If memory serves there are about 4000 independent data centers in US/UK/Germany/China ….total.
Obviously you can have concentrations of nodes in any particular data center.
Why are we limiting our imagination to the thesis that end user devices are cheap and inferior? Why are we even bringing up Raspberry Pi?
The access point and the pipe is agnostic to everything but the math.
What if u set up an access point in the data center?(like we’ve been doing for the last 30 years in telephony)…
Oh ….might that change the end user device?
Why yes it most splendidly could.
Why are we treating the term decentralized like they did in the early part of the century? Aside from being the ‘marketing spice’ is it not about the distribution of power among many participants?
That is the endgame consistent with the philosophy that I hear articulated by folks that preach decentralization uber alles. So let’s go there.
I cannot quite speak for the tweet you posted, but I think the reality is a composite of the following statements:
Dom sees Badlands as important but wants this to be something the community does want. There is a reason he wrote in his medium post the following:
Everyone, please note: this is a technical pre-post for those interested in the Badlands concept, rather than a formal post announcing details of the project. There are several competing demands on the teams developing the Internet Computer ecosystem at the moment, including work currently underway to directly integrate the network with Ethereum and Bitcoin using “Chain Key” cryptography. Providing opportunities for amateur node providers remains an important objective, so I’m sharing this post today to provide an opportunity for community discussion.
The Foundation has multiple complex inter-layer projects while also scaling the network and infrastructure so there is a high bar for tackling more complex projects.
One possible solution for a more decentralized layer on the IC would be to simply work with and integrate one or more other p2p compute projects geared towards consumer hardware. This would provide developers alternative options for building apps with greater levels of decentralization or lower costs compared to only using IC canisters. Building on a different protocol would of course be different and possibly more complex than building an IC Canister.
Second obvious question: how will it survive ddos ? throught internet identity ? or a kind of “the power I consume, the power I host.” Somehting like the good old bitorrent communities…Lets talk about it !
New in here. My humble opinion while this proposal migth not be a requirement in terms of technology it could have a great impact in ICP adoption. The barrier to participate in the project could be much lower, it will enable enthusiastic people and devs in general to participate in the construction of a real internet computer even if it has lower performance any way many applications does not have such high performance requirements (may be thats why there are so many types of hardware and devices with diferent characteristics). The sense of ownership and participation of the community should not be undervaluated, look at NFTs, people likes to own things and to colaborate even if it is with a raspberry on a basement.
In the more technicall side of things there could be layers of performance for subnets (NNS verified) with diferent fees or other properties and the developer of a DAP could specify which layer to target for the deployment of his canisters.
Really like to see a masive adoption of this project in the right direction.
I think making the badlands a subnet or set of subnets would be great for decentralization and just having the option of hosting with high performance but reduced decentralization or low performance but increased decentralization is amazing on its own and I don’t think any other network offers it. My only concern, which I haven’t seen anyone mention, would be inflation. So there should either be no rewards for badland nodes or rewards should be paid in a sub-currency that won’t affect ICP.
In my opinion badlands nodes should require an ICP stake that can be slashed in case the node isn’t reliable, so that providers are incetivized to provide constant uptime, if nodes where constantly being powered on and off it would be bad for the whole subnet.
I’d also like to see different types of badlands subnets, just like for standard ones. There should be very decentalized subnets with low HW requirements and subnets where an i7/i9 or equivalent CPUs are required. Badlands subnets should be paid based on the amount of “service” , be it bandwidth or CPU cycles, provided to the network.
I was initially against Badlands, because I thought it was a distraction from an already complex roadmap. However now on reflection I cannot see how we can have a truly censorship resistant ICP without something like badlands and as a separate network with a more limited NNS not just as a separate subnet.
Known data centres are easy targets for legal and political pressure,
Token holders can easily be coerced via either exchange delisting or personal liability.
Properties required:
Lower node and bandwidth requirements.
Different token (initial distribution should be based on ICP but also some “mining”) so can evolve own more censorship resistant culture.
Anonymous and “Permissionless” onboarding (Staking & people parties)
Bad actors automatically slashed or kicked of by in protocol mechanisms.
Designed to be more private and censorship resistant rather than compliant. (Either ignorance .plausible deniability or market based methods (pay more to host stuff others don’t want to host) )
Node and client IPs hidden via a mix network.
interoperable with main ICP network
But it is ok for it to be less performant and to require installation of client software to access it rather than just a browser.
It will create an interesting dynamic wrt censorship.
Protects main network from censorship as censored canisters will just move to badlands.
Protects main network from bad stuff as dodgy canisters will likely just launch on badlands rather than main network.
I think that sort of is what it is. IC is a completely different kind of platform to traditional L1s. It’s simply just going to take a long time for people to realize it’s its own new thing. Just like it took years and years for BTC and ETH to get major adoption.
I would prefer it to run as subnets and I’m ok with them having up to 10x decline in performance as long as composability is preserved. Idk, call me crazy.
Not sure I agree with that, there are A LOT of data centers and entities around the world and if you utilize all of them you can’t shut them all down. BTW I’m not against lower spec subnets I just don’t think it’s a priority.
BTW it’s not a requirement in the IC protocol for nodes to be in data centers or KYCed.
On the prioritisation side, I think having one badlands subnet of let say 100 nodes before end of 2023 would be nice and seems doable (without any prior knowledge of the technicalities).
Maybe someone from dfinity can shed light on the technical feasibility @diegop. It would be interesting to know as well, if the community decides to do it, do we have all the necessary resources (code, docs) available and open?
Yes, I also disagree. Decentralization refers to the transfer of control and decision-making from a centralized organization to one that is distributed. Being known or not is not that important, especially if providers are geographically dispersed to minimize censorship. I am more concerned about the decentralization of the NNS. Because of this, Badland implementation should be thoughtful