Topic: Proof-of-work problems designed to be solved by the reversible computers (Read 1949 times)

@op you'd be much better off reading the #trilema logs for six months, then humbly asking them your questions should they remain.

http://btcbase.org/log/

The premine is usually expressed in % of total coins, which you own from the beginning.

There are also other options; for example, Zcash implemented a 20% 'founders' tax', where for every ZEC mined 20% goes to the founders and the remainder 80% to the miners, for the first 4 years. Generally, Zcash's idea was not well received and different forks emerged, like Zclassic with no founders' tax. Another idea is to have the premine locked for a certain period of time, say 6 months or 1 year. This is to reassure miners/investors that the developers will continue developing the coin, before accessing the premined funds.

My opinion is that the premine should be as low as possible. Maybe 0.5% or 0.75% of total coins would be enough. The idea would be to use it for development but also for promotion ie. bounties.

Edit: I can help you setup a bounty campaign if you want.

So with the recent SEC ruling on ICOs in the USA, I am unwilling to do an ICO for Nebula. I now agree that the best way to fund Nebula will be to do a premine so that I develop enough funds to run Nebula (by premine I mean I reward some Nebula to myself as soon as it is launched). Would 3-4 months worth of Nebula mining be a good premine or should I do an amount lower or higher?

I will call the upcoming cryptocurrency Nebula. Here is the whitepaper for Nebula https://boolesrings.org/jvanname/2017/07/22/nebula-the-cryptocurrency-that-will-produce-the-reversible-computer/. The whitepaper focuses on the RCO-POW problem R5 since R5 will be the main distinguishing feature between Nebula and other cryptocurrencies, and the paper does not discuss other things about Nebula (such as block size, halving time, masternodes. . .) since those factors have not yet been determined. Please let me know what features you would like Nebula to have (for example, do you want masternodes? Do you want a new block to be produced on average every 2 minutes?). Further discussions about Nebula can be found at https://bitcointalksearch.org/topic/nebula-the-upcoming-cryptocurrency-that-will-incentivize-reversible-computation-2041169.

One can mine Nebula using CPUs/GPUs until mining power is controlled by ASICs. Since R5 consists of five different algorithms, one can probably get away with mining with CPUs/GPUs for a while (Problems 4-5 are somewhat ASIC resistant mainly because they change over time).

I am currently working on Nebula by myself in my spare time, so it is difficult for me to determine how much corporate funding will be appropriate for Nebula (and since I am an individual and not a non-profit, I do not know if corporations will be willing to fund R5).  However, from a corporate funding will help with the following costs as Nebula gets launched (I think corporate funding for one to two years after launch would be more than sufficient):

1. It would be advantageous to write a paper on the cryptographic security of R5 even though R5 seems to be very secure.

2. Funding to launch an ICO? (I am unsure if an ICO would be a good idea since Nebula should mint new coins by mining instead of through another process.)

3. Advertising for Nebula.

4. Developers salary.

Yes, I would also want to see the whitepaper. Do you have any idea on when you re planning to release it?
Thanks

I would like to get a copy of your paper about R5 (when it is ready), if that's ok.

How much "corporate" investment are you looking for?  Or are you saying you want a "corporation" to embrace your approach and invest in making it come to life?

You can setup a small premine, and use it to fund development and maybe setup a bounty campaign as well. I think the most important part is to make enough people aware of the coin launch, so that it is a fair launch and people will have an incentive to mine.

How will this R5 algorithm be mineable? I mean, apart from the RCO advantage, will it be mined with GPU/CPU?

I can probably launch the cryptocurrency without funding as well and hope that it does not flop. My concern is that if the cryptocurrency is funded through an ICO or through some other method, I can get advertising for the cryptocurrency, and the ICO (or other funding method) will give the cryptocurrency a positive initial market cap. Furthermore, funding can help with the early-stage development. But if you insist, I can probably just launch the cryptocurrency without funding and let it grow organically (and hope nobody attacks it when it still has a low market cap).

So out of all cryptocurrencies, it seems to me like Dash has the best model of governance, funding, ease-of-use, etc so I will probably just take Dash's source code and modify the POW to incorporate R5 as its POW.

What funding would you need? If you are close to finishing the design of the algorithm, you can simply launch the currency and see how it goes.

Reversible computation is more difficult than conventional computation since it usually takes more steps to compute something reversibly than it takes to compute it on a conventional machine. The algorithms in the RCO-POW problem R5 that I have in mind are lightweight and they can be computed about as efficiently as a standard cryptographic hash function. However, it takes just as many steps for a reversible device to compute R5 as it takes for a conventional device to compute R5. Therefore even if a reversible device were only 10% more efficient than a conventional device for solving R5, one would still prefer to use a reversible device for solving R5 since R5 does not give you any computational overhead when you solve it with a reversible device. Remember that the energy efficiency of conventional computing is limited by Landauer's principle and that we are beginning to see the effects of Landauer's principle as Moore's law is now coming to an end. The energy efficiency of reversible computation on the other hand is not limited by Landauer's principle (we will hit Landauer's limit in the 2030's so it is now necessary to start producing reversible computing devices).

-Joseph Van Name Ph.D.
boolesrings.org/jvanname

A reversible-friendly algorithm, e.g. not SHA-256, might always be simple to implement irreversibly and achieve both better speed and reasonable energy consumption.

loremipsum-I think that the cost of developing and mass producing an RCO-POW device will probably be hundreds of millions of dollars or even billions of dollars. However, since we do not have reversible computing devices, we do not know what technologies will be best for constructing reversible computing devices, and because I am a mathematician and not an engineer, I am unable to make an accurate estimate about how much it will cost to mass produce this future technology. Let me quote Mike Frank at https://intelligence.org/2014/01/31/mike-frank-on-reversible-computing/ about the difficulty of manufacturing a reversible computing device whose efficiency surpasses Landauer's limit.


It will only be practical to develop these reversible devices if RCO-POW problems take up a significant share of the cryptocurrency mining market. Of course, we do not need to make reversible devices that go beyond Landauer's limit just yet to mine cryptocurrencies. We currently only need to develop reversible devices which are more efficient than conventional devices in order to solve these RCO-POW problems.

I am not worried at all about the cryptographic security of RCO-POW problems since these problems can be made just as secure as SHA-256 simply by adding more rounds to the reversible circuit. Of course, there is always the threat of a 51 percent attack, but the use of 5 different algorithms in the POW will provide the decentralization needed to thwart 51 percent attacks. I currently do not have any funding to develop a cryptocurrency with an RCO-POW problem so I will need a modest amount of funding to launch the cryptocurrency in the first place. Even if I use an ICO, I will definitely need funding to set up the ICO in the first place.

David Rabahy-While all algorithms including SHA-256 can be made reversible, it in general takes more steps to run a computation reversibly than it takes to run the computation irreversibly. Furthermore, after one is done computing SHA-256, one produces much garbage information. In order to clean this information, one either has to uncompute or delete the garbage information (and deleting garbage information after computation defeats the purpose of reversible computation in the first place). I want an RCO-POW problem for cryptocurrencies since these problems will make it as easy as possible for reversible devices to solve. If it is not easy for a reversible device to solve a POW problem, then it is unlikely that such a reversible device will be developed in the first place for a very long time.

A possible problem with using a hybrid of reversibility and irreversiblity is that it may be difficult to make reversible technologies compatible with conventional computing technologies (since reversible computers do not exist, we do not know what technologies will be used to make reversible devices and these technologies could be anything from adiabatic circuits to quantum dot cellular automata to even nano-mechanical devices). Cryptocurrency POW problems should be used to advance new technologies instead of stockpiling current technologies and burning energy.

I do not think an established cryptocurrency like Bitcoin would want to change its POW to an RCO-POW until other cryptocurrencies are using RCO-POW problems.

Joseph Van Name Ph.D.
boolesrings.org/jvanname

Hmm, it appears there's been a misunderstanding.  Reverse engineering and reversible computing are very different things.  One cannot take a non-reversible device and add something around it to make it reversible.  One must reengineer a circuit from the gates on the chips up to make it reversible.

yeah but if they are using private chip like for example in the baikal miners, you can-t get anything out of reverse engineering which is the same as recersible computer to me, you can build the board and anything but the chip is what matter

Any digital circuit can be made reversible using components like the Fredkin and Toffoli gates.  This includes SHA-256.  As much fun as it would be, there is no fundamental need to develop a new algorithm for the Bitcoin PoW.  It would truly be interesting to engineer a reversible SHA-256 computing device and see how it compares in speed vs. energy consumption against the current generation of non-reversible ASICs.  In practice only the hottest portions of the circuit benefit from being made reversible.  A hybrid circuit of both reversible and non-reversible sections is likely to be the one that gives the best trade-off of speed and efficiency.

How do you think a corporation can be involved in this project? I mean, the coin has to launch first, people need to start mining it and the security of the RCO-POW algorithm has to be tested. At which point can a corporation be involved in this?

My thinking is that there are 2 options:

1) You can launch the coin as most POW coins launch ie. launch it and market it as a new coin that has a novel POW algorithm. You can keep a certain premine for your work and also bounties for promotion. It can be a successful coin, as it will have a novel POW algo that gives incentives for RCO mining. If you set the other parameters well, like block rewards, difficulty algo and other features (for example masternodes) it can easily be successful. Only after the coin's marketcap has grown a lot, will a company be interested in developing a reversible asic.

2) You can do an ICO. This is very weird for a POW coin but this case is different. You can use the ICO money to fund the development of a reversible asic. The problem is how to convince people to fund your ICO, as they will not get any immediate returns.

What is the budget we are talking about for research & development of an RCO device?

The idea behind RCO-POWs is to give a mining advantage to the reversible devices which do not exist yet so that businesses manufacture reversible devices without having to make their reversible devices so efficient as to compensate for the overhead that comes with reversible computation. Even though these problems are meant to be solved by reversible devices, one could still solve these problems using CPUs/GPUs. Since reversible computers currently do not exist, the mining advantage will go to the conventional ASICs once they get built (one of the five problems in R5 is a problem that changes every year and the other problem changes slightly every new block so these problems will be slightly ASIC resistant) and then only when reversible devices come into existence will the advantage go to reversible devices. Here is the hierarchy of computational devices for solving R5 or more generally for any RCO-POW problem that I can imagine.

CPU
I suggested corporate investment since it seems like one of the ways that such a project could be funded. I will only accept corporate investment only from corporations who are interested in manufacturing reversible devices though (and I will never accept a corporate investment from Google or Microsoft because Google and Microsoft spread lies and slander people. Yes. I have a personal grudge against Google and Microsoft and I hope they both go under as soon as possible). These corporations such as IBM have the greatest incentive to ensure the endurance of RCO-POW problems so that they can eventually manufacture and sell these devices. What other way besides corporate investment of obtaining funding is suitable for RCO-POW problems?

-Joseph Van Name

Im not sure I understand this. Would mining using the R5 algorithm be possible only on reversible computers or on any CPU/GPU? Is the idea to offer a mining advantage on running the algorithm on reversible devices or to have it ran solely on these?

Also, why seek corporate investment? I think a lot of people in this community won't like that. If there is a mining advantage in using an RCO device, at some point, if the coin marketcap is big enough, an RCO device will be generated.

Right now I have half-way written the program in C for the reversible functions f for the RCO-POW algorithm which I call R5, and I am writing the whitepaper describing R5 and a draft should be out in a week or so. After I am done with the paper and initial design for R5, I think that the best course of action would be to seek corporate investment for R5 if possible. Computational device manufacturing corporations should be willing to invest in R5 (their return on investment will be that they will get to sell reversible computing devices to miners) or a variant thereof in order to eventually develop reversible computational devices.

I am not too concerned about initially setting up the R5 RCO-POW algorithm, but I am more concerned about how well RCO-POW problems will thrive in the cryptocurrency community in the long-term. Since efficient reversible computers do not currently exist on the market, these reversible devices are probably very difficult and very expensive to construct especially as one eventually wants reversible computers whose efficiency in practice exceeds Landauer's limit. The use of RCO-POW problems will therefore only be effective if the market for RCO-POW cryptocurrencies is large enough to cover the R-D costs of initially developing these reversible devices.

Any way that you or anyone else can help get R5 or any other RCO-POW algorithm a modest corporate investment would be great (I am quite new to cryptocurrencies so I am just taking things one step at a time and hoping I am taking the right steps).

I will post an update once I get the whitepaper for R5 out.

-Joseph Van Name Ph.D.
boolesrings.org/jvanname

It will take a long time to write a new RCO-POW algo from scratch I guess. I'm not a developer but let me know if there are other ways I can help you set it up. Its a great idea if it works!