I’m curious if anyone has done some comparison of Sharding with Close Group Consensus? It sounds like there is some significant crossover with Disjoint Groups and Data Chains.
a sharding scheme on Ethereum might put all addresses starting with 0x00 into one shard, all addresses starting with 0x01 into another shard, etc. In the simplest form of sharding, each shard also has its own transaction history, and the effect of transactions in some shard k are limited to the state of shard k.
edit: for those with a reading addiction:
notes on the development of this feature in EIP 105
Nice find, I will certainly digest this doc. Skimmed it but i can see where you are coming from, very similar issues in many ways. I see Ethereum also would like to do POS to make sharding easier, but have also some big issues to solve such as colluding oracles etc. IIRC (see the video of Vlad posted here recently) which is good to see. I think they are getting closer to us in many ways, but from different angles and this is good news. We need many projects thinking about this kind of thing and I am pretty sure we have much to offer each other as we progress.
I would like us to take some Engineering time after alpha 3 (probably 4 as that will include data) to really dive into the sharding proposals, also POS to figure out the relationship there and why POW is more difficult for this proposal. In any case this is a good find and I believe worthwhile.
Ethereum are on a route to seeing a 1TB+ blockchain, so are thinking about ways of solving that. Blockchains are used for a variety of projects, one project is not the same as another.
Ethereum is also looking to create a decentralized internet, like maidsafe, but with different properties. I think there will be advantages and disadvantages to both and that some apps will work better on one or another platform. Even for ethereum a lot of stuff is still just on the planning stage and not on the roadmap until 2020+.
Both are working on their own browsers, in the future I think it would make sense to have one browser for the decentralized web supporting both.
I was also looking at a few other projects which also appeared to have different solutions relevant to this aspect. (to me anyway as a non-techie). I found reading the “Byteball” whitepaper interesting in particular, along with “Iota”.
These projects use a directed acyclic graph, or “DAG”
What I found interesting is that Dan Larimer has discounted this method as inefficient/unworkable for fast transactions. He is of the opinion that measuring/voting on changing “states” is not the way to go - his preference is for using messages (I only vaguely grasp this concept)…This is apparently not possible on most blockchains, however is on his Graphene based chains (capable of 100k tx per sec).
His approach is now to create “EOS” which will use parallel (or horizontal…not sure) blockchains on top.
To my mind, a number of projects are now chasing the same goals here…Maidsafe, EOS, Etherium etc. This routing issue is crucial to success here I think. I would definitely spend some tech person’s time researching these other projects, as I believe some form of hybrid solution .may be possible.
Not sure if any of this is useful, just seemed very relevant to me conceptually.
Very interesting to hear Vlad say that “it’s quite realistic that we might see the first version of sharding in a couple of years”. I’d say that gives you your answer
I completely agree with the vids, the crypto community really needs to wise up to how far Eth still has to go and how hard their challenges really are. Their choice to start with smart contracts and then try to make the blockchain square fit in the round, scalable hole gave them a kind of first-mover advantage on usable tech in the wild while it was small, but things will continue to get progressively more difficult for them as they grow. The high price and pressure might cause them as many problems as it has afforded them advantages if they start to creak under the weight of expectation and investment.
Tangle, data chains, sharding and graphene are all very interesting I agree. I think David took the right approach though; starting from the bottom up and thinking holistically about design rather than piecemeal additions to a single, seminal innovation. Slow and steady wins the race.
Yes, I’ve never held Ethereum and think is way, way overpriced. Personally I’ve narrowed the eventual “winners” in this crypto space to be based around either Safenet or EOS. Unfortunately, I’ve got to buy Ethereum to get EOS though…
I have some Ether … didn’t get in on the big run up, but I think they are undervalued relative to bitcoin. the next five years are going to see a big shake up of crypto’s IMO.
I’d love to see more collaboration between groups like MAIDSAFE and Ethereum. Given that they are both working on similar, if not identical, underlying problems progress could be swifter (ultimately) and a large market-share advantage could be gained.
Also given that Ethereum will probably have to go through some serious changes (hard forking?) in the future, why not just take advantage of SafeNet instead of reinventing the wheel? There might be big advantages for both projects to collaborate.
This last video in the series with Vlad is specifically about sharding and it really shows how much work they still have to go with the idea… it’s significantly less mature than the close group consensus mechanism on the safe network.
Perhaps that’s because safe starts with extreme scalability and aims to mix in robustness via data chains, whereas ethereum starts with extreme robustness (hmmm) and aims to mix in scalability with sharding. So their progress on this specific feature will naturally seem less than safe.
Ethereum started on an easier problem than maidsafe, make a better blockchain and then try to solve the more difficult challenges later. Maidsafe didn’t start with a blockchain so they were forced to start on the more difficult challenges immediately.
The advantage ethereum has now is more developer mindshare and hype. They now have third party companies like brainbot technologies working on core tech to help ethereum scale (raiden/lighting network). The vast majority of developers is working on ICOs and various apps though, many that could run on SAFE if it was out and once even a permanent SAFE testnet is out I expect we’ll start seeing many more developers trying out SAFE.
What if Ethereum turns up late delivering sharding, and is not 2 years away from solving the scalability issue like Vitalik thinks, but 5 years away ?
And what if in the meantime, Maidsafe delivers everything they promised, within 6-9 months ?
Wouldn’t it be ironic ? My experience is that very often, the market is wrong. Contrarian investors that go according to the value and not the hype can make a killing. Maybe Maidsafe will be early and Ethereum will end up solving scalability late, if ever they manage to.
In my opinion, the real question is at which point do we know that Maidsafe will deliver for sure ? When do we know that the hardest part is behind us and it just details from now onward ?
Is it after datachain implementation ?
Depends on which critics you listen to. For some SAFE can’t be a ‘sure thing’ until it has survived a few years in the wild with the economics and security being tested in practice. Personally, I’ll be happy once I see data chains, data republish and testcoins working as we hope. I feel fairly comfortable with the economics of it and I’m sold on things like the viability of ‘pay-once-store-forever’, so I think that probably puts me in an optimistic minority.
Great post and exactly my thoughts. There ain’t no fat lady singing yet!
IIRC, David Irvine is on record saying that all the key concepts have been proven in the test nets. It is all about maturing the solutions for delivery.
Safe net has so many good foundational concepts that it makes many a blockchain solution look cobbled together. They may work as POCs, but are they of genuine value in the long term?
Of course, safe net still needs to prove itself resilient in the wild, but it is the right forward looking solution, IMO.
Also, Dash’s ‘Evolution’ is looking to do the same thing as Maidsafe with account/pw stored on the network. So will be interesting to see how they approach the issue of limiting account creation. Couldn’t find any info on how they are planning to do it.
This is what a lot of folks miss in the crypto community, they think MaidSafe is still working on some simple p2p-storage implementation, missing the fact that there are problems being solved here. The implementation of Disjoint Groups was one of them. And quite an underestimated one if you ask me. The old idea would’ve probably worked as well with each node having it’s own perspective of the network with 32 close nodes. But DG removes a lot of complexity, like who’s gonna store what and what if that node (and its own view of the system) is gone. Let the scaling begin at Alpha 3 ;-).
Build a functional Sharding with Blockchain is like build a skyscraper with the shallow foundations of a nice vacational house.
Good look, but If just increasing the number of TPS to a few hundred are costing years of work (and we are still waiting). Creating a truly functional sharding system can cost decades (if ever they get it).
