Jehan Tremback Explores Shared Security on the Hub and Partial Set Security | Informal Systems

Transcript

let's let's move on to uh IC replicated security partial security actually yeah I mean this this was sort of the main reason why I wanted to get you on here is to talk about partial set security uh th this this idea of share Security on the Hub has been around for a while uh it was launched some time ago now I think over over a year and a half ago maybe even two years ago um time time seems to escape me these days but well it's been about it's been about a year that it's been live but yeah right okay and and so there's there's there's these three different terms right so this interchange security IC and there's replicated security and partial s security and I'm still confused as to

what these things mean and how they relate to each other so maybe just sort of clearing up the confusion here yeah um so I I yeah I do apologize for the confusion um I'm I'm also like I I think it's uh generally we should just call them all IC interchange security um um and replicated security and partial set security are just two different uh two different variations of it I would actually say partial set security is kind of like icv2 um and um so we're going to be doing some marketing work to kind of clear up those terms um but in general just it's interchange security that's the hubs reaking product so the um the original version of it which is in production now uh which uh kind of went live about about a year ago is

replicated security and replicated security every um every validator on the cosmos Hub runs every consumer chain so if a consumer chain joins every validator has to run it um it's actually not quite every validator 5% of the smallest ones don't um but in you know effect that is actually a large number of validators because there's many many small ones but effectively you know it's like every validator has to run it so um that is the highest level of security and there's actually there's actually an interesting I'll get into this a little bit later but there's there's this there's something called the subset problem there's a lot of thinking we've done around the security properties these different variations so I'd love to cover in more detail but just to keep a high level for now um the if the entire validator says running it you

have a full security of the cosmos Hub you have full like economic alignment basically um and it is the highest level of security however you know along with that high level security and the high level power you get from an app chain the expenses for validators are somewhat high as well so validators running another chain they kind of need to run another server um and you multiply that by 180 it's a lot of money so that's something where I think we we we didn't necessarily last year we didn't do a good job anticipating how and people were saying this um and I think we always knew there were going to be scaling limits to it but basically that's the major downside of of replicated security is is this is the scaling limits um so to get every validator run a consumer chain they all

have to or you know at least majority of them have to agree to it and pass the proposal and so we have two very high-profile very high quality consumer chains running Neutron is stride I'm very happy about that but it has been kind of a a high barrier entry to the the new chains just you know starting out and so that's a partial set security which is really just the next version of interchange security is addressing so with that you don't have to have every validator run every chain you can have uh a chain start basically permissionless Le and as soon as validators start opting in it starts running and so uh with with that it allows chains to sort of start up in a very lightweight way um have the cosmos Hub be a place to find a validator set and get started get launched and scale

their security as they need to and what that allows us to do is really ramp up S of like open the flood the floodgates to basically let everybody who wanted to run a consumer chain run a consumer chain and that's also you know our ramp up of the BD and growth efforts are you know enabled by that because then we can bring a lot of people in get them started get them going and there's not this big hurdle of the governments proposal so that's what that's what that that distinction is all about there there is a feature in partial security that's called top end so there it will there will be an option for chains to have guaranteed security so for Neutron and stride for example they're going to move to that um and what that does is you can have a setting that says for

this consumer chain we want to mandate that the top end percent for Neutron Insider the top 95% of validators have to continue running these chains and that's good for chains that do need to continue with that high level security um or if we get other sort of higher profile chains coming in they might also use that um so chains that can clear that governance hurdle um may want to do use that for the majority of chains though I think it's going to be they can just launch permissionless le validators start opting in and it's a lot smoother so that's kind of the uh yeah that that's to clear that up it's all just interchange security um and partial side security makes it easier to launch so so this top end uh thing you mentioned so that that would mandate that validators run or at least that the the top 95% of

validators Run Security for these chains is that a governance is that subject to governance yes yeah it has to be yeah so you know regular opt-in consumer chains uh they can um you know that's where there's no top end permissionless they they can launch perly I think in the first version there will be a governance proposal for every consumer chain um but that's kind of an implementation detail but uh that's you know it will be soon it will be completely permissionless and um but but yeah for for top end if you're mandating that a certain top percentage validators run it that there does need to be a governance proposal so top 95% is kind of basically equivalent to replicated security the the version um but you could also do something like top 66% 66% of the validators actually is only 24 nodes so um that's pretty

interesting because you get like the majority of the of the same security um but it is only 24 noes that have to run it reducing the costs by a lot but yeah I think it's going to be less important than opt-in security it's more just there if you need it the vision here seems to be kind of different from the original vision of of interchange security at least how I remember it I I I remember interchange security being this idea that the cosmos Hub would secure a handful of chains maybe 10 20 chains these would be very high quality projects like you know new the neutrons the strides the Nobles uh aligned and complimentary with amongst each other you no competing plat chains on the Hub and I kind of like that I I I felt that

this would would create sort of an ecosystem of of applications that would create a lot of value for the Hub and that this curation was really an important aspect of of interchain security it it seems now that that shifting to a more permissionless model where the Hub becomes kind of a security Marketplace and validators can opt in or opt out of securing certain chains [Music] [Music]