unseated or the group splintered through economic activity (and of course, by violence, though I hope that never happens).I would also love to replace PoW, to avoid the seemingly wasteful process for selecting validators. But it is interesting to note that difficulty correlates with price over time, so there does seem to be a commensurate investment in security of the network as the price rises. This incentive doesn’t hold in the same way for PoS, where we have to trust a relatively small and semi-permanent group to promote security / integrity of the network. Maybe PoW is not such a big waste, after all.

  18. Eddie J

    I think Bitcoin will be first to implement POS with segregated witness. It’s not fully realized yet but has a head start over Ethereum. Side chains, would allow proof of state on top of Bitcoin.

  19. MaxKK

    There are many PoS cryptocurrencies out in the markets. For example Nxt which soon celebrates its third birthday. It’s not a new thing.There are also variations of PoS, for example DPoS (Delegated Proof of Stake) which is even more lean and energy efficient, faster, has more consistent block times and is perfect for a large cluster of sidechains.Lisk is utilising it and will launch this consensus algorithm within a week.More infos:Nxt.orgLisk.io

  20. bernardlunn

    A bit off subject, but I struggled to decide on whether to focus on the conversation here or on Hacker News.

  21. bernardlunn

    Proof Of Stake tackles scale and cost. Perhaps Lightning Network or Open Transactions are better approaches?

  22. John C. Vernaleo

    Not to advertise something I’m involved in but we’ve been doing a mix of PoW and PoS in Decred (https://decred.org) since our launch. I’m pretty happy with how the network has looked with that mix.

  23. Mark Essel

    Dig that there’s motivation for a lower cost trust system. Will keep an eye out for interesting cryptocurrency solutions over the next few years.

  24. Lawrence Brass

    I am absolutely ignorant about why blockchains need a “proof” on top of the underlying Merkle tree, more than the obvious “makes them harder to break”. Why the Paxos algorithm don’t need an expensive computation?I am writing my 2017 wish list, and asking these questions to Leslie Lampert is at the top.

  25. Kevin Hill

    I think this is the best post you’ve done on blockchain technology.I agree that proof-of-work is the main barrier for this tech in changing the world. To your list of ‘cons’ I’d add that the need for the proof-of-work to not be too easy is one of the main reasons why transaction times need to be fairly high.I also wonder if there is active discussion around other forms of distributed security. One thing that always struck me as a good possibility with these distributed systems would be some sort of automatic-insurance auction similar to ad-words. Basically, you’d have an automatic bidding process where the person giving the coins would pay a little bit extra to have their transaction insured by a 3rd party. An added benefit of such a system would probably be a reputation-like system that could be useful for things like troll detection.

  26. Muneeb Ali

    The biggest problem with proof-of-stake is that the older data/transactions secured by the blockchain are no longer secured by proof-of-work i.e., an attacker can come up with many different histories without much computational cost.The biggest benefit of blockchains is that history cannot be tampered with. It’s cheap to tamper with history with proof-of-stake, which is a huge issue in my view.Also, a 51% attack on proof-of-work is very different from 51% attack on proof-of-stake. A 51% attack on a proof-of-work blockchain is a temporary attack (only impacts current blocks — it does not impact the past or the future). A 51% attack on proof-of-stake is permanent.

    1. moeadham

      Great points Muneeb. What do you think about economic incentives?For PoW, if a bad actor wants to attack a blockchain that is of minority size and uses GPU’s, their investment is not totally lost; They can likely point their GPU investment at another blockchain and amortize, making the attack “cheaper”.In PoW, the economic incentive appears to be more scortched-earth. After you attack the chain, the chain becomes valueless, and you lose 100% of your attack costs; Could a 51% attack on proof-of-stake be permanent, but more expensive?.This is by no means a promotion of PoS, but I do fear the explosion of new PoW chains that use commodity mining hardware.

      1. Muneeb Ali

        I see the economic incentive there but need to think more about how game theory will work out in these scenarios. Also, you can get to 51% by a hack of private keys as well ($0 of your own money spent). Kind of like mining with hacked computers (again temporary vs. permanent).I’m frankly not that concerned about “wasting electricity”. Datacenters cost so much money & electricity anyway and mining is a very small use case of datacenters right now. Also, investment in mining can be reused over many chains (as you mentioned).

    2. Ned Scott

      This is not the case when Transactions as Proof of Stake (TaPoS) is employed in conjunction with other PoS, such as DPoS.https://steemit.com/bitshar

      1. Muneeb Ali

        Not sure if I buy that argument. You can just replay the transactions with different hashes retroactively and construct thousands of alternate blockchains. What’s the limiting factor? The CPU-time required to construct alternate blockchains is pretty small.

  27. Evan Van Ness

    PoW mining is akin to buying equity. PoS “mining” is akin to buying a low liquidity bond.A PoW miner must front substantial capital for hardware, then hope that the MR (blocks mined) > MC (electricity costs) for long enough to get an ROI. It is pretty risky. Meanwhile, PoS has much lower hardware requirements, as well as lower electricity costs. It has much lower risk, and will likely pay out X percent per year, but it will take some time to get your capital back when you want to “sell the bond” (stop mining).It remains to be seen how network security PoS will provide when interest rates return to historic norms. This seems to be an underexplored topic, as the proposals I’ve seen call for static issuance rates

  28. bernardlunn

    Like that analogy