DFFML Rolling Alice: Engineering Logs: Volume 0: Architecting Alice, 7 May 2022

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Architecting Alice: Volume 0: Context: Part 13: Thoughts on/as NFTs

Description

https://github.com/intel/dffml/discussions/1369#discussioncomment-2706807

A

All right, we are back and we're gonna spin up firefly.

A

And we are live fantastic, okay, so.

A

Let's get to this so yeah, we had talked about okay check content link, um so this was we were implementing. We were experimenting with um you know. What does uh you know? What does the ideal flow with the system, local resource management look like or the the the you know, locality-based resource management and our abilities there. um You know with regards to you know basing something off of this.

A

You know an open architecture spec where we'd have uh you know uh essentially dynamic levels of reconfigurability and if, if you're thinking about this really um all we're saying here is it's just a name for not having to bang your head against the wall when you need to recompile several pieces of, say a linux distro in order to deploy your application right.

A

So if you've ever had to do that, you'll know how much of a pain that is right, um and so you got to go, get all these user space components. You got to set up your build environments for each ones. You got to go. Do your compile? Okay! Basically, what we're saying is overlay boom right? What are my patches?

A

I want to throw on top of this thing right now make those patches work with whatever the distro build environment is right, so you know, maybe I need to run changes to firefly cli on top of um on top of like alpine or on top of arch right, so we're gonna have uh underlying um orchestrators that uh facilitate the rebuild via the uh the package management system. That is for that distro right, and that way you get a cross-distro application of these overlays all right.

A

So um so, yeah check content length trigger operation to request disk quota.

A

Resume download or receive body content body uh wants a ticket for once uh quota confirmed available.

A

Once content length confirmed available within quota system, local resource management, which is you know, uh aka ida, um traverse input, parent.

A

And interact with backing.

A

A

A

Okay, um yeah yeah, so right now we don't have that so we're just gonna. Do it like here.

A

And we're going to store the cache, oh and the other thing that we'd looked at was basically okay, so say that we want to um snapshot the state right and, if we appropriately so right now, what we have is this little firefly config right essentially defines its own interface right. Well, it's its own plug-in right, um so so, and and yeah there's a lot of awkwardness happening here, maybe we'll just spin up firefly yeah we're just gonna spin up firefly!

A

There's a lot of awkwardness happening here, trying to implement this as an operation, um because these are this. The awkwardness that we're seeing here is the stuff that we're going to fix by doing the turtles all the way down sort of thing. This is just like an example. This is now an example of why the api is awkward and and and the things that need to be done to fix it. So we're just going to go and uh nuke the rest of this or no, let's not do that.

A

We should make sure that this is reproducible.

A

A

Okay, so we're gonna say: maybe this is yeah firefly import.

A

So this thing is going to.

A

A

Where's that gen funk stuff.

A

A

That's in data flow as class, so oh so that data flows, class.

A

That will also help us with this type of thing.

A

So when you define a data flow so um so because there's- and I think, we've gotten into this because of the context- the double context- entry and the way that the operations are really just like a single function- call um the stub the function prototype the operation right is can be different than the operation implementation and therefore you can have the operation, implementation introspect, the operation and understand you know if its behavior uh should be different. For that.

A

um You know for that particular operation that it is uh in because it your your mapping, operations to insta implementations, to instances right and you can reuse instances. You can reuse implementations for different instances. If you wanted to right um or you could scope them and you could basically reinstantiate, um and so this is an example of where we have separate instances.

A

And um but we're using, let me just show you so run dataflow so run data flow default inputs, flow inputs, okay, so, basically inputs to add to the data the key value, mapping of context, string value pairs. Let's check out our examples: okay, so basically yeah, you know. So if we pass basically what we're doing is we're passing this here right uh same thing, only value definition. So it's not input object, they're, they're, they're, raw they're, they're, they're, unconverted types.

A

That's even worse,.

A

Something's not ideal with the comment text and and it's barely visible in that dark, blue, um so yeah.

A

A

A

Less on my path, the following shows how to use run dataflow with custom inputs and outputs. This allows you to run a subflow as if it were an operation. Okay- and this is what we're interested in. So basically we can say okay, so this last path is an operation which is a subflow we're going to instantiate, run data flow within our data flow and we're going to say you know the instance right.

A

So this is the instance right as exists in operations, which is our prototypes is we're going to modify the instance and change the inputs and the outputs, and then we're going to you know change the config or yeah we're going to instantiate using that config right so that it uses the subflow and then basically, what will happen is that when we come in here and we'll end up, triggering this run, custom right and run custom is going to go. Look at the inputs right.

A

Self parent op inputs items move string.

A

Context, input name context, input definition, so this is some really rudimentary stuff right now, where basically we're saying the first input given in a custom is the context, um which is why we did this definition: url, okay, so we're. Basically, this is sort of like our record depth. So this all needs to this is why all of this we have so many implementations of the same thing right. We just didn't know that.

A

Sometimes we didn't know that, and we didn't know how to unify now. We know how to unify right. So this is what we're doing we're going through we're unifying um okay, so essentially yeah. So this is the point. Basically, the point is that you, you know you can have. You can have a custom custom run function if there are custom inputs, so we're actually gonna just copy this directly. So.

A

All right, okay, so this is from dfml and we'll make a note so.

A

Okay, so um data flow from peer id, okay, yeah. We were talking about that um and then we have you know firefly, okay, we were making this operation to down or we were making this context manager which downloads firefly and okay. We don't need a context manager for that, because uh cache download on pack archive is going to be.

A

Oh, we do need a context manager, okay, so to do, remove context manager when okay, you know what, when okay, so to do, remove context manager when we fix op imp enter and ctx enter to not attempt context entry before setting return value on parent uh key, given remove context range, remove context manager when we fix op enter in ctx, ensure to not attempt contact surgery before setting return value on parent key. Given if return value is not a context manager.

A

Is there anything no context, I'm not sure if there is.

A

um That could be helpful there. So basically that way you know uh instead of so right now. What happens? Is that um whatever you return from this function, since it's a lambda right, you have to return something that then can be called and then the return value is a context manager which we enter. So if you just pass a regular function, it's just going to blow up so.

A

We want to uh call means instead of entering their context.

A

So basically, if we see a co-routine we're just going to go, call it and sit into the context and so we'll let this run or we'll let this we'll let this sorry.

A

A

We're going to take this run dataflow context, we're just going to find our place.

A

Firefly supernode.

A

Or maybe with a firefly import.

A

Or maybe we'll call this.

A

Inputs to odap.

A

Remember we'll just call it odap.

A

A

A

Okay, so let's go read a little bit more about this.

A

Remember always read the second series consideration the first tbd, not good red flag.

A

Red flag, major red flag: this thing expires in a few days and they haven't written anything about security, major red flag.

A

Man, that's unfortunate, so what are we gonna do to get this reviewed? I guess we could sit here and review it.

A

Don't really have time for that, though,.

A

We could cross our fingers.

A

And hope that somebody secures it by the time we go full blown with this thing.

A

I think we can basically assume that.

A

I said lock in your escrow: okay, I'm glad that there's locking.

A

Virtual asset transfer between okay, so basically this is defining between blockchains and then we can extend that concept to between.

A

Decentralized trust me: oh, hey, decentralized! Trusted computing base. Hey that's interesting! Oh ned, ned ned. What are you doing? I just talked to you.

A

Well, this is perfect, I talked to ned several years ago and that's how I learned that this is the way things are going and look here. He is hardware roots of trust. This is all of the stuff that we're going to be leveraging. This is exactly so. We need to go talk to ned.

A

A

A

This is perfect. This is a road map is what this is for. The different underlying secure, further considerations: when did you publish this december? 2019 wait.

A

I think I read this yeah. I think I read this a while ago. Okay, so we need to go, ask ned what is the most recent thing that he's written?

A

It goes prolific. um I should have thought of that. I thought I talked.

A

I need to go talk to him again, all right, so this is three years old.

A

Yeah I just talked to him about the centralized identity. What did he say.

A

Maybe I forgot: I've dropped a lot of packets lately.

A

A

Yeah he he also. He said that you know he would like to see the you can stuff support. Seabor and I talked to brooklyn about that.

A

Oh, we should go check in on that stuff. So, let's see.

A

So this looks like something that we're going to really tie into our um identity, aware context stuff in terms of like transferring around user permissions as we run these things. So where is hey? Look at this web3 storage?

A

What is this nine days ago.

A

Okay, so this is another.

A

Okay, so if you're building an std market- okay, so basically so just like these nfts right, so the nfts all of the data, basically anytime, you create an operation. There's going to be an nft for the output is effectively what we're doing here um and then that data, the data for the nft lives. You know off chain.

A

And it lives off chain as accessible by the data flows or the open architecture which you know gets embedded right as the dna. The encoded dna is context aware, embedded oops, I'm knocking over the server over here. Okay, um are we on the internet? We are on the internet, okay, great.

A

This is three years ago. I think I remember reading this.

A

A

Talk to him about the ids, he said, cbor.

A

Would be great, uh and so I'm guessing that's because dice stuff there's this dice stuff.

A

I think is device identifier, composition engine, so this is robust, resilient recoverable iot, so this is um basically we're going to tie this dice stuff in with we're going to tie the dice stuff to the um dids.

A

That's that's our goals, part of our goals. Right here we want to tie dice to dids to cicd to ml right, um okay, computed device identity. This is what okay.

A

And if we can do that, then we can end up with robust resilient recoverable alice.

A

Okay, so let's see so participants when you can offer identified by a key fair, which is this private signing key, corresponding okay. So.

A

Syrian lord keepers.

A

Let's go into this: take care to save creating a uten token, creating a root. Token derive token, create child token.

A

Restored for nfts.

A

A

A

So this nft dot storage? Okay, because if we can basically say the data flow is the nft and the nft is in a.

A

A

File coin decentralized storage, network.

A

So are these people using dids.

A

Okay, so it looks like there's some pretty good cross compatibility here.

A

A

uh Decentralized.

A

We want to know about decentralized identity, the road to decentralized identity, falcon spec, I'm not seeing I'm not seeing this look like it has.

A

Nft storage, okay,.

A

So what is this? You can storage and house? I don't understand using you cans with nft storage use of you. Cans to delegate upload permissions in nfc storage is currently using you cans to delegate upload permissions in nft storage. If you have any feedback discussion, let's go check out this discussion, so this is telling us that we can use these ucans to authenticate to the nft storage api.

A

A

A

A

See this is what we want yeah. So maybe we need nfts.

A

We want nfts on dids on pure d, ids.

A

If we can do nfts on dids, then we can do pure dads and then we can build our graph.

A

I'm just wondering I'm wondering if there's something: that's like okay, so this is great, so we want to intercept with this stuff.

A

We want to intercept with the ucan stuff, so we know we want.

A

It could be good, so c intercept our intercept route is um uh identity, aware context.

A

uh To um uh you can to seabor, which is you know via seabor, uh you know dice, uh so this is gonna work, a device at the station.

A

So this is how we're going to end up with our hardware root of trust, um so we're going to tie our identities and our credentials and we're going to hopefully end up with this uh authentication mechanism via ucan via c-bor, which is interoperable with the authentication and attestation mechanisms that our devices are using. So we're going to prove that we are, um you know, have certain claims, just in the same way that devices prove that they have certain claims, and then this is going to allow us to have this.

A

This opera interoperability between um you know, execute sort of execution as a service where you have some like human in the loop stuff right, and so you know you have uh tested credentials for which humans are doing, which things- and this is what allows you to swap out basically subcontract any piece of your open architecture or have a device to it all right.

A

A

A

To share uh crit attestation via open creds type stuff.

A

Across humans, slash machines.

A

When we need to mitigate when our strategic plans are attempting to mitigate for um when our strategic plans are attempting allows us to share our decisions, we open credit service stuff across human assessment teams, when our strategic plans are admitted in to mitigate risk via uh execution of operations uh via implementations which have a tested properties.

A

So it's just going to simplify our overlays because they can just look at ucan stuff instead, so let's go check that thread. We will spin up firefly.

A

A

Okay, so here's this nft storage thing. So this thing: free decentralized, storage and bandwidth for nfts and file clients. So examples you can node, okay, basic output using you can token okay, so this is great. So here's something this is going to be. This is you know, part of our on-ramp off-ramp setup right when we deploy these to a live environment. um You know, maybe this can be part of our our bridge stuff, so um I mean unless we can just so. What do we want to?

A

Do I mean okay, so we want to understand so what what why? Why am I off looking at this stuff right now? um So why am I off looking at this stuff? So if we want to be able to mesh, we want to mesh. We want to mesh the goal immediate.

A

Goals um uh leverage existing uh off-chain storage interfaces provided by nft storage.

A

uh And authentication.

A

Via ucan, based on dids, to support two support uh or to act as gateway as portion of gateway.

A

A

Reverse engineering.

A

Environment, where we attempt to get data into firefly network firewire dlt based on dids okay, so um we could go spin up the firefly thing right now right um and we could go uh dump all of the. um uh We need to talk to brooklyn too and find that discussions thread so where's that you can. You can.

A

Because I promise that I would provide feedback when I have time and now we have time- and hopefully we have a little more- we can have some feedback right. So let's go give some feedback.

A

A

Transparent format- agnostic, you can's; um okay, yes, so this is the thread.

A

Let me link to the recording here we're at 35 nights in just because this could be helpful if we go talk back.

A

A

Where we we are talking about this is about 30. What was it probably 30 minutes in 30 minutes in two blank?

A

uh The fully connected development model, we're gonna capture, all this information, um so let's go spin it up all right. So um what are we looking at here? So we are looking at.

A

Okay, so thanks for the comments- okay, so more discussion february 3rd, the p2p okay- was there anything more down. When was this? When did I last comment on this?

A

So this was back in march okay, so we need to okay, increasingly we're custom performance. uh Okay, so um looking at this again uh really excited about it, I would like to see seabor support still, as that will simplify.

A

Okay, so, but that's not the question, we're not answering the question, so what is? Let's just put a note. I think I'm too in the zone on this right now, so um arbitrary resources, degrees of freedom. You can version.

A

Yeah there's we looked at that the other day.

A

Auto code wreck, essentially referencing web central, monitor.

A

A

I believe what we're suggesting is this sort of auto codec? Is this open architecture.

A

Auto, oh yeah, we're basically so so. The auto codec, auto codec maps to.

A

A

Maps to um our shim discussion uh manifest encoding as dna discussion right. So that's. The dna is the codec that we're encoding too. If that's to write the words, um oh god. Oh no, um so.

A

All right, so, let's go look at this okay, so we would like to have the nft storage thing. So if we can spin up the nft storage to point at a did based blockchain, we can go prototype all this very quickly and then dump the chain and then reconstruct the format on the other side. So let's go look at this nft storage stuff again, so this is.

A

A

So we've created bulk imports. Okay, that could be interesting.

A

We want to do some bulk imports, import list of cids, so ipfs, okay, so we're going to store it in ipfs, but where is the chain? So what what chain.

A

About okay, yeah offchain data.

A

All right, data's content address using ipfs, meaning the uri pointed a piece of data is completely new that data okay, so we're gonna run this thing.

A

So we're gonna run this thing we're going to so what? If we want to host it ourselves.

A

Okay, we have a slight problem: if we clone it get repo, the computer will blue screen.

A

A

Okay and if you showed but I'll, give you content, I do upload when minting an fat nft to make it truly immutable and then performantly available. Okay, performantly is quite the claim there. I'm really curious about that. uh We'll see so uploading to answer.

A

Smart contract base url is this something feels good to discover this sound a lot of time. If you guys come into use for fuel, you have no time.

A

Okay, so token, uri based token uri, all right, okay, so big, okay, so base url api start linking to a directory okay. So um this didn't explain anything to me. I mean I didn't really read it, but.

A

Storm and nft is using custom. Metadata sounds up our alley.

A

Okay, single file.

A

A

So we can upload multiple files, vfs content, hey look, 100 megabytes single size request, I'm preparing your metadata. Once you have your all your assets stored. You can upload your meditation.

A

Maintain your nft.

A

So you can win tokens using the block flame chain platform of your choice.

A

Avoid storing http urls on chain: let's continue your way to retrieve nft data from ipfs may involve using an nfpa fest shape, http gateway. You should avoid storing http gateway links in smart contract or other blockchain record, instead store ipfs uri, which doesn't need a single gateway provider, rewrite this uri to a gateway link at the last mile when displaying a a a when displaying the nft on the web. So um okay. So what is this saying? So this is basically saying you know. Your ipfs is like this distributed.

A

Data store right, and so uh your http interface is really this web two thing, and so, if you happen to, if you find yourself displaying, if your context, if your context is such that your parent contacts in some in some parent is such that you are going to render this thing to a user via http, you need to make sure you change the reference length to go through the http gateway for ipfs.

A

And so this is the type of thing that we're going to have the visibility into you know we would store things yeah in ipfs or we would probably actually store things in an operation which will then yield us something. But we need to see so story directory of files.

A

And tft projects: okay, creating a project, install nft storage files from path, adding the upload code store directory.

A

So basically we can say: okay, nft storage, token, okay, so we can store a whole directory in nft storage using youtan tokens for uh dedicated authorization using you can.

A

A

Energy storage, api.

A

So I'm confused is this nft storage? Like is this? I don't, I still don't quite understand whether this is open source or not, because we need this to be open source.

A

Can we run this development.

A

Getting started development.

A

Okay, the api server, okay,.

A

A

Let's update the plan.

A

For this recording.

A

Deploy nft storage with.

A

A

And interact with it via you can token auth ah then uh ensure, or so.

A

Ensure nfc dot storage can interact with did based blockchain.

A

Firefly: okay, we're going to use firefly to do that. Firefox transact uh question mark because we're not sure uh dump resulting blockchain dump resulting chain reconstruct um uh chain via python code.

A

In I do female operations pure daddy.

A

So the tree has it wrong a folder named peer id instead of what it should be.

A

A

All right, so maybe we'll fix that too.

A

A

A

So, let's see so, where is this.

A

Okay, so what endpoint for production, postgres, cluster nodes, ipfs staging database, postgres postgrest.

A

Oh, it turns postgres into rush. I was like okay, okay, yeah token, for what is this. This is some kind of telemetry pinata.

A

A

Sure you're nft learning how all right, let's just think of an nft thing, dead cargo api. What is that cargo.

A

Nft storage tag, cargo, we're going to work in progress; okay, so this looks like it's under development. This name will not survive. Do not pull gold rush with maintainer. What is this person working on?

A

Oh, this person is familiar.

A

Why are you familiar.

A

Possibly forming a cyclic direct, directed acyclic graph, around streams, parts stream, dagger.

A

Walls criteria, so this sounds somewhat similar.

A

Okay, it looks like maybe abandoned. Okay, um you can private key. Okay. Great um front end also needs environment, so just cloud for pages secrets: okay, so they're gonna production, environment or else local, uh okay, so magic secret, key.

A

Magic clink app.

A

What is magic link, a personal account, magically delightful auth for web3 onboarding. Are you kidding me.

A

A

Eventually, it'll be ucans.

A

So what I don't understand, how they point at a blockchain, where are they putting it on chain.

A

I see them putting in an ipfs.

A

So I really don't understand something probably using energy storage.

A

Before you can create the blockchain record, you'll need to store all the data off. Change comprise the energy pack. Once everything is stored, you can use the ipsf file metadata link to form the on-chain token. Do everything else, with lower examples of starting ft assets metadata using javascript and http api both use the store, so we can use the store.

A

So there's there's this year, three standard metadata, so we need to make sure that we conform to. This, probably is probably what we're going to see.

A

A

Using custom metadata.

A

Okay, so what are the metadata schema.

A

Uri tokens token: okay, the url value for the id substitution by clients, the string ids exists in any uri. Clients must replace actual decimal form; okay, so basically so metadata, uri okay. So this is saying uh basically um standard interface definition, standard interface detection. This sounds like something we care about. We care about standard interfaces, standard interfaces like the erc20 token interface.

A

Okay, publishing detector center is detection.

A

Okay supports interface, id.

A

Interface id interface, identifier,.

A

A

Okay, so basically they just assigned this supports interface. Is this okay, so the source contract makes a static call to destination, address with input data and guess this corresponds to okay, so this corresponds to contract support. So basically we say call this function.

A

uh If it doesn't, then it doesn't support. So basically those contract challenges, so there, okay, so all right. Okay, so you see one. Oh sorry, ruffles.

A

Okay, so this is the interface, so maybe we need to encode the data flows to the smart contracts if we encode the data flow to the contract.

A

I don't want to set this to gas.

A

Okay, so basically so this is cool, so basically, I think we can use this type, so we can encode to this type of thing.

A

On tracer detection of methods for smart contracts, uh should we encode two smart contracts.

A

Okay, so we're trying to figure out where does this? Where does the data flow execution engine fit into this?

A

This on chain off chain, blockchain model.

A

A

Put this builds on that.

A

Multi-Token standard standard interface for contracts to manage multiple token types are deployed. A single deploy contract may include any combination of fungible tokens or non-fundable tokens or any other configuration okay. So these are the different data types that we might be working with right. So the contract is the data flow. I think, and then the data flow types of the inputs right, the nfts are the inputs or the or the fts or the inputs.

A

So your transform.

A

My safety call save transfer from safe batch transfer phone balance from battery okay, so.

A

All right, so we just want to ride on top of this stuff we want to put. I think this is just going to be the tokens and the blockchain and all this stuff. I think this is going to end up being. You know, uh I think it'll be it'll, be yeah, it'll be good to be based off of um because then you can leverage that for the reward system for execution.

A

So basically you can issue. I think you can issue, you could issue a contract. That says please execute this thing or please.

A

Like you could issue a contract that says.

A

Pay this address address then it does this thing right and that ends up creating an input object which is itself you know. So, basically, if you, if you execute the contract, the contract will so basically you you want, you want to have some other nodes in the network, run a data flow right, so you issue, I think you issue the contract or you yeah.

A

Somehow you expose this interface, you offer like the bids right, so people can bid on it and then people can execute it, and so if they go execute it, then they.

A

So, basically, when they execute the contract, they have to be able to prove.

A

So when they execute the contract.

A

They have to be able to prove that.

A

What is it so? They execute the contract and the contract is to do stuff off chain.

A

And so to do that, you need to have some kind of check on the risk tolerance ahead of time, so to execute contracts which will go on fixed off chain data. Okay, so executing contacts.

A

Execution of contracts we're executing where um uh reward rewarded party executes off chain data flows off change system, contacts.

A

Execution of contracts where rewarded party.

A

Executes off chain system context needs oracle needs or needs information within network to make decisions to upfront upfront.

A

To make decisions about if a party is able to accept a reward.

A

So basically you put this thing out there you say hey, you know I will pay you this much money to do blank. So you put out blank. You put out system context, so process flow.

A

Publish strategic plans, publish system context they want executed to chain.

A

There's no contacts they want executed to chain and so flow for rewards. So this is a we're exploring a possible flow here. Where you know, agents are communicating together, um but they you know, they need to be compensated. Somehow uh you know maybe for their compute time or something.

A

So- and this is a tangential thing so just exploring how this might work so.

A

A

Okay, they want to execute it to the chain. Okay, so they basically, you know. This is like a call for proposals.

A

To execute right so others publish their proposed system context, which they so others publish, proposed system context.

A

um Others, other agents in network.

A

Publish proposed system context in uh proposed uh applicable.

A

System context: hey, I want you to do something all right. This is how I could do it. Okay, so this includes, you know, includes provenance information that that would be provided.

A

um So basically, then they say hey you know I can do this thing. Here's the provenance information that I could provide. If I did it right references.

A

To previous executions,.

A

A

A

Includes what provenance information would be provided so, basically.

A

So what salsa level can you meet.

A

Right um so we put out this thing, we say hey, you know I want you to go, run some code and uh all the agents out there looking at the chain. They say: well, okay, well, you know I'm doing similar work to that based on alice's brain is thinking and she's thinking. uh You know I'm doing similar work to that uh on.

A

You know this these trains of thoughts that I'm operating on because, of course, alice is across many devices, um and uh she knows she's sort of forming this this this just this yeah just just yeah, okay, um so basically uh everybody's got their own alice for different purposes right. So all these little uh versions of alice are sitting out there and they're listening to the chain and they're trying to complete their. You know move their strategic principles in the right direction until they're, you know turned off right.

A

So um uh when we do that, so we have this possible flow where we we and- and we have to think about when these contracts are issued, all the data needs to be on chain already right, because the decisions that are being made here- and we also want to think about the properties of trust that we have in the in the physical world and extend them to the digital world and not really, you know, come up with too many unreasonable expectations for the for the digital world right.

A

So um how are we going to do that right? Well, so, uh before any money changes hands or any rewards system or any token or whatever that that exists? I I'll be honest. I don't really understand blockchain stuff, that's what we're doing here, we're going to figure it out. um So uh once once um you know, please correct me if I'm getting something wrong, um but once um the so so we can put these contracts out onto the network. We can basically issue a contract. We can.

A

We can issue an nft, we'll we'll mint, an nft. The nft will um uh was. It will mint an nft on a did, backed network right, and the contents of the nft will be a encoded system context right and and we we are going to encode that system context in a uh you know, in a context-aware way right. This auto codec sort of thing that that was referenced in that other thread, which is basically okay.

A

So if I'm looking at something on a blockchain and I'm pretty sure it's a system context, I should make these assumptions about decoding, and then you know start from there and that's what the shim will do right. um So, basically we're going to publish so we're going to mint it we're going to mint them in nft we meant an nft and the nft is basically like a request for execution right uh uh or so we can mint nfts that are um so basically so call for proposals to execute or hypothesize right.

A

So process is mint nft for system context on did backed chain.

A

Right then, other agents in network publish proposed applicable system contacts which, including what provident provenance information, would be provided you know, so you can map that similarly to to to to salsa levels um and uh if, if you know, can that agent meet so basically I'm going to tell you, uh okay, you know I'm going to publish another thing to the network right and- and this is my bid on the nft and um the bid is like a reverse bid. So it's like okay, you pay me.

A

If I buy this nft, then you pay me, then you accept this contract right. So uh bit is like so uh like a reverse bid right. um So if I buy.

A

The nft you just minted, um then you then we have agreed to a contract which says if agent buys the nft the original system context the the system context originally minted within this example right, and so we can sell. Obviously, so we can sell um you know we can, I think, isn't there some concept. I really have no idea, but I think there's some concepts um where you know you can have very similar versions of some.

A

I mean, ideally, what we could do is we could allow multiple people to buy the same thing um you know. So maybe uh we put a note here. Ideally, uh multiple people, multiple agents, ideally an or nft- can be purchased.

A

Multiple times so maybe it's not an nft. Maybe we want a fungible token.

A

Right so maybe the use case.

A

Yeah, maybe this use case is that we want a fungible token right. I don't know what fungible versus non-fungible means, but you know. Suffice it to say we. This is. This is um the the input objects themselves should be static right.

A

What is a fungible token.

A

Concept of fungible verses not or whether or not token like objects or changes as far back as roman empire. Okay, so.

A

Utility tokens give holders access to projects or services or blockchain-based security tokens or traditional assets like stocks and shares represented by digital tokens on the blockchain, the most distinct type of tokens are fungible and non-fungible, so fungible tokens are divided. The divisible and non-unique okay, so, like money, interesting non-fungible token are unique and non-divisible. They should be considered the type of deed of title or ownership as a unique non-replaceable item. So maybe we use nfts so.

A

Nfts might be good for results. Outputs and ft's might be good for issuing requests for bot.

A

Do which is basically do something with the system context.

A

And so what what? Why might be so so nfts uh uh are likely also a help to us in solving our um uh locking issue right because explicitly they have to be transferred to be used. So this kind of you know could help with our distributed, locking.

A

Situation needs so you know inputs with lock equals true are likely nfts. Okay,.

A

So and nft for system context, okay, so.

A

Okay, so we know that we want to put the input objects that have locks as nfts, and my guess is that a fungal token is a simple case of a non-functional token.

A

I'm trying to make sure we're working down the right path here.

A

A

She will be using nfts.

A

The nft thing is attractive because of this off chain link. Here.

A

We'll take a look at the data.

A

All right, um let's just see if we can spin this thing up for god's sakes,.

A

Okay, what we want data transfer value data, okay, so non-fungible tokens store um data. So we know we want data. Okay, so.

A

A

Metaflex can deserve people. I feel token. Okay, so bring your metadata metaplex token standard. Okay address here, basic austrian letter, foreign.

A

Yeah, so going back to that example, I think that um so we issued this request. Hey. You know I've thought of this thing. I would like someone to think more about it. Could somebody go run their models on it or could somebody go execute this flow, so hypothesis or execution?

A

You know what what what's your bid for that right? So then another nft is issued in response right.

A

Another nft is issued in response where the data within the nft is this stuff. That's stored off chain right and that's okay, because we just want to go inspect it right. We're going to do you know, basically, is this system context valid right when translated into our environment?

A

Does you know we're going to go, run our analysis and we're going to see you know all right, so there maybe somebody's proposing an aligned system context where it's going to work to it's going to it's, it's different they're, doing different work, but they're moving the strategic principles in the same direction right so we wanted to go. You know clean up the code base and this person is going around doing refactoring. Okay, so well, you know their refactoring. Work is eventually going to clean up the code base.

A

So then we would say: okay, you know maybe accept contract right. So there's sort of this proposal counter proposal flow.

A

And then acceptance, and then um so and when the counter proposal is issued, so the proposal maybe could just be the single hand like single thing. Where then, all of a sudden, it's just good, but it could also be. You know this case where the counter proposal is issued, right and and uh yeah.

A

No, so the counter proposal has to be issued because we then need to go um analyze the off chain data within the counter proposal to decide if the provenance information that's going to be provided with uh to us is acceptable and check the history of that um that executor right, so we're gonna basically make decisions uh of who to reward right who to actually.

A

A

A

If agent buys this and kind of resume it sound that they have agreed to congress says they will preform the will hypothesize.

A

Or execute the system context uh outlined in there, so we're just gonna.

A

Scratch this out.

A

A

Okay, so we said ideally, you know they can be purchased multiple times. We don't know about that, but we do know we're mitigated anyways via mitigated either way the call response, protocol system or proposal system. uh So so, basically, essentially you know uh where we come to an agreement on what system context will be hypothesized or executed.

A

And then accept.

A

Contract on that.

A

Okay, so if agent buys the system contracts originally minted, within this example, we have agreed to contract, which says they will hypothesize or execute system context outlined in there, as outlined via their response system context. Counter proposal.

A

Okay, you know so uh counter proposals evaluated uh using evaluated uh using.

A

By mapping by mapping or by uh analysis of historical data agent execution data to calculate risk, tolerance, review system to calculate risk, tolerance, counterproposal is evaluated by analysis of historical agent execution data of historical proposed agent execution agent with counter proposal.

A

Data to calculate risk tolerance, you know, based on.

A

A based on review system of previous quality of of execution or via uh attestation.

A

Via attestation right so we could, um you know they could basically prove to us that they have have the hardware somehow right that they they could. You know they could issue a counter proposal with an attestation within the counter proposal, which is a valid attestation which says you know hey. You know I have um you know I have uh you know the hardware that is a test. Is it tested to run this thing.

A

And I you know have not have not, I I you know basically because you're you're, you're you're you're outsourcing your work, the work that you want done right, and so you, basically you have to vet it somehow right, and so you can do you know you can use whatever mechanisms you want to do that vetting right. You can pull in. uh You know third-party data, that from people that you trust to say, hey. You know who you know have people that I trust.

A

So if it's google that comes back and bids on it and you're saying create me a vm and google comes back and bids on it says I'll, create you a vm. Well, you can. Basically, you know you might have a strategic plan in there. That says uh you know, accept, auto, accept offers. You know that are within these areas from these you know or auto, auto let these offers through the gatekeeper right, um because I trust uh you know, I trust, google to run a vm.

A

um You know you may or may not trust rugel google to run a vm, but you may you know, trust google to to at least you know, spin up the vm for you. You know you you. If that's what you know and obviously you know you have different threat models for different things and you need to and you're going to take that into account within the evaluation um and and that's probably happening within the parent system context. You know these these uh or within the system contact or so the the plan.

A

The plan that issued or no the prioritizer, the prioritizer would prioritize, which system context gets executed as and and also those counter proposals. So those counter proposals counter proposals, counter proposals, uh you know so counter propole.

A

Are fed directly into regular execution.

A

Regular data flow execution, loop.

A

Oop and prioritize.

A

And ran through gateway, uh just like anything else, I can apply overlays to do uh context, slash threat model, specific uh vetting.

A

Right becomes trivial, so.

A

Great all right.

A

All right, I think, I'm going to get outside.

A

No, it's not going to rain.

A

It's like a rain, we're going to start the blockchain all right. Okay, so we found out nfts we found out. Nfts are a good way for us to go and we don't want to store those nfts in it in a did based blockchain. um We want to. uh We found out some ways that we might execute dots in a distributed setting um and uh you know, and so- and so you know, obviously uh you know uh yeah um so.

A

And you know it should be noted that basically you know it may be needed for the or.

A

Yeah they issue the counter proposal, which is effectively just another nft. So the counter proposal.

A

uh You know I got a proposal.

A

Enter proposal is effectively just another nft minted and then appear did linked to original proposal.

A

And I think, uh can we accept when we form linkage right? um So you know uh you know, does the did com message require signing.

A

A

When original context, when did for original context, uh signs uh linkage, peer link is prompted so, when did for for original context is prompted.

A

To sign peer link is where we are. We do our vetting uh right so get keeper.

A

Right and then we uh form link form peer, slash link. If we accept right so basically, then, then you do your you know, and so you can pay via. um So then you pay. So basically they say here's this system context. I want to link it. You know we'll link the dids, basically by saying well mint one ft, we'll mint an nft right. We watch fermented nfts, I'm in the nft system context.

A

um We uh then you know other agents out of the network say: oh okay, yeah I'd like to do that. Work sounds good um and then they, you know mint their nft. That's a response and via the the peer peer did interface, you have to have one peer agree to be linked to the other peer. I believe this is how it worked, and so when we do that agreement, that is our point where we can jump in and run our vetting flows and uh which we can overlay on there um or whatever.

A

A

Once payment is received, um agent will start work and then uh ask mint, nft and then do whatever what the results was asked all right. So maybe you're meant to another to maybe mint another ft with results, maybe never hits chain.

A

Just run that data flow.

A

Right: okay, oops sorry, ruffles stay very into his tail poor guy, okay,.

A

Okay, all right, okay, I need a break. I need a break we'll be back now. I have to change this again, all right. So what was this uh nfts so uh thoughts.

A

As nfts, instead of thoughts on nfg's yeah thoughts on slash ass.

A

Right because our system context will be nfts, which are our thoughts all right, see y'all soon, bye.