Collaborative Proof of Work (PoCW) which I published its idea a while ago was a preliminary example of how it is possible to have POW that is not based on winner-takes-all. I have merged the core idea with a scaling solution (based on hierarchical sharding) and it is working (conceptually speaking) fine. It is not about ASICs, has nothing to do with ASICs, it is about pools and mining variance.
can you provide a link to your paper on "collaborative proof of work"?
That proposal is abandoned by me because, in the process of implementation, I realized that there are opportunities for taking care of the scaling problem as well and surprisingly this is how we can implement the core idea of PoCW. It is exemplary as the first alternative to Nakamoto's winner-takes-all PoW design, as far as I'm aware.
In PoCW miners compete for transaction fees in the first phase and the block that has won the race is chosen by subsequent shares that are generated by miners committing to it more and more.
In the original design, there was a finalization phase in which rewards should have been distributed, later I found a way to eliminate this phase by leaving this duty to the next block. I didn't edit the original article for historical reasons neither published a new version. The whole idea is now merged into a new work.
you can check this link for details:
https://bitcointalksearch.org/topic/getting-rid-of-pools-proof-of-collaborative-work-4438334you're speaking in (purposefully?) vague terms. i don't think you'll get much support without at least describing the sort of technical changes you're proposing.
BTW, It is not vague, it is just about the big picture:
tackling centralization and scaling problems at the same time. We are used to thinking about each problem separately or worse, as a challenge for solving the other one. What if they both are phenomenons of the same
flaw in bitcoin? Above thread I have introduced this flaw for the first time:
winner-takes-all.
we think that way because changes imply trade-offs.
No, they don't! It is true for super matured technologies or basic nature rules but not for bitcoin. Winner-takes-all was a wrong choice from the first day, Satoshi didn't care because he was not ready for developing a system that is supposed to live forever, imo, he was just releasing a proof of concept and you don't wrap a concept in a sophisticated multi-layer architecture like what we have to do for a collaborator-takes-share model.
There is no reason behind such a claim, changes imply trade-offs, just analogies, and presumptions about how GREAT bitcoin is.
To be clear: bitcoin is great but it is in its infancy, the fact that this system is so exciting shouldn't be an excuse for manipulating minds about trilemmas and trade-offs. You can't tell a child to do a trade-off and choose between growth in his knowledge and his brain or body.
let's take your "hierarchical sharding" idea as an example. can you elaborate on that? because the consensus is that sharding is highly centralizing since it drastically reduces network redundancy.
They "say" that sharding has security implications because of the distribution of hash power but what if there is a workaround for this threat or other issues? Hierarchically sharding the bitcoin machine state is one such workaround. Now they may suggest centralization threat, yet there may be a workaround for this, actually, there is:
Suppose we have segmented the UTXO (IOW, bitcoin machine state) in just two layers a top layer (full state) and two odd and even segments. High-end full nodes may choose to maintain the state in the top layer and machines with less power should stay on either odd or even segments but does it lead to centralization of mining (we will be back to the centralization of full nodes later)? No, absolutely not. We could simply ask miners in any layer to do the same amount of work to prove their blocks, couldn't we?
Now, let's take a closer look at our simple 2 layers hierarchical sharding model and become even more surprised by figuring out that mining is not centralized and on the contrary, we have a more decentralized system! How? Just focus:
In our naive model, the difficulty is distributed between 3 segments and in each round, we are introducing 3 blocks with less merit to the network, these blocks cumulatively have the target difficulty, tho. Yes? But lower difficulty means better distribution of luck and it is what we desperately looking for to keep the network decentralized: low variance, less pooling pressure, less need for pools.
Interestingly, we found ourselves in PoCW territory again. Defining a block generation round as being broken to small parts and accumulating the work done is the key to Collaborative Proof of Work. Only in the new model, the transaction space is partitioned and work is done more efficiently and in a smart way.
I'm aware that it is not a formal description or a proposal but it is more than enough for the purpose of this topic to convince the reader about one simple argument: Decentralization and scaling problems are neither paradoxical nor even substantially different problems and have one basic solution: Distribute the work!
As of fears about full-nodes (not mining) becoming centralized, one should be absolutely pragmatic:
1- We have UTXO commitment and pruning potentials for bitcoin, currently not unleashed, that can reduce cots of maintaining a full-node drastically but we have absolutely no idea to fix centralization of mining and getting rid of pools other than PoCW and this new version we are discussing here. So, the whole full-node issue is not on our priority list, actually, it is not a problem at all.
2- Costs does not grow linearly with UTXO size, anyway.
3- Bitcoin is and remains permissionless and it is up to the user to choose the level she wants her node to take part.