Bittensor Novelty Search :: Network Governance

listening to that launch which was insane. We really didn't know what we were in for. It worked though. It was incredible and yeah so one whole year of DTO guys.

And quite a revolutionary idea to take a blockchain of one token and turn it into 128 new tokens in an instant. And of course, everyone who participated in that year knows that there were some ups and downs. We went from an incredibly low volatility, sorry, a low liquidity regime to now a hyper high liquidity regime.

With very little MEV resistance, a functioning market, tens of millions of dollars of daily volume or millions of dollars of daily volume, tens of thousands of Tao per day. We've had Ponzies. We've had financial schemes. We've had rugpulls. All the things that people said that were going to definitely happen definitely did happen.

All of the negative things that people said that were definitely happen. The rugps and the ponzies and we knew going into dynamic tow that that was going to happen because there's just no avoiding it. If you want to have permissionless markets, if you want to use the wisdom of the crowd to solve a problem, you're also going to have people.

And you're going to have those people run scams on other people. And, uh, that did happen. We had 10X. We had Subnet 28, a very initial uh Ponzi idea, you know, pay to mine. Which we hammered down and there was a lot of scams. The word scam was thrown around a lot, which was good because I actually think that and I tend to lean on the side of, you know, being a more toxic community in that sense, being willing to critique.

We had all of that happen. We had rugps. We had financial schemes. We had people play around with different economics. One thing that we saw was the subnets learned that they needed to burn some of their minor mission in order for their subnets to make economic sense.

We didn't stop that because we couldn't without too much oversight from and we also believe that in the end it was probably beneficial for these subnets that went from a regime before DTO where basically everyone was just incentivized to spend as much money as they could you know just take the tow mission and just burn it.

We've really really really tightened up shop. The subnets have learned how to build mechanisms that cap the cap the burn uh like the amount of cell pressure that they have via changing the incentive landscapes. So it's not just automatic burn.

We've got Penna from Nova here. So, Nova learned this one very interesting lesson about how they had an incentive mechanism which just forced miners to put even more and more compute and it create uh created a lot of sell pressure on their subnet going into about halfway through last year. We learned those we learned those lessons and actually we tighten up shop in such an such a massive way.

It's like the the valve is now opening up again for miners as you know teams are turning what their digital commodities were into productive assets in 2020. three 2024 uh we had subnets that were just burning emission for they were not burning emission they were just spending on open AI credits that was the best subet on Bit Tensor was an open AI credit subnet no offense to to Mog and team that but that was actually better than the subnets that we had at that time a lot of them were broken we didn't know the art uh we didn't know the art of of of building digital commodities.

And then and then the best subnet was an inference subnet and it was a very narrow band inference subnet. It was a fast inference subnet and it had issues itself. And now we have now we have shoots.

We have we have general purpose inference. We have an infrastructure that is that is servicing more more queries for open source models than any other provider on earth. Which is unbelievable and that's done in a decentralized way in an economic in an economically sound manner. That was a consequence of moving to dynamic towel.

Sure we have minor burn but now we have functioning economics. Another change that came out of dynamic town. Another thing that we had to learn was what really has value. People started questioning revenue and and it wasn't it didn't come from the top down.

It wasn't validators. It wasn't groups of individuals that had vested interests in the ecosystem. It was actually perfectly incentive aligned individuals. You know, the DJ community, the Price Chats, the Sunnet Summers, the Fruit Brigade. I don't know if you're in the DTO channel.

They came to actually run Betensor and there's a lot of weird characters in there, a lot of fantastic characters, a lot of not very smart people, a lot of really smart people. And in coordination together, they've they've made Bit Tensor itself intelligent. In a way that would have been impossible for validators to do.

In fact, comparatively, when you look at where we were with root network to what we created with this functioning market, there is no comparison. It is maybe a hundred times more efficient at allocating towel than it was more than a year ago. Well, just just actually just under a year ago.

And that came from the coordination of many aligned incentivized individuals who thought who looked at mechanisms who found teams who who called out scams who promoted projects. One of the first things that happened also was that instead of promoting internally and this was something that happened last year at the end of root network everyone would just come to us right the valuers hey this is what we're doing this is what we're building look at this now we're fully pushing out into the into the world right so every subnet team needs to have a marketing uh department or a marketing plan a way to go a way and going of going to get themselves out of this ecosystem and we've seen subnets that have gone and integrated with they're they're on bit tensor but people don't even know they're on bit tensor Shoots is a good example. Synth is a good example.

These projects that actually build they build tools that are valuable generally valuable into the rest of the world. And we've captured a lot of that value. It's come back into Bit Tensor. Of course the miners have gotten a lot better as well like a lot a lot better.

We we we didn't think that was possible. There's now mining companies of 50 to 100 people that in different countries in the in the world that run there's organizations that are open source that that run create open source mining software. There's probably tens of thousands of individuals across the globe who have mined bit tensor in China in India in Europe in North America and South America.

People that live off Bit Tensor that that spin spin their wheels day in day out optimizing these things. We we also hit some benchmarks just in the field of decentralized artificial intelligence that nobody in the world has ever achieved and we're we're so far ahead that you can't even see our competitors in this in this regime in this regime including decentralized training.

We have the largest models in decentralized training. We have the best algorithms in decentralized training. We have the best algorithms in decentralized reinforcement learning in in inference and verifiable inference. Just about anything that you name it, Bit Tensor has proved that if you want to solve something, if you want to solve anything fast, if you want to solve it best, if you want soda, build an incentive mechanism and let the mining communities on Bit Tensor solve it for you.

And we've done it in a way that we don't even care who those people are. I know for a fact that there's some pretty shady organizations that mind Bit Tensor all the way from Lazarus Group in in North Korea to hacking teams in in in in Russia.

And I I say to those groups you know, we welcome you in some respects to to align yourself and build real commodities instead of being nefarious organizations. Maybe we've done some good in the world via via via that dimension. I don't know.

I actually just don't even know what's going on because the minor community is actually fully blackboxed and that's part of our ethos. It's permissionless system and it's been permissionless for good or ill. There's been a hell of a lot of chaos.

There's been teams that have come on that have built absolutely terrible things from from you know not terrible things. We haven't actually had any cesam yet which is fantastic but if we did I believe that the community has proven that we would back we would back together. we would work together and get rid of that type of thing.

We we've we've we've proven to be libertines but not sorry we've been we've proven to care about liberty but not be libertines in entirety and and that shows itself by looking at the top 20 subnets on Bit Tensor. Below that it's a bit harder like it's a bit more chaotic.

But the top 27 bit tensor are functioning functioning projects with with real code beating best in the best best in the world and their particular domain pushing out to the world. And that alignment came from this community. Didn't come from me. It didn't come from Mog. It didn't come from Allah. It didn't come from any of the validators on Bensor.

And so I would say it's a massive success in proving out that these economic fundamentals, econom economic primitives of markets can work. Cool. So one year Ally, anything to say about that?

Honestly, it's like I was telling you earlier const. It's very interesting to see these power law distributions kind of happen in real time, especially when you look historically at the earlier subnets like when we first created evolutions branch, right? That was back when what 2023 isish. Yeah. When we had the the Kazuna coins of the world kind of join. We had all kinds of the exact same stuff.

And the community ended up sort of rallying together and building the really really interesting stuff that even I couldn't even think of in the first place. And you know sort of DTO came hit everybody the head with the hammer but also was a very needed hammer and ended up kind of in a way resetting the status quo.

And it's really good to see the exact same progression happen from the very beginning up until today. It's it's extremely extremely inspiring. Really really makes you proud to be part of the community. But at the same time, it's really good to see that we're getting better at this at the holy.

We're getting better at decentralized AI. We're ahead of everybody really and really right now it's all about momentum and keeping that up. Another thing too is also is also integration. You know shout out to the shoots team right top on bid tensor 18% emission John Durban fantastic developer better than me by a thousandfold who has come into this ecosystem and taken charge and and also the the team around them they were on novelty search last week building technology that can be the foundation of an integrated bit tensor you know there's no I said this a number of times there's no incentives in bit tensor for direct integration that you don't get more mission if your subnet uses another subnet.

But we have the teams on Betensa that are building towards each other and on top of each other and as foundations of each other and stacking lay layers of digital commodities all the way down to the just the compute substrate, right? like the the raw hardware all the way up to high level fed high fallutin agentic workflows and it turn looks like shoots is really playing this playing that role very well as a backbone I use shoots every day you know building a aentic loads running LM inferencing and so you know I think it's also a testament to bit tensor that the subnet that is the most integrated the one that is the most interconnected with all others in bit tensor is the top subnet on bit tensor and that's what we should hope is the case that the subnet that being used that is integrated into the ecosystem the most is the most successful and and well-rewarded one and that's that's not written into the code that's epiphenomenal of this ecosystem that we're building so yes I'm not here to talk about ships actually I'm here to talk about betensor as a whole so last year was dau killing it scary you know it it went up it went down we hit one everyone freaked out and now it's been a very steady growth since some of one uh in a great way we upgraded from from the original price base to TF flow which added some more dynamism into the emission distribution. Some people don't like it.

But I believe it's been over overall quite positive for the ecosystem. And now it's another year we're one year past and so what what what is Bit Tensor going to be going forward? And I want to hearken back to when we wrote the detail paper. There was actually a couple sections here in that paper that we didn't implement and there's the grander vision for Bit Tensor that needs to come to fruition that needs to be built and it's not what it is right now.

So right now Bit Tensor is it is actually not decentralized. runs the chain nodes at great cost basically funded via via donation from from Allen and myself and also from people that are delegates to our validator. Thank thank you so much. It runs a lot of infrastructure. It runs entry points.

We fund the the chain development. We we are the main developers. We we spend multiple millions of dollars per year hiring developers, maintaining this this infrastructure. And and we've gotten to a point now where actually there's not that much more to do. Which is amazing. It was falling forward for many years. Alec can can can give testament to that.

We were always upgrading and changing and fixing new things and pushing the chain faster as fast as fast as we could and probably we were the fastest upgrading blockchain in the world at any point during that period. I don't think there is a blockchain that had as many upgrades as us and we really took advantage of the polka dot substrate that we had.

But this technology is so important that it needs to be decentralized because it's something that can creep up on you and eventually we we discover that oh my god you know there's governments that want to stop us. There's people out there's really nefarious individuals that want to take out this chain. They want to stop what we're building here because it threatens monopolies and it threatens uh the power structures that exist in this world.

and and it's not now and we don't know when that will happen, but we know it will happen. And what's the point of working towards a goal if you're not being honest about what you're building? We should be honest about it. We're building decentralized technology. We're building something that is unshutable, unstoppable, that lives on the premises of its of its code that uh freedom and permissionless and liberty and ownership.

And these qualities are mostly there. definitely have come out to be this way in the subnet domain, but at the chain level it's not the case. We still run the chain and we still write the code and we've done a good job, but it's time it's time to really to stop that.

So I mean a bit of an announcement. I'm I'm stepping down from CEO of Open Tensor Foundation. I will no longer be associated with F. Actually it's already happened. So I am no longer F's boss. I cannot dictate the rules for the developers. I cannot run the validator. I do not have control there at all except perhaps through soft power and respect from the individuals there. Thank god.

And F is this year going to be decentralizing chain chain nodes. That means that there will be number of organizations that'll be running the chain nodes across across the globe. By the way, this means I'm not leaving the ecosystem at all. I'm actually stepping into the ecosystem. I'm going forward.

So I'm going to be maybe I'm going to be mining or maybe I'm going to be building subnets, but I'm definitely here. I it's it's fantastic news for me. Because there's nothing there's nothing more fun than building a game and then being able to play it and participating in this in this ecosystem, this fantastic fantastic ecosystem.

So yeah, I mean I've I've built a number of other subnets and I will continue to do that mining and so I'm not going anywhere. I'll participating in this in this ecosystem. But F uh with MIG gone is going to be decentralizing the chain nodes. That's the that's that's on our road map for this year.

I think look forward in the next six months. it'll be a nominated proof ofstake mechanism to begin with just uh uh running the chain nodes uh but in a nominated proof ofstake manner and then we will move it out to three or four organizations in a couple months after that and then finally it'll be an incentivized fully incentivized blockchain uh un unstoppable unshutable difficult to fix harder to break perhaps in some ways and and that that's it's important that we make that step because at this point you know Bit Tensor is under the under the eye of of governments that may not be interested in in in what we're building here and they know about us and they probably want to stop us or maybe they want to work with us maybe they want to arrange run chain nodes but we just can't take that chance and so it's important to be decentralized and become truly the Bitcoin of AI

But there's more than just the decentralization of the chain nodes. There's also the governance. And so this is actually what I want to talk about. This was the name of the the talk for today was talking about about governance.

And it's in the paper from dynamic tao that was written a year ago talking about how the first thing that we've done the most important thing we've done with dto is perhaps separating the validators from the subnets and as a consequence of this allowing for true decentralization of power separation of powers and we've built it so we have actually built out the tools on the chain that allow for true decentralized governance governance.

So this means that in 2026, by the end of 2026, we will have a decentralized governance and a decentralized chain. That means that every single one of you will be up for grabs, so to speak, to to participate as the controllers of this of this ecosystem, which is the way that it should have been.

And a bit of a tangent here. The the idea of having subnets actually came from us at F. going underground and looking at the ecosystem so to speak with pseudonomous identities and asking ourselves, hey, why are these guys running this [ __ ] Why are they only the ones that can make subnets? And it was because of that conceptual flip that we let we let subnets be built on Bit Tensor, which was an incredible revolution for the technology that is Bit Tensor today.

And I believe that this this can be a similar a similar upgrade, a similar revolution in the ownership of individuals. So every single subnet own subnet owner on bit tensor will have the opportunity to be participate in the triumvirate which is the three individual group that does the initial ratification of upgrades.

We're going to be using a form of blind voting a random voting where the some of the top subnet owners will be selected at random every six to eight months to replace one of the previous triumvert members. So that will be cycling. In addition we will also be pulling from the top validators and they will have the opportunity to run on the triumvert in exactly the same way.

This this means that also all submit owners and all validators in addition to being able to run on the triumphvirate will have the ability to hold votes to delay and stop chain upgrades of any type. So this means that subnet owners that are in the top 25 or so subnets and the top validators that are in the top 12 or so validators will have true authority over what is built with that in conjunction with a decentralized chain.

means that Bitensor is truly leaderless, truly decentralized living organism out there beyond us, beyond f beyond the foundation and myself. An opportunity as well, right? In addition to being decentralized in this way, it means that that there's way more participation on behalf of every single person that's probably on this call. Your voice matters.

you can run a subnet, you can get on the triate, you can control this. And that's what we wanted to build right from the very beginning. We set out here to build an AI company, a decentralized AI company that could own artificial intelligence, own the best-in-class artificial artificial intelligence, and make it open, truly open, because there's nobody, there's no company, there's no organization, there's no government that will allow that same thing.

They're all going the other direction. They're all going towards decent decentralization. None of them are going to build it so that you can control it truly. And we're going to be the first to do it. We're going to be the first to make it so that every single one of you every single one of your voices matters in the same way that every single one of your voices mattered during um the last year for for DTA.

It's going to be this way for the governance of a tensor from the ground up. Yeah, we have the technology and yeah. So yeah, guys, everyone's saying you're gonna continue to say I'm not going anywhere. I'm not going anywhere. I'm actually going to be more here more involved, more coding, more building subnets, more speaking in my mind.

In fact, I I actually hope that as a consequence of not having this position as someone that does have, you know, like let's say veto or absolute totalitarian control, not totally, but I do have a lot of control like physical legal power in Bitensor that that I'll actually be able to express my voice even better now.

And people can't just say, well, hey, you know, you're in a unique position to to you know, the chain isn't decentralized because that's that won't be true. I won't have any press special privileges aside from the fact that I do have a voice here and I know this technology inside out not disappearing and so you don't need to worry about that. Great.

So one of the first things and this is actually maybe where some people are going to outcry. It was all very inspire inspiring up to this point. But one of the first things that is going to be on the table for governance in the next 3 months is that we are going to allow votes on the stopping of emission on some subnets.

There are currently on bit tensor a number of subnets that are u manipulating incentive taking away from other teams. An example will be 104. Subnet 104 is forced validators out of the subnet. There's not even rooe emission. it's running a scam and it's pulling away from the real builders on Bit Tensor.

One of the advantages of us having a decentralized governance of having these tools on the chain is that we as a community can defend ourselves because up until this point it's just not been correct. It's not right for the author authoritarian control without that being liquid, without that being decentralized. It's not right that we would make such decisions that affected individuals and but we can do that through a proper governance system that has veto power and is decentralized and people can vote and people can disagree and people can unstake and people can can can sell subnets and val and unstake from validators that they disagree with and the voices that are made and the opinions that are put forth.

At the same time, it also will free up some emission for new teams and make it so this this project is is uh pushing even faster, exponentially faster, into the proper teams that are producing digital commodities as aside from the ones that are extracting. We will continue to hone this ship and we will continue to make this technology as fast as we can and better every day.

And I think that we're going there and I really hope that this upgrade is alignment with what the community wants. You know, it's difficult. It's it's a little bit it's a revolution in in Bitensor, but you know, we're we're we're giving it back to the community. We're giving it back to your voice.

And we think we believe in you guys. We believe we've seen how the the community, the detail traders have turned this technology. We've made it faster through your your collective voices. So, I think that's going to continue.

So, that's the first thing that's going to go on the plate and that's exciting. That's probably going to happen in the next month or so. So watch out if you are a subnet that is exploiting. There's probably going to be a vote to to pull out your emission and and save Tao and and redirect it into the into different different places.

I mean, hey, if it's 120, let's do it. I'm ready. It's fine by me. Yeah. Okay, cool. Well, that's that's my spiel. All, do you have anything to add?

Um, maybe we do like an AMA or something and people have questions. The only thing I'm going to add is say that 120 kind of sucks balls. So, I'm glad it homeop people kind of vote you up. That's going to be good. Very happy. Put out a vote. Put out a vote by 12.

No, I was I think it might be good to explain to people as well just how gradual intentional this is going to be. Maybe via like a just a general timeline of how this is going to look like. Like you're obviously not we're obviously going to be like decentralize everything tomorrow. It's still going to take a little bit of time.

Exactly. So the first roll out is actually a very minor upgrade allowing this vote to occur for subnets a subnet cutting their emission via validators requiring more than 50% vote. The more will be spoken about that specific upgrade. It's not correct. It's not quite ready for too much discussion about how that's going to happen. That'll be in the next month or less.

As for the full governance system involving subnet owners and validators and triumphvirate and voting for triumphirate and randomized voting and and the design that is in the detail paper coming to fruition that is in the works. It's actually ready to go, but we're going to take that out to an the end of Q1. So maybe actually that's coming up. Sorry. Perhaps maybe three three months from now.

And then the decentraliz decentralization of the chain on top of that let's say 6 months that's dependent on you know really internal testing we we have to be careful about the centralization of the chain because if things break you know it can be very difficult to put Humpty Dumpty back together again when in a decentralized environment.

But we're we're very confident that we I mean we're quite confident we can do it maybe even faster than that. Technically we could launch all the stuff today. But but we're going to do this differently. We're going to do this properly. We're we're going to bring people in. We're going to build the tools. The culture is here and that's the hardest thing. The technology is just about here but just springing it on people is not a good idea. So so let's say 6 months and yeah thanks that's a great question.

That makes sense. And do you want to give any details about how the voting mechanism is going to work or just generally if it's talated voting or otherwise? Yeah. Okay. So yeah, so we we built a system of governance that we believed struck a balance between speed and decentralized power.

So in in all governance systems on basically every blockchain out there, they've they've mostly all failed. and and we know this firsthand because it was one of the reasons why we moved away from the Polka Dot ecosystem was because their voting process was going to take 3 months to get a very minor vote through that we were required for actually running on the Polka Dot ecosystem and we just decided to go away from them.

The the truth is that almost all subnets fail sorry not subnets all blockchains fail at governance because they are far too idealistic and they move too fast. Rome wasn't built in a day, right? The democracies of that we see in the United States, that was built off many many many many iterations of bad versions of democracy and and republics.

And and humans have in fact needed a very long time to build functioning governance systems. We we we think we can move faster because we have all that knowledge and we're in crypto and we're in blockchain and we can move fast. But but we need to understand that that there's a there's a balance to be struck between executive power, how fast we can go, how fast we can change things, can we change things, and can people stop things, is it decentralized?

And so the way that we did that is is we actually built a voting based on there are proposals that are made and sent to the chain. Those proposals sit on the chain and they are ratified by the triumvirate. So that's the first line of defense needs to get to it of three votes from the current triumvirate and once the the triumvirate gives those votes it goes on a track and that track is basically sitting between the two houses of bit tensor the validators and the subnet owners and if the validators or the subnet owners are capable of getting 50% quorum that votes against the proposal then the the proposal is stopped.

So the the upgrade does not go through in that time. However, every vote of nay delays the proposal doubling the time it takes for the proposal to go through. At first it's very fast. So it's very it will be possible for the val the triumvirate in coordination with the developers that are making bit tensor to push something out and have it on the chain within 5 hours.

If there's nobody disagrees with that proposal, if nobody wants to vote to slow it down, if they look at the code and they say, "Hey, this needs to happen. We need a hot fix right now," that can happen quickly. But for larger upgrades, contentious upgrades, we expect those will be delayed. And and for for good reason.

And it will be great if they are delayed so people can have conversations about them. And if they're able to get 50% of the voting subnet owners or 50% of the voting validators, that will be blocked. So that is the initial design. And we think that we believe that this will allow us to slow things down when we need to if we need to stop if we need to oify betensor and I believe we we probably will oify betensor eventually.

there'll be possible for subnet owners to inquote for everything doesn't change and then it will be impossible to upgrade the chain and it becomes a fixed crystalline system in space that flies to the end of the universe right for a thousand years. That's what we want to build. Right now it's we're going to be heavy on executive power, the ability to move stuff.

So that's an aspect. Another aspect of the of the design is the triumphvirate will be voted in vial via blind vote. So there is actually no voting. You simply will become a triumphirate member randomly using block height. So there's no need for election campaigns. This is something that's like not possible right in normal governments.

But we can do it on a blockchain. We can use things like commit reveal and dand etc etc. So you have the opportunity to literally sit on the tri. You didn't even know that you were going to sit on the tri. You didn't vote for trying to get this triate. And this is actually quite a quite a powerful defense, right? Because it means that no no group of individual can can guarantee outcomes right that that they it's unless they are able to pack the full submit owners and and validators which is a very interesting idea that comes from from um from um you know kind of experimental governments.

And yeah so like that's that's the that's the general idea the gist of it. Obviously when we when we we push this out in in small pieces people will be able to explore every aspect of it and and make their voice known right so owners will have their voice known the teams miners everyone will have their voice known in this ecosystem and be able to speak and see it play out in real time so yeah does that answer your question is that a good description of it think so uh do we want to talk about how the proposals are created or is that too as we have it right now.

We will have proposal keys that are controlled by different individuals initially organizations that run a GitHub that is associated with the subnet us with with substrate and that we believe is running all of the necessary like upgrades and CI pipelines to make sure things are are passed properly. And then eventually our idea is to fully decentralize that. So so upgrades can come from anywhere.

However, all in good time, right? Getting this off the ground will be fantastic. And and it will, you know, I think heighten and bring a lot new energy and fire into this ecosystem. There's people that have ideas, great ideas about what they want to do. Yuma just came to me the other day about building a lending protocol on top of the pools. Very cool idea.

Right now they just can't do it. In the future they will be able to do this kind of thing. Hypothetically everyone can. And I want a point here is that if we get this correct, if we do this properly, it allows us if we decentralize, if we build the governance, it means we can do things. We can have privacy on on Bit Tensor.

We can't do that right now. It's it's too legally impossible. We can have privacy on Bit Tensor. we can prove that this this this these tokens are immutable. We can show that they're they're locked into a decentralized network. And this will be better than even what Bitcoin has, right? So Bitcoin doesn't have that. it doesn't have the governance around its repo and we will have that and and that will actually I think take us a step further than even Bitcoin got in terms of decentralization which is pretty exciting. Agreed.

Ag great um with regard to decentralizing the chain nodes I know that you know it's effectively expectably ready but do we know what protocols can be used? How are people going to be incentivized? What's that going to look like? Uh nominated proof of stake. So it'll be a proof of stake mechanism.

And the there'll be more details about how that stake will be delegated to those to those validators. It may be the validators themselves. It may not be the validators. It may be any individual. I believe it will be any individual. That's the the end goal there.

The there will also be transaction fees to I mean like has basically just been shelling, you know, footing the bill for six years, right? And it's been very expensive. So so yeah, there will be transaction fees and that's I think that's really good because it'll maybe mean the infrastructure gets better. We're looking at probably a couple million dollars of transaction fees as it is today easily.

So the transaction fees that are being burned, are those the ones that are going to be basically