Given the upcoming t-ecdsa functionality going beta, and the HTTP interface enabling using canister signatures( Tackling CertifiedData in Motoko - #32 by bjoern), the restriction on canisters not being able to hold neurons is a merely an inconvenience.
A good bit of code needs to be written to enable either t-ecdsa ownership of neurons or certificate ownership. To avoid wasted effort, we recommend removing the restriction at the following lines that will enable canisters to hold neurons(deleting 12 lines of code vs having to write thousands of lines of code):
Pros:
We don’t have to write a bunch of hoop jumping code.
We can write trusted canisters-hotkeys that do things like auto-close spam.
DAOs can more easily control neurons.
Cons:
Neuron marketplaces will emerge and this may undermine the security guarantees of NNS governance. Unfortunately, with the coming tech, this is inevitable(unless we design some other method to enforce this.)
Action Items:
Discuss the proposal
Decide if we want to take extraordinary measures to restrict neuron marketplaces.
Prepare final proposal text
Note: @LightningLad91 expressed intent to write this proposal and just did it…give credit to him for pushing the issue.
Note 2: We have upcoming bounties that touch on these things and a number of other ideas that we’ve had that could use this that are non-neuron-marketplace related, and writing this proposal seems like less work than writing the code…we’ll see how discussion goes.
Note 3: Personally I’m very wary of neuron marketplaces as I think they break the governance model. I also don’t have a solution and this thread should be a place where we try to reach some consensus about how to address it. I’m guessing the foundation has had these discussions internally, and I’d love to hear their current thoughts on how to deal with this.
One proposal we could put in place(I’m not recommending this, just putting it on the table to help frame the debate) is to allow the censure of a non-self-authenticating neuron via the NNS so that if one appears on a marketplace we could freeze it. T-ecdsa still gets around this though, so all it really does is make it harder to implement.
I’d say that generally censure of neurons is a place that no one wants to go to.
Thank you for posting this @skilesare. I was going to post it myself but I doubt I would have communicated it as well as you did.
I really hope we can find a way to make this work. I think canister-controlled neurons have huge potential to accelerate decentralization of the NNS.
What do I mean?
By allowing a canister to control a neuron you could have a DAO manage it directly. You could have a service running under the DAO that monitors the NNS for new proposals and surfaces the ones it cares about. Maybe this DAO is a consortium of protocol devs and they only care to see replica changes. Those specific proposals could be surfaced through the DAO’s governance portal so that DAO stakeholders can vote on it. This means that the final vote could be the result of consensus amongst hundreds (potentially thousands) of independent community members.
Why does this help decentralization?
One way this helps decentralization is that it removes an incentive to bribe, lobby, or influence individual representatives. I can even imagine a more complex system of DAO-controlled neurons that supports elected representatives, but also provides a way for community to hold them accountable and remove them if needed. Ultimately, this gives power back to the individual community members.
How is this different from the NNS, why not just have everyone follow/vote through liquid democracy?
Because I strongly believe that communities want to have their own environment to discuss topics and debate outcomes. Not everyone wants to argue on the forums. Communities need to be empowered to have their own culture and their own leaders. If we give DAOs this capability then they can build their own forums and develop their own customs. Community leaders can encourage their followers to participate in governance; not because it helps the network, but because it helps their common goal.
I think neuron transfer (especially neuron marketplaces) are a major security risk. Hence, I think we should make them very hard or impossible. The community has talked about this idea a lot in the past, but I don’t feel like I’ve heard DFINITY comment on it very often. I would want to know their position.
The issue is that once you let a canister hold a neuron you can just change the controller of the canister and you’ve changed the owner of the neuron. A marketplace could automate this. So you quickly end up with a system where you mint your neuron through a canister and then can list it on a marketplace when you want to exit your position. Locking for 8 years no longer means much and you can still get 8-year returns.
Right. I realized my error after responding to @wpb. my apologies.
I’m still not convinced this current limitation stops that. The individuals who would benefit from this most have the resources to establish their own OTC neuron trading market. So some central authority manages the books and the keys for them. How do you prevent that?
@wpb just to address your point about prior community discussions on the topic. I don’t believe DFINITY and the community have had an earnest conversation that accounts for the line quoted above. These marketplaces will happen one way or the other. The question isn’t if, but when.
Hi @skilesare, many thanks for raising this proposals.
I do not have yet enough insights in all aspects of this proposal to provide an answer straight away. Let me review within DFINITY and the come back to the forum.
Doesn’t this just create another gatekeeper scenario? Unless, like @Severin said, you make the list so dynamic that anyone can join easily. In which case, what was really accomplished?
I agree this discussion hasn’t really happened, so I’m glad to see the discussion starting again and DFINITY is engaging so far.
I’m not ready to conclude that it’s just a question of when instead of also a question of if it will happen. I don’t recall DFINITY indicating that they agree in the past. It has always been the community that makes this claim.
The security risk is pretty big and DFINITY has always held to the position that neuron transfer should not be allowed, presumably because of this risk. Even if it is technically possible, I also wonder if DFINITY has the legal right to pursue litigation on behalf of NNS investors against any marketplace that forms.
Hence, unless DFINITY comes out and formally states that they believe making it free and easy is the right path forward, I think methods should be designed to prevent it including code limitations and legal recourse.
I can definitely be convinced otherwise, but I really want to hear an official DFINITY position instead of this just being a community discussion like it has been in the past.
I think this is would be a very slippery slope for what is supposed to be an open protocol / ecosystem. If DFINITY is separate from the network then this sort of enforcement (at least from the legal perspective) doesn’t make a lot of sense. I do agree with code limitations if the community agrees on them.
Sorry, I’m not trying to be annoying. I just want point out that @bjoern has already described how a canister can control a neuron today in this post. which is what inspired this proposal. As you’ve said before, he is the foremost expert on the matter and works at Dfinity.
This proposal aims to make the development of this capability simpler. The only thing preventing it today is the complexity. So I would classify it as a matter of when and not if.
Another thing DFINITY should consider is the effect of this on the price of ICP.
If I can go to a marketplace and buy a neuron holding ICP at a 5-10% discount if it’s locked up in a 4yr neuron, and at a 15-20% discount for an 8yr neuron - the demand for ICP on exchanges would dry up and the ICP price could potentially plummet, especially given the anemic demand side of the equation over the past few months.
I think this decision of “when” should be driven by consulting the data science and the economic wings of DFINITY as well.
I don’t think that neurons should trade at a discount compared to liquid ICP since the expected rewards will be included in any calculation of the present value of the neuron. In order for an 8 yr neuron to trade at a discount to an equivalent amount of liquid ICP you would need to assume you can make more than the ~20% annualized returns that the NNS is giving you by using some other form of risk free interest generated from your liquid ICP.
I feel that this may actually create an additional incentive to move ICP from exchanges into 8 year neurons since your neurons will remain somewhat liquid but holding your ICP will net you a higher interest rate.
Also worth pointing out that none of the existing neurons would be in play. This would only be for newly created neurons; and even then, it would only be for neurons created by a canister. So if these NFT markets were to appear, it would require a significant amount of staking to take place first.
Also, how difficult would it be for public blackholed canister (contract) to use edsca set up not just a neuron swap marketplace, but an identity swap marketplace? (with neurons attached to the identity and NNS wallet).
