Fluence Labs Conference talks, 24 Nov 2022

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Peer-to-peer compute: the missing piece

Description

The talk by Tom Trowbridge, Fluence Labs co-founder
Singapore,
FIL Singapore conference 2022

Original source: https://youtu.be/FfU0fSII0m0?t=6746

A

Foreign, thank you um it's great to be here. Thank you, Dewan. Thank you to the Falcon foundation and protocol Labs team thrilled to be here. I'm Tom, Trowbridge um co-found influence Labs. We are a peer-to-peer, compute protocol, an application platform. You may recognize some of these investors I'll, obviously point out protocol Labs, very important investor for us and excited to be partnered with them. What is peer-to-peer compute so that may be academic, but I just want to level set with the audience here.

A

It basically is compute without centralized hosts or the cloud, and where does most compute happen? It happens in centralized host or the cloud right. So that is, you know, a very important limiting factor in compute today. How do you do peer-to-peer compute, they're?

A

Basically, two options: you can either do unchain smart contract compute and that's what most people think of with peer-to-peer compute or you can do general purpose compute and it's worth realizing that the smart contract based compute is slow, it's redundant and it's expensive um versus, and it's also a tiny fraction of the overall compute. It almost doesn't even register and I'm going to show something later about that, so we're focused on the general purpose.

A

Compute, which is basically 99 of the web, importantly also realize that where is where's, ipfs storage data, it's not on chain IPS ipfs uses blockchain for incentive mechanisms and to compensate, but it is not stored on chain. So general purpose, compute is where we think the future is. Let's back up for a second, what's wrong with centralized compute. Well, you've got companies that do it, and then you run the risk of censorship.

A

I think everyone is aware of people that have been either banned or Shadow banned on Twitter, and that's just one example of the type of censorship that any particular company has is completely within its authority to do, and frankly, exercises a lot, the other clouds and with clouds you have massive dependency and so AWS Azure and Google control 60 of the cloud. There is actually something regulatory released recently in the UK where they have 81 percent of cloud.

A

Revenues in the UK are those three companies and when that's the case, the companies on the cloud have massive dependencies. Next Issue, which is related, is de-platforming. Centralized applications can kick anybody off and clouds can also shut off applications. Now, why would a centralized? Why would Twitter shut somebody off or why would a cloud shut? Somebody off government when companies get large governments have to get involved and always do you can go back to any industry the larger it gets, the more relationship with government.

A

It has whether it's media, whether it's railroads, it happens throughout history and it's just the way things generally operate, and so we've seen government censorship, Drive, WhatsApp decisions and censorship in India we've seen parlor taken off I, don't think people's hear lives have been impacted negatively by parlor being de-platformed, but it shows what governments can drive and that more and more Revelations come out. It seems every week about the government influence on all of these platforms, but also, what's important, is what we don't see, which is the closed systems that do not create.

A

The renovation hasn't happened that otherwise might have, and so Twitter and Facebook closed off their their apis around 2018. Farmville, you can think of. You can think of all the Twitter applications that used to exist. Everyone in this room. Think of the Innovation, which has happened since 2018.. If those platforms were open, it's hard to conceive of what other applications and Innovations and usage we might see on those platforms, and so that's within their right to do no doubt.

A

But it's hard to imagine what could have happened if these platforms had retained their open apis and made it safe for developers to build on them with confidence and that's an issue that is really core to centralized compute. Also, de-platforming risks don't exist just in web 2 companies web 3 companies. A lot of them rely on the cloud right now.

A

The cloud by the way is not some puffy white, benign Cloud I like to see it as a storm cloud with lightning, because it's dangerous and that's how I will always show the cloud infuria relies on AWS as we know, and centralized even even a lot of applications. We know, like you know, snapshot as an example rely on centralized servers and it's no one's fault.

A

It's just that peer-to-peer compute hasn't evolved to a level that people are comfortable, deploying it and is this the internet that we want I, don't think this was Tim berner's Lee's, you know vision for an open internet, but the problem is that the internet rewards scale and the winning businesses become massive and when they become massive, governments have to be involved and the companies have to listen to shareholders have to grow and have to avoid jail right, and so, when you have those type of incentives and the governments recognize the strength of these companies, there is going to be a relationship, and so that is something that we that is sort of inevitable.

A

And so the question isn't: how do we stop that from happening? It's how do we create an alternative that that basically, governments will not have that same leverage point to access, and so how do we get here? We got here through um through basically an evolution where people started hosting internally we're drag Kicking and Screaming into the cloud many companies resisted for a long long time going into the cloud.

A

um I know myself and lots of others were very excited to invest back in 1998 in data centers, and that was way too early as companies hemmed and hot about moving to them. Now, that's the default, but peer-to-peer platforms are the next Evolution and probably the final evolution.

A

So why is open peer-to-peer the future? It's infinitely scalable. It is higher security and has better resilience. So, let's talk about each of those in turn quickly. Scalability, if you have a hardware Marketplace, you can respond in the market globally, can respond to price demand and demand a much faster than any centralized company, and so anyone can contribute Hardware to the network and turn it on or off. Given price signals, I, don't think anyone would think that a global network is less efficient in responding to demand than any one company.

A

No matter how big that company is, you also can tailor offerings with a Marketplace to users, far more specifically than one company and security. You know when you have a peer-to-peer Network, it's far more censorship resistant, because there's no one owner, it's a network and anyone can host or use, and so there's no legal entity to compel there's no management to threaten no directors to influence and no shareholders to pressure so you're, far more censorship, resistant and you're also far more hacking resistant as well I think people haven't focused as much on.

A

Not only can companies be be vulnerable to explicit government influence, governments have infiltrated. Companies and Twitter has the long-standing case with Twitter, where Saudi Arabian agents have effectively effectively infiltrated Twitter to docs Saudi Arabian activists leading to to jail for for several of them resilience.

A

I was thinking here about resilience in terms of the network, outages that we've seen happen with Amazon with the cloud with Facebook Etc, but there's also other kinds of outages which are deliberate outages and again, when you have a centralized, compute and centralized Network, it's far easier to shut it down with a decentralized network you're protected from both inadvertent human error, but you're also protected from explicit, um deliberate attempts to shut down a network, and so that I think given what's happening in Iran happening in many places around the world is bringing to the Forefront the absolute need for peer-to-peer Networks Where Do We Stand.

A

Well, we have peer-to-peer networks in payments and in storage, but we're missing the compute layer.

A

And let's talk about traction, you know for peer-to-peer payments, I think everyone is very familiar and knows that Bitcoin and ethereum and many others currencies have now a very long history, significant assets and lots of volume that problem or that that use has been taken. Care of storage.

A

We're all here, learning about ipfs and filecoin's terrific success in bringing on usable storage with over 200 petabytes of data stored and much much more coming, but what about compute and the compute Market's hard to gauge, but somewhere between 100 and 400 million or so right now growing to about a billion in 2030.? Is that a big number or small number? Well, the cloud last quarter did 126 billion in one quarter. Now this is not exactly Fair because it includes content. Delivery services include.

A

Storage, includes a variety of services that are not Apples to Apples with just smart contracts, but you could quadruple it or cut it by 10 and it's still shows how incredibly tiny smart. Smart contracts are in comparison to the cloud, and that shows why there needs to be another solution to peer-to-peer compute Because by the way smart contracts aren't new right, they've been around for what since ethereum launched. So this isn't a brand new technology here and why, then, is open peer-to-peer compute hard?

A

It has to be decentralized, verifiable and scalable, and the problem is that being decentralized and verifiable without blockchain and without being Unchained is difficult, and so what fluence is doing is using combination of a web of trust, Quorum and ZK proofs, and that gives us the scalability so we're not bound by the smart contract constraints by the on-chain constraints, and we look at the competition globally as and where we want to compete with, ultimately is Amazon web services in the cloud, not blockchain, not smart contracts, and then what does? What do we get say?

A

You know when we're successful and everything is launched and humming. What does that get us? Well, you can then see in an open, transparent, Cloud what is executed and where and by who and the code can't be shut off. So that means no closed ecosystems and any code uploaded can be used and can be built on that is incredibly powerful. That is I, think a creativity accelerant, because developers can build on other applications.

A

They can build on top of existing code and that open source authors can share in that hosting Revenue, because it's clear who is using what code that is and that's a whole talk in and of itself, but happy to discuss it further and also, very importantly, it's Unstoppable as long as there is one peer available in the world to serve the code or the application that you're looking for. You can get it. That is hard to underestimate the importance of that feature of open, peer-to-peer, compute now for filecoin. Why does it matter?

A

Why does it matter for falcoin miners? Well, because filecoin obviously is well suited to incredibly large data sets, as we've learned when you have a large data set you if it's useful data you're going to need to run computation on it and computation is easier run by the storage provider and storage providers already have this computation sitting.

A

There they've used it to seal, so this unused computation is wasted currently, and so we expect storage providers to have a new model or an enhanced model where they can monetize their unused CPU resources by selling compute services to on top of the data that they store, and so the computation can be run locally.

A

On top of that data, and if you think about what does that data mean people here may know better than me, but we're talking about satellite imagery, video medical data and what is compute on that mean it can be removing blurring faces in video, that's computation, it could be moving moving away clouds from satellite imagery. It could be. You know, running studies on on data or determining in out of sample efficacy.

A

Large amounts of potential there and far easier to do on the data where it is and far faster. We're convinced that compute demand is going to grow dramatically. The data on filecoin is growing and we've seen those numbers and the more data that's stored. The more computation will be required and we think that storage providers are going to be incented to and will actually earn a great return selling their unused computation to these providers. If you think about it, we've already seen auctions for unused storage, where there's negative storage costs, as there's unused capacity.

A

Well think about that for computation, you've completely unused computation across the filecoin network, so you may ultimately get to a place where quite quickly where storage providers can offer computation at a minimal price. It probably won't be negative because you have to pay for electricity et cetera, but there will be. It will be by far cheaper than computation offered by the traditional cloud, and when this happens, um some of that Revenue will be shared with the authors, which will drive usage and and creativity and modules and more um uh useful compute.

A

On top of that data, and when you combine the storage of the filecoin networking of ipfs with the compute affluence and other providers, you end up with a decentralized cloud, and that is incredibly powerful and that's what I think we're all working to build these individual pieces are terrific, but we all need to figure out how to put this decentralized Cloud together, because that's the real goal, and so where is fluence fluence is here.

A

We've already launched the peer-to-peer Network we've launched our language Aqua, which basically abstracts out the complexity of authoring and composing peer-to-peer applications. So this isn't a 10-year out Vision. This is existing now and this year we're launching a Swiss Foundation, which is going to be our Dow governance and next year, the compute economy, which will provide the incentives for um for storage providers to charge for compute and for software providers to be compensated as well.

A

So what kind of world do we want? We want three companies controlling internet hosting and content moderated by nation states. um With you know, creativity, stifled and huge barriers to building independently. We think that you know incentivize open source is the only architecture that can work, and so I think everyone here wants to support. Peer-To-Peer and I think that is, you know the size of this conference. The size of token 2049 show that there's interest in this and I just hope. Everyone recognizes the importance. Why are we doing this?

A

We want to build this because we don't even know the type of creativity can come afterwards. It's as if someone built you know, building the HTTP protocol could imagine Uber no chance right, but when we build open systems, the creativity and the applications which can develop are very hard to imagine and that's what fluids. What we're trying to do is to empower developers and Empower this next future of creativity. So with that, thank you follow us on fluents.network on Twitter influence underscore project. Thank you very much.