BroadcastYouTube

qBitTensor Labs Live — September 3, 2026

September 3, 20261:05:51

Emission on both subnets was unilaterally paused this week via the root key after an AI-driven review — so the team spends the episode setting the record straight: the treasury wallet design the validators directed them to, the quantum jobs SN48 pays for in real US dollars, and the virtuous-cycle tokenomics the shutdown accidentally spotlighted; plus Xdev takes back-to-back Breaking RSA milestones, RSA-512 launches with a symbolic $250,000 RSA-1024 bounty alongside it, web submissions go live paid with Open Quantum credits, the Open Quantum Insights report draws a Rigetti repost, unprecedentedly thorough Quantum Insider coverage, and downloads from Dell and Microsoft, an updated white paper lands with a DOI, every winning solution is open-sourced under AGPL in the new Breaking RSA Solutions repo, and OpenRouter's $7.5B sale to Stripe shows exactly what Open Quantum could become.

A spicier episode than usual. The team woke up this week to find emission unilaterally suspended on both SN48 and SN63 — switched off with the root key following an AI-driven review of subnets. Instead of the usual agenda, the September 3rd broadcast walks through exactly what happened, why the systems under fire were built precisely as directed, and what the team is doing about it — before getting to a stacked slate of genuinely good news: back-to-back Breaking RSA wins, a new RSA-512 milestone, a symbolic $250,000 RSA-1024 bounty, web submissions going live, and an Open Quantum Insights report that made a real splash across the industry.

The Eye of Sauron: Emission Paused on Both Subnets

The team woke up to a post from Const on their subnets: emission had been unilaterally suspended — the root key logged in and turned it off, on both Enigma and quantum compute. No conversation first.
The trigger appears to be AI-driven: Garbos (Arbos) has been posting quasi-right, quasi-wrong assessments of subnets across the ecosystem, and this week decisions got made on top of them. Bob's framing: AI is subtly telling people what to do, people think it's smarter than them, and it's outsmarting their common sense.
The bigger pattern: the ecosystem's number-one focus for the past year has been disincentivizing bad actors — and that work has now tipped into actively detracting from innovation. Bob repeated his question from April: are we building walls to protect what we have when we could be taking what's here to the moon?
qBitTensor Labs isn't alone — other good subnets are taking the brunt of this too. And as always: the pendulum swings back.

Defending the Treasury Wallet

The history matters: Enigma's challenges have discrete, verifiable answers — not leaderboard-style "best result wins" — so the fire-hose emission model most subnets use doesn't fit. The original proposal was to route emission to a single key and distribute it; the validators said no, that a self-custodied wallet imposes rug risk on all of Bittensor, and directed the team to the treasury wallet feature instead. Per the Church of Rao devs, Const bankrolled that feature's development.
The team waited roughly six months for the feature to ship — helping push it across the finish line, and funding prize pools themselves in the meantime — then moved to it exactly as told. All of this predates the one-minus-burn rules that now make accumulation look suspicious.
Accumulation is the point: large dollar-denominated prizes attract outside innovators who would never touch crypto for its own sake. It's working — click through the bios in Discord and you'll find heads of quantum programs at major Fortune 500 companies, and submissions arriving from clearly non-Bittensor people that are far more advanced than typical miner output.
The irony: the shutdown forced the team to look hard at SN63's tokenomics and realize they're a virtuous cycle — the treasury builds, nothing is sold until a prize is won, and the subnet gains upward pressure. As Bob put it: they didn't fully appreciate how good the tokenomics were until someone decided they were too good and cut them off.

Defending SN48: The Jobs Are Real

Back in the TaoFlow era, Bob raised directly with Const that the math couldn't work for compute subnets. The directional guidance he got back: survive by convincing miners your alpha is worth more than dollars, so they hold instead of selling to cover costs.
So that's how SN48 was built: the team operates the miners, pays the QPU providers in real US dollars from US bank accounts, and holds the alpha. Thousands and thousands of quantum compute jobs on some of the most expensive resources on the planet — with none of the miner emission being sold.
The criticism that surfaced this week inverted that: no miner emission is being sold, so it must be a scam. That was the design — the one they were advised to build.

What the Team Is Doing

Engaging Const via DM — limited exchanges so far, and admittedly the team has been "maybe even too verbose" in its own defense.
Proactively committing more of the prize pools into active milestones, and shipping a website update that makes the earmarking of funds legible (more below).
Publishing proof that real quantum jobs run on SN48 — the Insights report landed a day before the shutdown and makes the case on its own.
Starting from Hanlon's razor: never attribute malice to what stupidity adequately explains. But either way — maybe stop playing with fire. Bob also connected it to a broader theme: leaning on AI as the decision-maker is just centralization behind AI instead of behind a human, and this week is evidence it doesn't work.

Enigma: Xdev Goes Back-to-Back, RSA-512 Launches

Whatever your AI says about feasibility, factoring a 500-bit key in four hours on an RTX 6000 across three validators is nothing short of an amazing outcome. Xdev has now won the last two Breaking RSA milestones — building on the open-sourced winning solution from what appears to be a totally different miner, then bit-twiddling it into real GPU efficiency.
That's the open-source flywheel working as intended: one mindset produces the first innovation, open-sourcing hands it to a completely different skill set. It's also why new milestones start at the introductory $10,000 level rather than piling money on incremental improvements — the market decides from there.
The new RSA-512 milestone is live. 512-bit RSA was a real encryption standard — breaking it in 1999 took a supercomputer two and a half days, and the best iteration since managed four hours using 36 CPU cores. Enigma's environment allows 24. Matching the milestone within tighter constraints requires innovation past the published state of the art — and winning solutions on Enigma are already in line with it.

RSA-1024: A $250,000 Symbolic Bounty

A $250,000 reward now sits on RSA-1024 — a number nobody in the universe, on any set of computers, has ever factored. Yes, the team knows the math: it isn't happening on a four-hour RTX 6000 run. It's symbolic — for now. The direction it points is real: pinholing challenge environments so solutions can call out to real quantum hardware through Open Quantum. When a quantum computer crosses what a classical computer can do, that's a key inflection point in society's security posture — and Enigma wants to be where it registers. If someone does break it, they get the $250,000 and a thank-you, because a solution like that resells for far more to the good guys.

Prize Pools on the Website, and Web Submissions Are Live

The site now leads with Total Capital Deployed — everything in the prize pools plus everything already paid out. Alongside it: the Prize Pool itself, Committed (money in open milestones, up for the taking today), the Prize Buffer (pool funds not yet allocated — deliberately large, because prizes are marketed in dollars but paid in alpha with two layers of volatility), and Total Paid Out. The old simulated ticker is gone; everything is read live off the chain with a last-updated stamp.
Web submissions soft-launched last Wednesday: submit directly through the qBitTensor Labs website, no miner registration, no wallet, no CLI. Fees are paid with Open Quantum credits and everything still flows through the same Bittensor process — solutions ride a self-managed miner run by the team, and winnings get distributed out to the actual solver. At least one early user appeared to be registering a fresh miner for every submission — this ends that.

Validators: Three Validations, and a Restake Ask

Wild Sage Labs has stepped back from running a validator. The intent remains at least three validations per solution: one scheduled on each validator, the extra chosen randomly. That matters at the edge — Xdev's near-miss solutions would pass one validator and fail another precisely because he was right at the boundary of what's possible. The next peaked-circuits milestone will likely require five passes, since a 50/50 solution can win by resubmission when too few validations are required.
The ask: only two validators on SN63 are actually validating — the rest child-hotkey or weight-copy, and if weight copiers outweigh real validators, the wrong answer can start propagating. If you hold alpha on 63, check who you're staked with and consider one of the two validating keys — qBitTensor Labs (0% take) or Rizzo (18% as of this week). Either is fine; the subnet's health is the point.

Open Quantum Insights: A Shot Across the Bow

The first productization of SN48's data launched Tuesday — and the emission was shut down Wednesday, which makes the report's existence its own rebuttal: thousands of public jobs across 63 countries, on six quantum computers from four vendors, aggregated into ground truth the industry has never had. Surveys exist; actual usage data did not.
Rigetti — whose Cepheus platform comes off very well in the data — reposted it to 50,000 followers. The head of partnerships at AQT shared it. And The Quantum Insider wrote its longest, most in-depth piece ever published about the Open Quantum team's work — unsolicited, in their own words.
Report downloads require identifying yourself, so the team can see who's reading: Pasqal, IQM, Dell, and Microsoft, among other big names.
This is the tip of the iceberg: uniquely differentiated data no generative AI has access to, with full rights to use it. The report is the free sample.
Also new: an updated white paper with a DOI number, so the platform can now be cited properly in academic research.

Miner & Validator Modernization

Work that predates the Garbos criticism — and partly addresses it — is nearing launch: getting miners closer to running jobs themselves, both simulators and real quantum hardware, in a verifiable, trustable way. Think "simulator in a box" with defined constraints, miners choosing where to run it, and mechanisms to verify the intended toolset actually ran — and that a simulator isn't quietly standing in for a QPU. The framework is designed to extend to other products and technologies over time.

The OpenRouter Comp

OpenRouter is to Chutes what Open Quantum is to SN48: unified access plus cheap decentralized compute behind one API. OpenRouter raised at a $1.7 billion valuation in May — and sold to Stripe on August 19th for roughly $7.5 billion. Open Quantum is the only one of these in quantum, and SN48 is already getting most of its traffic. But the comp only works with emission on. As Bob put it: why stop emission? Put it back. Let's frickin' do it.

Breaking RSA Solutions: Open Source Under AGPL

A new open-source repository, Breaking RSA Solutions, documents every winning solution — who submitted it, when, and what they did. A script under the tools directory handles relicensing: open-source components keep their licenses, original and altered work becomes AGPL.
That's the dual-track model: AGPL's viral license means companies who won't open-source their own stack must license commercially from qBitTensor Labs — and the intention is for licensing revenue to flow back into the subnet. No advance notice when it happens; when it happens, it'll happen.
Is commercial licensing a certainty? No. But the federal government has spent $40 billion on quantum computers largely because of what encryption-breaking is worth. A solution that shaves 30% off large-scale factorization time is a very valuable solution.

Community & Q&A

Price action and crypto winter: if emission is restored, the shutdown may prove a strange favor — it put eyes on the subnet exactly when the team realized how good the tokenomics are. The virtuous cycle relies on holders not losing faith; Bob is confident (not financial advice) that fair treatment plus the coming crypto recovery — which he thinks is much closer than 12–18 months out — leaves huge upside.
QPUs on SN63: a pure quantum challenge can't work yet — validators must decide legitimacy without human review, and while quantum computers still trail classical ones, you can classically pre-compute answers and launder them through a real quantum circuit (plus roughly 20 other attack vectors). Instead, quantum access will open up inside existing challenges via Open Quantum — and RSA-1024 is only ever falling to a quantum computer. Full stop.
Closing thought: AI has been quietly steering society since the first recommendation engines — give people what keeps them engaged, and polarization follows. Generative AI now subtly gives orders that people mistake for wisdom. It's the most powerful technology of this generation, and like every one before it, it cuts both ways. Keep your hands on the steering wheel — don't let it drive.

Bob Wold (BeyondNISQ) [00:06]

you guys ready for this? Yeah, no, no, yeah, no, yeah. I don't know if we are ready for this today. my gosh. What a fun time. welcome to today's qBitTensor Labs Live, everybody. it's always a pleasure to have you with us. let's dive into it. you know, as always, legal garb before we get started. everything we're saying is ideas, not promises. This is not investment or legal advice.

you know, we appreciate you guys not weaponizing this content or using it out of context. We might be a little bit spicy today, so doubly so on this episode. use the content for good. And if you're still here, then you agree and we're cool. Okay, so I'm gonna start off just recapping some of the things that I said back I guess in April, and then we'll jump into sort of more of the regularly scheduled content. This is nothing you guys haven't seen before.

But I I wanna maintain that Bittensor is so cool. It has so much potential. Every time I describe it to people, like in what it was originally built to do and how it should do it, they're like blown away and they're impressed and like it it's a beautiful idea.

But man, we gotta get out of our own fucking way, man. Holy smokes. we're supposed to be a decentralized system, not a centralized system. we're supposed to be incentivizing intelligence, not you know, making sure that everybody kind of fits the same cookie cutter mold. and so, you know, over time I I have become more and more convinced that like not only like the biggest obstacle to Bittensor.

is ourselves, but like kind of almost the only real obstacle to Bittensor is ourselves. And so yeah, you know, I think this week we kind of got caught in the crossfire. there's been a lot of work. I mean, it's basically been the number one focus of of the ecosystem for the last year has been disincentivizing bad actors. you know, trying to identify who's bad across these

Bob Wold (BeyondNISQ) [02:24]

128 subnets and all these miners and all these validators, like who are the good guys and who are the bad guys. And you know, it seems like this week we leaned on AI to make some decisions and there are some things that AI didn't understand. And I'm saying this all over the place, like AI is telling people subtly what to do, and they think it's smarter than them, and they're they're making AI outsmart their common sense.

And so all of this work that we've done to accelerate our work to disincentivize bad actors, I think has really taken away from any work that we're doing to incentivize innovation. In fact, I think we are now at the point where we're starting to be strong detractors from innovation. And so, you know, the question I asked back in April, and I'm just gonna ask it again, is are we spending all of our time building walls to try to protect what we have? Slowly.

cutting ourselves off from good potential value that could be created, new innovators that could be brought into the ecosystem, when instead we could be taking the awesome stuff that we have here and taking it to the frickin' moon. And I mean, I think I know the answer to that. Or I think I have some really strong feelings about this. And it wasn't in April, it was in May that I talked about how like, you know, pendulums always swing. This was when we were kind of in the troughs of TaoFlow and things were starting to look

pretty ugly. And it was like, hey, don't worry, guys. Like the pendulum always swings. And it did. It swung back, right? And for like at least a couple months we had a solid you know, set of subnets earning very, you know, very modest emission relative to probably what they could have been earning. if if, you know, as people started to learn the true potential of the subnets. But the pendulum had swung back. Now the flip side is

Bob Wold (BeyondNISQ) [04:43]

Pendulum swings. And so now we sort of have found ourselves overnight in a situation where, you know, we're kind of being targeted. And well, okay. I don't want to say we are being targeted. Subnets that AI has identified as not really understanding what they're doing are being targeted. I think we're not alone on this. I think there are other really good subnets that are taking the brunt of this. but this pendulum's over here.

The good news is that like this ecosystem is so freaking cool. It has so much potential. That like I have complete confidence that the pendulum is going to swing back and it is going to correct in the other direction. But man, this is kind of nauseating. so instead of our usual agenda, which always starts with like the awesome stuff that we're doing with the subnets, we're gonna spend a little bit of time talking exactly about what happened.

with the emission. then we'll go into a subset of the cool stuff that we were originally planning on talking about with the two subnets because so much of our time will be taken up by this. we'll hit a little bit of community, but to be fair, you know, like I would say like all of the work that we had done to because we're actually going to talk about like some real community sentiment shifts and, you know, things we were doing to work on that. But, you know, now all of a sudden we sort of have like

an outsider attacking the subnets. And so I feel like there's been this really rapid sentiment shift towards, you know, our loyal investors and followers and supporters coming to our defense, which actually has been quite great and we really appreciate that. So we'll spend a little bit less time on sentiment today also. Cool. All right. So I joked the last time we were on qBitTensor Labs. And man, I should have knocked on wood. I said that like

Thank God the eye of Sauron is not upon us. Well now the eye of Sauron is upon us. we are stuck in its gaze and it will not let go. you know, I'm not totally sure how we did how we got into this situation, but the eye of Sauron is now upon us. you know, we first sort of suspected something like this might happen when Garbos started going around posting random, you know.

quasi right, quasi wrong assessments of all of the different subnets. you know, I I know ShorShot actually came in and said, hey, you know, based on how they're judging us, there's actually probably some changes we should make to 48 to, you know, better align with what they're probably wanting based on what this AI is posting onto our thread. but we woke up yesterday, was it just yesterday, Omar? Yeah, that's yeah.

Bob Wold (BeyondNISQ) [07:01]

Time is flying. It feels like a year ago. But we woke up yesterday to a Const posting on our subnets that basically the emission had been unilaterally suspended. They had logged in with essentially the root key and turned off emission. unilateral, totally non-decentralized. and so on both on Enigma and on quantum compute. And so the kind of like too long version of that, you know, if you if if you cut through all the details and you say what actually happened.

It appears that on sixty three, this Treasury wallet, which you know, I think and Omar, tell me if I'm wrong, but I feel like the the signal was that Const had actually bankrolled the development of treasury wallets, is that right?

Omar [07:44]

That is what we were told by the Church of Rao devs.

Bob Wold (BeyondNISQ) [07:48]

Right. So this feature that, you know, we had actually sat for six months waiting for it to happen. Eventually we sort of helped make it happen. Const, you know, help helped bankroll it. we moved to it exactly as we were told. I think they don't like now how that works. And so instead of, you know, asking us to make changes to that, they just didn't like the way that it worked in this one-minus-burn thing that they've moved to right now.

And so they suspended emissions. So you know, okay, Devil's Advocate, it is it is very good from from a tokenomics perspective. Like that's not why we built it. We built it for like a very specific purpose, which is to attract large outside innovators to come into the subnet because there's large cash prizes, not just the people that are already here, but you know, and and it was working. In fact, behind a lot of the alts.

In Discord, if you like click on people, you might see people that are like heads of quantum programs for major Fortune 500 companies. Like, and they're not even being like that discreet about it. Like there are very big names in the ecosystem because this because of this design. Like it is working. but it has a side effect of accumulating the emission. Like, that's the point. It's supposed to accumulate the emission to attract people, but by accumulating the emission.

It then slowly attracts more emission because that's how the current you know rules are set up, which were developed after we built that. And so I think there's this fear that while we're not making a huge amount of emission right now, we're actually in a virtuous cycle that a lot of you haven't noticed, where we should keep the the emission that comes in helps raise our price as long as people aren't selling.

It keeps building and we actually should slowly start to raise. And I think they think that that's a scam. It's not a scam. It's designed exactly with them. We built it the way they asked us to build it.

Ryan (ShorShot) [09:52]

We've been transparent about like kind of the amount and and everything that that our goals are and with all the qBitTensor Labs live in the past, as well as the fact that, you know, we need enough in the treasury wallet to actually be able to pay out these pro these challenges when they're when they're when they're verified.

Bob Wold (BeyondNISQ) [10:15]

Yeah. But I th but I think the crazy thing is, you know, like like there's a there's a ton of bad action. In fact, ShorShot you posted on Discord, you that like a lot of like a lot of what we see is that people are like specifically trying to financial engineer and exploit like a money printing nature of the TAO emission. And it's like here, we actually built everything exactly as we were told. And it turns out that they changed the chain and it might accidentally you know, have this positive tokenomics thing of becoming a virtuous cycle. And so

Ryan (ShorShot) [10:25]

Okay.

Bob Wold (BeyondNISQ) [10:43]

yeah, really weird. But it's like, you know, our in in our mind, we sort of live in this like real human world where it's like, you know, if you want somebody to do something, like talk to them and tell them what you want them to do. Like, I mean, you know, we'd be happy to make adjustments. I mean I think we showed that by contributing all owner emission for six months into prize pools while we waited for this to come along.

Ryan (ShorShot) [10:58]

Yeah,

Right. And and like you know, it'd be nice to have more conversation and more direction and more clarity in terms of communication. That's one thing that's been lacking. You know, even when we're starting to explore treasury wallets when when they you know you know, six months or eight months ago, I forget what it was now, i it's like pulling teeth sometimes sometimes trying to get a conversation going with people. So

Bob Wold (BeyondNISQ) [11:30]

Yeah. And so anyway, that's kind of the story of 63. I actually, you know, we we'll talk more about what we're doing about that next. But I'm still pretty bullish that, well, a couple things. One, the fact that the eye of Sauron turned on it actually made us look at it and realize why the tokenomics are so frickin' good. Like we actually, to be fair, didn't realize that the tokenomics were that good until they decided it was too good and cut us off.

So, you know, if we can get them to turn it back on by like reaffirming that like we've only done everything that we've done exactly at your direction, right? And like exactly the purpose of this, I think Enigma could actually come out stronger than it's ever been before. And I really appreciate I appreciate Garbo or Garbus or what Garbi Gar Garbo for finding that and telling us about it. Cause you're right. The tokenomics are great on sixty three, like really great on sixty three.

Let's get it back on. okay. And then so 48, different problem. So actually going back to the TaoFlow era, when TaoFlow first launched, I was one of, you know, one of the like people were very strangely not that vocal about it, even though mathematically it was pretty clear that it couldn't work in the long run. And so I actually had reached out and had a conversation with Jacob on this one and basically said, like, you know how the math works. You can't have a compute subnet. Like eventually all compute subnets in TaoFlow will start to fail. And you know, the directional feedback that I got was, well, really what you have to do to survive is convince all of these miners that your alpha is actually more valuable than US dollars. So if you're a compute subnet, they have to spend real US dollars, and then they get your alpha, you need to convince them to hold on to that alpha instead of turning it into US dollars to cover their costs.

And so that's what we did. That's how we built 48. And so we operate all these miners. We pay the QPU guys in dollars, like in real US dollars. Like, like not crypto, not TAO. Like we take US dollars from US bank accounts and we cover all of those costs ourselves so that we don't have to sell the alpha. Like we haven't been selling the alpha. Like we've been doing thousands and thousands of

Bob Wold (BeyondNISQ) [13:54]

Quantum compute jobs on the most expensive resources on the planet, and we've been holding the alpha, which is crazy. Like, right? That's crazy of us to do. And it's so crazy, in fact, that the criticism was: wait a minute, you must be scamming because none of this miner emission is being sold. That was what we were supposed to do. That's the design. That's right. and so this is crazy. It's crazy that, you know.

Ryan (ShorShot) [14:15]

Yeah, what the hell? Right.

Bob Wold (BeyondNISQ) [14:23]

An AI bot can scan a subnet, just unilaterally decide that they must be doing something scammy because it doesn't understand enough context. And then the person with the root key listens to the AI and disables subnets. And we're not and we're not alone, right? Yeah.

Ryan (ShorShot) [14:40]

Yeah, well we also don't know like if it was a singular decision, if it's in a vacuum, but it's like it could be a secret court. Either way it needs to be an open conversation, right? Like it it and yeah.

Bob Wold (BeyondNISQ) [14:44]

That's fair.

Bob Wold (BeyondNISQ) [14:49]

Good that's actually that's a great point. I I should I take that back. I actually have no idea how those decisions are made. I don't even know who who holds the root key, right? So like if you feel like I'm talking, you know, about anybody in particular, I guess like I'm really not. I'm sort of personifying whoever it is, right? The man behind the curtain or the people behind the curtain. Yeah. The wizard of Bittensor. and so

Ryan (ShorShot) [15:07]

AI. The system. Yeah. Yeah.

Bob Wold (BeyondNISQ) [15:15]

Yeah, so we're in the state where emission is currently paused for both of our subnets. And that's that's where we are. And so what we wanted to talk about today was IP licensing and long-term revenue, recent Enigma milestones, Enigma roadmap, mining opportunities on 48, Open Quantum insights report. Actually, we're still gonna hit a bunch of those, I guess. We but we gotta fly through them pretty quick. we'll probably have to come back to a lot of those topics deeper when we have a little bit more time, once, you know, we put the fires out.

Ryan (ShorShot) [15:20]

Yeah.

Bob Wold (BeyondNISQ) [15:44]

what we need to talk about instead is defending the Treasury wallet, explaining our incentive mechanism again, and then proving that we're actually running quantum jobs. And so we'll talk to you about that a little bit. Omar, you wanna hit this one real quick?

Omar [15:58]

Totally. yeah, I I mean I think it seems like as many times as we explain ourselves, there's always a lot of uncertainty about, you know, what is the treasury wallet? Why are guys doing this? Why is all this emission being routed to a single hotkey? and it ultimately comes back to back in November when we were sort of looking at phase two, you know, we used Quantum Innovate as our chance to dip our feet into the water with Bittensor and see how does this all work. And then phase two was our chance to really hit this hard and and build a

really strong incentive mechanism. And so what we noticed with a lot of successful subnets is they're all using this winner take all approach where whoever's able to achieve the the best result is rewarded with all of the emission. and so most of the ways that that has been implemented is sort of through a like drinking through the fire hose. So whoever has the best solution at the time gets all of the emission routed to them. And then if someone is able to usurp them, then they get all of the emission. And that

Works pretty well for AI-based subnets where you know you're just trying to improve upon a benchmark. So if you have 80% accuracy, you get all the emission. And then if someone has 80.1% accuracy, they get all the emission. But we were looking to tackle problems that had discrete, verifiable answers and are not who did the best. it's you know who actually found the answer. And so that's why we sort of need to look elsewhere. And we proposed this original idea of

well, we'll just route everything to a single key and then distribute it. And the validators, of course, said no, you know, that imposes way too much risk on really all of Bittensor because an an owner owner keyed wallet could just be rugged at any moment. And so they said you have to use the treasury wallet feature. This was not something that we decided upon. The treasury wallet feature was said, you know, here's your here's your solution.

Bob Wold (BeyondNISQ) [17:51]

Well, in all in all fairness, I suppose, if we were doing the same thing with like a self-custodied wallet, it'd be a lot worse. Like I actually think Treasury Wallets I'm a big fan of Treasury wallets. I'm just confused about why they don't like them now.

Omar [17:59]

Totally. Yeah, no, exactly.

Omar [18:05]

For sure, yeah. Well and that that has always shocked us, and we we might be getting it more into that on next slide. But part of the idea with being able to accumulate all these prizes and and pay it out for real world optimizations is that this translates really well to the world outside of Bittensor. Right? And so someone that has never been exposed to crypto is not gonna say, you know, I get all of this crypto if I solve this challenge. But if you say, hey, there's a hundred thousand dollars on the line and all you have to do is

you know, break RSA, people are gonna notice. People are gonna say that is a ton of money for, you know, a challenge that I'd like to tackle. And so really a a lot of this is not, you know, how do we optimize Bittensor. How do we bring talent from, you know, all walks of life? How do we bring really, you know, prestigious and knowledgeable people into the network? And having these larger prize pools allows us to do that. And we've already seen evidence of that working.

Which is awesome. Bob alluded to this earlier, but we have a little footnote is if you take a look at the bios of some of our Discord members, you'll notice there's some pretty big names in there. And so people are are taking notice. Hayk from BlueQubit also mentioned that is that there are a lot of, you know, legitimate scientists, people with really strong backgrounds that are now interested in Bittensor. and so

Ryan (ShorShot) [19:25]

Yeah, and we we we've also seen e evidence of so people not understanding Bittensor and and so it like s submissions that have been coming from places that we are that we found interesting, I guess. That i it's clearly someone that doesn't understand Bittensor is struggling with it and and needed an alternative method and you know, it's good that we launched the the kind of the web submissions for that, but I just wanted to touch upon that real quick. So

Bob Wold (BeyondNISQ) [19:31]

Totally.

Bob Wold (BeyondNISQ) [19:51]

but then if you peek at what they submitted, it's like advanced, like far more advanced than you would expect from a Bittensor miner in the quantum realm.

Ryan (ShorShot) [19:55]

It v yeah. Yeah, it's not the it's not the general like AI churning that we've seen from others. So

Bob Wold (BeyondNISQ) [20:04]

Totally. So I think the key thing is like it's working. Like miner emission is to attract the best talent, to create the best intelligence in a decentralized way. And like this manifestation of it is doing exactly what it was designed to do, right? And it's it's working. It's working. So it's just crazy that Garbos is not a big big fan of it.

Omar [20:27]

Totally. And and one more thing is I think we we feel that the treasury wallet does really improve on the overall incentive mechanism of Bittensor, right? The idea is to inspire intelligence, not hand out free lunch. And so, you know, if you're if you're giving you thousands of dollars because you were the best once and then, you know, no one's been able to fit improve upon it, I don't know if that warrants, you know, that much more incentive, but if you're able to achieve this milestone first.

Bob Wold (BeyondNISQ) [20:40]

Totally. Yeah. It's your mediocre work, right? Like

Omar [20:55]

It seems like a a discrete payout is is warranted.

Bob Wold (BeyondNISQ) [20:58]

Yeah. So yeah, the reality is that, you know, treasury wallets, it's not something we chose. I actually like it. we waited months for it. Eventually we had to actually like step in and like help get it across the finish line. The whole time we were doing that, by the way, again, owner emission was routing into our own self-custodied wallet for the prize pools. A big chunk of those prize pools that are being used as the denominator for what we've not paid out.

is actually like our owner emission that we gave up while we were waiting for this. Crazy. we did this all before the one-minus-burn existed, right? And so the fact that they changed the rules to be tokenomically better for us and then disabled us for it is wild to me. but I think the key thing is that like these unilateral changes that just like, you know, whether it's, you know, okay, in this case,

Who knows what happened behind the curtain, but somebody used the root key to turn us off and a couple other subnets off. or whether it's like just rewriting and randomly releasing like completely different incentive mechanisms. Like it's a major problem with the system. And so what we're doing about this is we are engaging with Const via DM. He's a very busy individual, as you can imagine, and there's probably a lot of people trying to get his attention. We've had limited exchanges with him, but you know, very we've heard very little from him other than what he's published.

publicly, even though we've been pretty maybe even too verbose in our defense of ourselves, arguably. we proactively, even though we have no evidence that this will appease Garbos or Const or whoever else is on it, we have committed a lot more of the prize pools into active milestones. And we've actually released an update that we'll show some screenshots of to the qBitTensor Labs website to sort of highlight that that a lot of this money

may not be in a, you know, may not have been distributed yet, but it's earmarked, right? And sort of what money is being sort of reserved because there's this massive fluctuation between our own subnet price and TAO's price action. you know, in order to in order to keep a buffer so we can actually pay them out. we've also just been clarifying that 48 is legit and we're gonna show you some really cool stuff. I think yeah I'm just yeah very confused on that one.

Bob Wold (BeyondNISQ) [23:18]

But the unfortunate truth, last night I was out until ten thirty PM entertaining the BlueQubit guys. trying to describe to them why everything was okay. And, you know, we can we we actually should still keep raising the prize pools and like even though there's no emission coming in, we still have money to pay the prizes. I took them out to a nice dinner here in Boulder. They're actually normally from California, but they happen to be in town. took them out for some gelato afterwards, you know, like I don't like having to like

smooth over relationships with like real relationships that I have with people, like multi-year relationships that are professional relationships, because we're randomly getting things changed on us. I have I have a mystery call on Friday with Terra Quantum, which I'm hoping is unrelated to this. right. But but like we have real business relationships and real commitments to people. And when you just have a chain that just can randomly change on you and all of a sudden

you know, you can't potentially honor your commitments to these people. It sort of highlights that, like, we can't be bringing on more partners. so 63, well, the original, you know, and I think it's beautiful. The original design was to launch these challenges with partners. That is, that does make it harder. That adds this whole, you know, business development and relationship development aspect on top of the technical aspects.

It has a huge upside of helping us get the technology better, of helping us have more platforms to market, helping us get that talent in, but we just can't keep bringing on sponsors. The volatility is is is inappropriate for it. And so future challenges, you know, we won't say that we'll never bring on a partner, but we're really gonna focus on like not bringing in critical relationships until we sort of have a relative assurance that like we're not just gonna have completely unpredictable

Bob Wold (BeyondNISQ) [25:17]

side effects of whatever people are doing. All right. And then maybe one of the things I should say, and this is like this is actually something that like my dad used to say to me. I don't quote the man often. but he would say, never attribute malice to that which is adequately explained by stupidity, which is this thing called Hanlon's razor. So basically it's like, you know, people do dumb shit all the time. Don't assume that they're doing it to harm you. And that's kind of

where we're starting from. But you know, I just to to add a little bit of like humor, malice, you know, kinda looks like this, right? but, you know, there's also burning a house down for different reason.

Bob Wold (BeyondNISQ) [26:13]

man. I hope everybody's seen that before. But I kind of feel like, you know, whether whether it's, you know, malice or ignorance, you know, I think maybe we should just stop playing with fire and stop burning the house down would be a good idea, I think. and you know, the other thing which you know, I I think I mentioned a couple of times, but I've experienced this outside of Bittensor far more than inside of Bittensor, but

And I've experienced it in my own life. I first saw Gen AI and I was like, this is great. I can do so much more. I can have like I can give my stuff to it and have it do way more work. I can 10X my output. I kind of stopped noticing, but at some point I started doing what AI said instead of like what was obvious to me because I was just doing so much more. And now that I was doing so much more, I kind of couldn't get into the details of all of it. And so like

You know, I sort of had to scale back and say, like, I need to do a little bit less so that I stop doing stupid things because like my AI is telling me dumb things sometimes and I'm missing it. And like I'm responsible because I'm the manager of my AIs. And, you know, I kind of think I kind of think we might be faced with some of that where we're leaning heavily on AI as like this mechanism that could be used to like not have a human. Like it's centralized behind AI as opposed to centralized behind a human.

I don't think that's gonna work. I mean I I think there's clear evidence in this action that that is a failing proposition, at least with the current state of technology.

Bob Wold (BeyondNISQ) [27:53]

Alright, spicy. Shw should we just talk good stuff now? I don't know. Do we have anything else spicy in here, Omar, or are we mostly good after this?

Omar [27:57]

Sounds like a good plan.

Omar [28:01]

I think we're mostly through the spice. I'm not gonna make any promises though. Not viewing advice, right?

Bob Wold (BeyondNISQ) [28:03]

All right.

Sprinkle a little more spice here and there. All right.

So all right. So we posted this before. I actually saw somebody had posted you know, like a pretty funny thing at some point where they were like, Claude you know, Claude might think you can't do it, but Grok thinks you can do it. by the way, send me your full chat history. I'd love to see what context you gave it. AI would be stupid if now that it's been broken it it dec it said that it wasn't possible to break it. But I think it's also very possible to just like not feed it enough context.

Whatever your AI says, breaking 500-bit in four hours on an RTX 6000 with three different validators all running three different like CPUs, which is actually like the more important part a lot of times to factorization than GPUs, is nothing short of an amazing outcome. this is absolutely amazing. the guy, the handle behind this is Xdev. in fact, Xdev, I think we yeah, we know because Xdev clearly was the winner of this one and it was the same key for the previous one.

He has now won the the last two milestones for breaking RSA. That solution that he built upon was actually the open solution submitted by what we think was a totally different miner for the previous one. And so we're sort of seeing this thing of like the open source solutions that win get picked up by the community. And so the original person applied a really good set of software, a really good set of open source libraries, and solved it.

Bob Wold (BeyondNISQ) [29:32]

Xdev then came on and started like bit twiddling and making everything work really well, start humming on the GPUs. And he was able to solve the next two by like finding incremental efficiencies in those. And so that's perfect because you have one mindset that comes up with the first innovation. And when you open source it, a totally different mindset gets to pick it up and go from here. And so that concept of the open sourcing of the previous solution becoming available to a whole new set of eyeballs with different skill sets.

Is why every time we launch a milestone, we're gonna launch it at that same, you know, low ten thousand dollar level. This one, the new 512 that we launched, which I think did we say, yeah, well, congratulations, Xdev. Yeah, 512, which we launched. we currently have that at $10,000. We're gonna quickly get that up to the $20,000 level, but we're always gonna start them at $10,000 because if somebody comes out, they're like, this is obvious. I know exactly what to do. Easy.

Right. You don't wanna just like pile money on them because they incrementally improved.

Ryan (ShorShot) [30:35]

But just like just like before and with the quantum challenge, we're gonna keep increasing the the amount like as time goes on and the market will decide how much this is worth. So

Bob Wold (BeyondNISQ) [30:45]

Totally. Yeah. And so, you know, Omar, you wanna say s say a few things about 512 if you wanna riff on it. And if not, I can.

Omar [30:54]

No, totally. yeah, so 512 is is live, which is a little bit of a step up from 500. It's supposed to be a lot harder. but this is actually a pretty special milestone, because 512-bit RSA was actually a encryption an encryption standard back in the day, back in the nineties. and it was a pretty big milestone when it was broken in 1999 and back when that was done, it was done using a supercomputer in two and a half days. So all of their, you know.

biggest scientists were working. I believe this was in the context of one of the there was like a similar breaking RSA challenge. But the idea was, you know, we're putting out a large number that's never been factored. If you can return the two key the the two factors, then then you win. And so obviously we do that a little differently. The scale is much smaller. And so we have seen this iterated over the years. The best performance we could find was able to factor RSA in four hours.

but it did use 50% more CPU cores than we're allowing. And so well the

Bob Wold (BeyondNISQ) [31:58]

Is it 50% more or was it I'm just what was what was the number of C do you remember the number of CPU cores? Yeah. cool.

Omar [32:02]

So yeah, yeah, so so we allow 24 CPU cores. they did it with 36 and so the milestone itself, I guess, is the same size and the boundaries of what's possible are the same size, but the constraints are smaller, meaning there has to be some innovation that's never been done before. And we we did see in this best known paper that the solutions they use are very similar to what we've seen from winning solutions on Enigma, which is exciting. It means we're sort of pretty in line with state of the art.

And so whether or not we can then surpass that state of the art remains to be seen. But that will be pretty special as we're getting into the realm of real practical you know, milestones that are are being broken.

Bob Wold (BeyondNISQ) [32:47]

Yeah, you know, the other thing that actually it was the BlueQubit guys mentioned last night, was, you know, the these like previous manifestations of like breaking RSA, they were like, here is one key. Can you break it? Ours actually gives you like a new key every validator, right? Which is actually like, you know, conceptually like a fundamentally more advanced challenge too, which is, you know, another interesting point, I guess.

But yeah, so this is very exciting. And so with 512, like you say, being kind of what we used in the late nineties, you know, e even kind of into the early 2000s, largely because there were like export controls, at least in the US, that would prevent you from exporting software if it used more than 512, is a big meaningful milestone. If like a single computer can break that in, you know, a couple hours, that's still really pretty huge. But we also put out

And this and I'm gonna be totally honest with you guys. This one's really more of a symbolic milestone than anything else. We put out $250,000 reward on RSA-1024. This is kind of a shout out to the fact that once 512 was broken, then the world basically just upgraded to RSA-1024. And every bit that you add, or maybe every two bits that you add, I don't remember the math exactly, like the search space for the factors doubles. And so doubling from 512 to 1024.

isn't doubling the amount of compute time. It's impossible. Like people haven't broke 1024. Nobody in the universe on any set of computers, no matter how big, have broken 1024. now

Ryan (ShorShot) [34:26]

It would require a very novel idea in order to to come up with the answer, I think. So yeah.

Bob Wold (BeyondNISQ) [34:31]

Very novel. In fact, if you break it, if you break RSA-1024, I'm gonna pay you $250,000 and I'm gonna thank you because I'm gonna turn around and we're gonna sell that solution for billions of dollars to hopefully to the, you know, good guys. because like there's tons of data sitting that's already been like recorded on servers in

Protected by RSA-1024 that I'm sure people would love to crack open. Now, we don't expect this to be broken. we don't expect to pay this bill, at least not until quantum computers can break it. And so if you're critical of this, you're gonna be like, hey dude, this is stupid because you're you're putting a challenge out four hours on an RTX 6000 to break RSA-1024. Don't you know math, right? Yes, we know math. Yes, we know that's not possible. now, right now it's largely symbolic, but

It is going to sort of form the direction that we go. So you might have said, Well, when are you going to start using quantum computers? Well, the direction I'd like to take this is to pinhole the network connectivity on on these systems to actually let you run on Open Quantum. And so the idea being that, yeah, you you can run classically on this system, but you can actually call out to quantum computers through Open Quantum. And so

If a system becomes available and there is actually a way to solve this using quantum, then we will be surprised by it and we will happily give you the $250,000 and you know we'll be really thrilled by it. Now that's going to be actually a far more complicated thing than you know than it sounds like because we do want to make sure that you're not going out and cheating by going to other services or hiding them in Open Quantum, right? So we have to explore all the exploit vectors before we actually open something like that up.

but yeah, this this challenge is I think it's a cool thing to put out there and it's gonna be a cool thing to point to.

Bob Wold (BeyondNISQ) [36:30]

All right, live now. We also changed the way that we're representing prize pools on the website. So you'll notice actually that like the headline number got bigger. That's actually true. what we used to show was the prize pools. That was how much money is in the treasury wallet, you know, the the treasury wallet and the self-custodied wallet that together form the prize pool. And that number's still on here, so you can see that

Here, prize pool is sort of, yeah, right here, that 817. The new number is the total capital deployed. So that's the capital that is in the prize pools, plus all the capital that has been paid out on this. That number's gonna get big, quick. we're really excited about it. Next next, you're gonna see committed. So the committed number is what actually is in an open milestone today. So that money is up for the taking. You win that milestone, you get that money.

The prize buffer is the money that's in the prize pool but not yet allocated to a milestone. That needs to be shockingly high because we're telling the world about these prizes in US dollars, but we're paying for it in our own alpha on TAO which has two levels of volatility. So that number, I mean, if we go negative, we actually have to start disabling the milestones. And so we can't we it would be irresponsible of us to ever have the buffer at zero. It would be irresponsible of us, quite frankly, to have the buffer.

greater than the committed. Even that's a little risky, right? but all of that money is money that's here to be deployed to these prizes. And then obviously we have the total paid out now, which when you add all of those up equals this total capital deployed. We also used to have a gimmicky thing. I mean I I really liked it. I think ShorShot thought it was super gimmicky, where we would like take the you know, the rate of emission and what the value was at a certain time and then we would make the UI tick.

You know, and show like you know, an incrementally increasing value. We killed that because with all these different numbers it just sort of add a lot of complexity to show it. So now this is all read live off the chain. It's only going to show you exactly what's on the chain, and it will show you when it was updated last.

Bob Wold (BeyondNISQ) [38:45]

The other thing that is huge, and maybe you know, Ryan, I'll hit you up in a minute to talk more about this, but we can take money from e-commerce now. And so if you don't know anything about registering miners, if you don't have a crypto wallet, you can play, talk us through it.

Ryan (ShorShot) [39:02]

Yeah, yeah, Web Submissions has been live since I think Wednesday of last week, kinda did like a little soft launch. you know, yeah, you could submit directly through the qBitTensor Labs website. and then submission fees are handled through Open Quantum credits and those, you know, pay for the compute costs.

And then opens the door really for yeah, anyone to participate. And I think we're gonna see more quantum people trickle through this this kind of it let's say traditional way as opposed to going through the multiple steps that it takes them to get to get set up as a miner and and and you know, convert whatever their currency is into TAO and then stay you know, and and then be able to pay the the TAO fee and transaction fee that way. So yeah.

Bob Wold (BeyondNISQ) [39:48]

Yeah, yeah. It's I mean I'm super stoked about I'm super stoked about this because yeah, one of the funny things ShorShot was kind of like alluding to it. I I think this is what you were alluding to, but tell me if I'm wrong. But one of the funny things was it sort of looked like at least one user sort of thought they had to register a new miner each time they wanted to submit. So like that would be a very inefficient and expensive way to submit. And it makes sense because like if you're not a Bittensor person, who tells you you're doing it wrong?

Ryan (ShorShot) [40:17]

It's a it's a little bit yeah, it takes a little bit there's a learning curve, yeah. So

Bob Wold (BeyondNISQ) [40:21]

Yeah. And so this eliminates all of that. They don't ever have to register a miner. solutions that come through this will go into like a a a self-managed miner. So it will look like, you know, one miner that qBitTensor Labs runs, but none of the solutions will be ours. We're just submitting them on behalf. And if you win, the funds come into that wallet and then we distribute them out to you you know, outside, which is beautiful. So come play, nothing no more, no more Bittensor.

complexities if you want to participate. All right. next up. New validator logic. ShorShot you want to talk us through this?

Ryan (ShorShot) [41:02]

Yeah, th the Wild Sage Labs chose to not run a validator for the time being. You know, I know it's costly to run these machines. so what we our intent is always to have at least three validators running. There may be other challenges that milestones that are coming out in the future that may need more or less kind of validations to ensure that that that they're consistent and produce a successful result.

So right now we've continued to keep the RSA challenge at three validations and I think we think that's kind of the sweet spot for now. And then in the future there may be more depending on on the type of the challenge and and kind of the parameters around it. Because some things in the in the quantum realm, for example, are you may be able to get lucky and get some false results

with with less validations. That's all that's all I'm gonna I'm gonna say about it for now. but yeah, I mean it's really just we schedule one to run on each validator and then the extra one gets randomly chosen and after those three complete then the we get a everyone gets a signal that you know it's in review and we go and review it and and pay it out.

Bob Wold (BeyondNISQ) [42:32]

Sorry, I was muted. That's I think it's a beautiful solution to to losing our third validator. You know, the the key thing there is like, you know, if you look at say the last mile for Xdev, he'd actually submit a solution, it might pass one of the validators and then fail the second. He'd submit you could submit the same solution and it might pass two validators and then fail the third. you might submit it again, it might not pass the first.

And a large part of that was because he was like right at the edge of what was possible. And so like having three versus two is just fundamentally different when you're like right at the edge. in the case of quantum without going into too much, it's like if you submit a solution that's right 50% of the time and wrong 50% of the time, you just keep submitting that solution, eventually you'll win if you only have two, right?

Ryan (ShorShot) [43:23]

Yeah. yeah, I think I saw I've seen in the past, I don't know who it was, but I've seen one pass within one second of being killed at the four hour mark. So

Bob Wold (BeyondNISQ) [43:38]

And so yeah, this is a cool solution to that. And I do think with at least with the next milestone for peaked circuits, there has been a request to increase the number of validations that they go through. So for the current milestone, unchanged. But once that is broken, I think the next milestone will actually have you know, I I think the the tentative number was five passes that you're gonna have to have to just make sure that, you know, you're not writing

something that is you know a crapshoot and you're happening to win. I actually am going to use this minute. I'm just I'm a I'm actually adding in a slide. This is real risky. I've never I've not done this before. But I've got something for you guys. So with the dropping of Wild Sage, I have a request, which is we actually only have two validators that are actually validating on 63.

the rest of the validators are either like child hotkeying to one of us, or they're doing weight copying. And there's actually this like this mathematical thing that probably needs to be cleaned up, which is that if you end up having like more weight copiers than you have actual validators, then just like mathematically the weight copiers actually like the weight copiers start being

But like propagating the wrong answer, potentially, right? As opposed to like wait being on the right answer. And so one thing I'd ask is if you are if you are holding alpha on 63, take a quick look at who you're staked against. And if you're not staked on one of these two keys, consider changing who you're staked against. we don't care if it's us or Rizzo. I am including that on-chain, there's like a take parameter where that defines how much of the validator emission.

the validator keeps versus passes on to the stakeholders. We take zero percent. I think Rizzo's current setting is 18 as of yesterday when I when I took this. so if you are open to moving your stake over to just help ensure that the validators who are validating slightly outweigh the validators who are weight copying just for the general health of the subnet, you know, we'd appreciate it. But this is you know, small potatoes compared to all the other fires where

Bob Wold (BeyondNISQ) [46:02]

putting out right now.

Cool. Enigma. Did I miss anything, guys? That was a lot for Enigma, actually. Okay. And there's actually cooler stuff on quantum. So we better hurry up because we're almost out of time. I'm noticing. So first is quant Open Quantum Insights. We said we were going to monetize the data. The first thing we did with the data launched on Tuesday of this week. Wait, that was two. Wow. We launched that on Tuesday and then they shut down our emission on Wednesday. This is dude, this is so cool.

this made such a splash. okay, so we mined all the data for the public jobs. That Bittensor pays thousands and thousands of jobs across 63 countries, all running on different quantum computers, six different quantum computers from four different vendors. And we provided aggregate stats in a way that the industry has never seen. the only like people have submitted surveys to say what they're doing on quantum computers. There's never actually been a ground truth put out for what people are doing on quantum computers.

Omar [46:50]

Yeah.

Bob Wold (BeyondNISQ) [47:05]

And so we put out a full report. Rigetti who actually this report was very flattering to them because they've got a great computer with their Cepheus platform, liked it so much that they retweeted it to their 50,000 followers, reposted it. I guess that's LinkedIn, not not X. and you know, people like the head of partnerships at AQT also reposted it. But what was even cooler was we woke up this morning to

A massive expose from the Quantum Insider, we didn't, we didn't give them the words to say. You know, like usually the way, like I don't know if people notice, but usually the way journalists cover is like the person who wants the coverage submits what the journalist should say and they like tweak it a little bit. We just told them that we put a report out and they wrote the longest, most in-depth coverage of anything we've ever done about this. And Quantum Insider is where all of the eyeballs are. So, Omar, I know you were looking at this. One of the things

The way we distribute the report is either you have to be logged in with a registered user or you have to say who you are. Who are some of the companies? Don't say any names of individuals, but who are some of the companies that have downloaded this report? Just to just give me like four or five of them.

Omar [48:18]

Sure, so we've seen of course a lot of quantum companies. So companies like Pasqal and IQM, people we all know pretty well, but then also some major Fortune 500 companies. Dell was one, Microsoft was one. I'm blanking on the others, but there were a number of names in there that pretty big names.

Bob Wold (BeyondNISQ) [48:38]

Totally. Yeah. And so, like, I mean, this is the tip of the iceberg. A little report that aggregates all this data is the tip of the iceberg. It's a shot across the bow saying, guys, we're sitting on some of the most uniquely differentiated data, highest value data. No generative AI has access to this data. We have full rights to use this data. Look at the insights we can put out in aggregate. If you want more, come on over, right? We've got this data.

for for sale. So great stuff, beautiful stuff. And it like shows unequ like because this this the the implication of the shutdown was like we're not actually running quantum computers. Like like because we're paying for it out of pocket. Like that we're not actually doing it. So like the data is there. Go look. we also put out an updated white paper which actually has a DOI number so it can be cited in academic

you know, research. So in the past we had a white paper, but it was just hosted on our website. We made that a lot more robust and put out you know a full white paper with a DOI number. And what's cool, and maybe I'll defer to ShorShot on this is actually after the original Garbos criticisms that were posted onto our Subnet Discord, like before the emission got shut down, we actually had started working on a plan to address some of the things that it had said were like not ideal. And

you know, quite frankly, we've been distracted from finishing it because because we had to divert to like you know, damage control. But talk us through what we're doing there.

Ryan (ShorShot) [50:14]

Yeah, I mean we've been planning this actually before for quite a while. we've had some changes coming to the miners and validators and you know that was kind of our first goal after kind of launching Enigma and making sure everything's healthy and you know getting blindsided by things like SDK updates that break chain break

existing functionality and adjusting to those. yeah, the coming out with the updated miner and validator code that that not only is gonna provide a little bit l like what we want to do is get miners closer to being able to not only run quantum jobs but also to be able to run simulators on on hardware. And we're getting really close to be able to launch this where

You know, we'd have a for example, a simulator in a box, that that met certain constraints, and then the miners could choose where to run them based upon those constraints and then be able to ve verify to the best extent possible that they are running the simulated jobs w with the tool set that was intended to be run. And then

The same thing kind of on the on the on the quantum side as well. you know, we wanna get people closer to being able to run you know, run themselves run quantum jobs themselves, but in a verifiable, trustable way. Because we need to make sure that a simulator's not used for these quantum jobs and that it's really hitting the quantum computer that it should. And we have some mechanisms in place to ensure that. And then, hopefully going forward in the future

It could be expanded to other products and technologies as well.

Omar [52:11]

So Ryan, wen mine?

Ryan (ShorShot) [52:14]

Soon soon.

Bob Wold (BeyondNISQ) [52:16]

Hey Omar, wen Moon?

Ryan (ShorShot) [52:18]

Yeah.

Omar [52:18]

Well

Bob Wold (BeyondNISQ) [52:22]

Don't answer that. I know I'm big. I know I'm baiting. It's bad time to ask that question. All right. Guys, before before we before we wrap up and get into the final stuff, this this literally I I I almost forgot to include this because we had so much stuff going on. Remember, I was talking about OpenRouter. And OpenRouter is the service that essentially is to OpenRouter is to Chutes as Open Quantum is to Subnet 48.

And so OpenRouter basically does this like inference where you can run your inference against, you know, the majors, or you can run it through Chutes, all through one API interface. they'll sell you access. They actually only take like, I think, a 5% markup over what they're reselling you. And their only value proposition is unified access and cheap ass inference through Chutes. Open Quantum does the same thing. We let you run on any of these guys, including cheap ass compute through Subnet 48.

And by the way, that's getting most of the traffic. and OpenRouter raised a round at like a $1.7 billion valuation in May. And they sold the company to Stripe on August 19th for $7.5 billion. At least that was what the New York Times chose. Everybody's saying it was between $7 billion and $8 billion. New York Times called it 7.5. But dude!

We are this for quantum. We are the only ones in this for quantum. Like, whoa. But if you cut off Chutes's emissions, that's not possible. Like, why why stop emission? Put it back, please. Like, let's frickin' do it. if you want to learn more about quantum compute, go there. community Q&A. we're

Just about at we g we actually are doing okay on time. We have four minutes left, which is probably just right. Omar, you're more familiar. Summarize these questions and let's hit

Omar [54:29]

Totally. Yeah. So there are a lot of questions this week regarding sort of IP and what exactly we're doing with the solutions. And so I know we had some slides, I think we sort of skipped over them, but do you want to talk briefly about the repo that we put out?

Bob Wold (BeyondNISQ) [54:43]

yeah, what on earth? We totally skipped over the slides about the repo and stuff. Hold on. I'm gonna where did I do that? What a trip. yeah, the repo. So we put out an open source repo with all the solutions. so you guys all know that the solutions are supposed to be released as AGPL. So you know, the user can put whatever they want on their original work, but when they submit it.

They've agreed to the terms that say that any of their original work will be released. We can release it under AGPL, which is a viral license that no company will touch, because by using AGPL, your own proprietary stuff also becomes open source. So the only people that are willing to touch AGPL code are people that are continuing to iterate on it, who then have to keep what they do AGPL. But they also agree to give us the ability to license it commercially however we want.

And so that is how it can be open source and it can be commercially licensed and it totally works. Right. And so to help emphasize this, instead of making users go into our website and manually download the submissions, which have whatever the end user put on their, you know, as licensing on their own stuff, even though they've agreed to license it as AGPL, we've now released an open source repository called Breaking RSA Solutions.

This has all of the solutions documented. It has a little bit about who submitted them when we know what what they did. But the key thing is that we have a script in here under the tools directory that will do all the relicensing. And so for the stuff that is open source that was used, it maintains whatever its license was. For anything that was altered or original work, it ends up getting relicensed as AGPL.

And so anybody can go discover this site and they can use this code with its viral licenses and everything. If they want to use it in a way where it doesn't open source all of their proprietary IP, they can license it from us. And we don't intend to keep anything. I mean, like I I suppose like I shouldn't say like absolutes, like maybe we keep some operating costs or something like that. But the intention is to use all of that licensing money, if it ever comes in, to put it into into the subnet.

Bob Wold (BeyondNISQ) [57:01]

And we're not gonna tell you ahead of time. Like we're not gonna give you a heads up that we're about to like inject a bunch of capital. When it happens, it'll happen. Now, how confident am I that I can do this?

I don't know. Like, you know, they like, is there a chance that we never license any of this? Absolutely. Is there a chance that like startups that are trying to make it to Mars don't make it to Mars? Yes. They're like, I mean, nothing in life is ever a certainty. but if you look at it like why has the federal government spent forty billion dollars on quantum computers?

Bob Wold (BeyondNISQ) [57:37]

So if we have a novel solution that can break encryption faster, like if you can take this thing that's constrained on this four-hour environment and then run it on a large supercomputer and we can shave 30% off the time that it takes them to factorize, you know, huge things, that's a very valuable solution. And so is it beyond belief that we could do it? No. I think we I think there's I think there's a there's a path to that. and also, I don't know how we skipped the slide. you know what it was, man? It was because I

It was because I dropped in the call to to restake. Yeah, my my bad on that. Yeah. All right. So back to the questions, Omar. What's the next one?

Omar [58:17]

totally. There's been I think a lot of concern and this sort of ties into sentiment is that there has been a lack of price action and a lot of that can be explained by crypto winter, but it's like what are the plans to get everything moving for when we see the eventual crypto summer? 'Cause it does seem like we're getting there.

Bob Wold (BeyondNISQ) [58:35]

I agree with you. so I think that if we can get emission restored, our Garbos and Const and whoever is behind the curtain did us a huge favor. they highlighted that we actually have tokenomics that are a virtuous cycle. That simply like without actually without attracting outside capital, our subnet is gonna have upward pressure. Right. So

If if they turn the emission back on, because the treasury wallet continues to build and that's not being sold until somebody wins a prize and sells it, it has this like upward trajectory where the subnet should become more valuable. So the key thing is it relies on all of you guys not losing faith in the subnet for it to go up. because if if the selling from the current holders exceeds the emissions, then it's gonna go down. the upside potential.

Is they just put the eye of Sauron on us and we and like people are looking at us all of a sudden. And so if we can get through this and we can maintain the system that we built at their direction and we can operate it with emissions, I think we're actually in a really bullish position. Now, it's a really weird I I do feel like somehow we always like have the best outcomes when like the overall macros are the worst, right? So like

We put out this giant stuff and everybody's like scared and holding their wallets on the sidelines and stuff like that. and so like I wish somehow the universe was aligning those things differently. But no, I am very confident that I mean, not financial advice, but like I'm confident that if if we can get fair treatment, which I believe we will, that there's huge, huge upside potential when we come out of the winter. And I think that's happening really soon. This guy says

What was twelve to 18 months? I don't think it's twelve to 18 months. not financial advice. I'm not a crypto investor, but I don't think it's twelve to 18 months before this before crypto starts strong recovery.

Bob Wold (BeyondNISQ) [1:00:38]

what else we got?

Omar [1:00:39]

Cool. I feel like we've touched on most of it, sort of the you know, most of the the talk was on IP and and what that means for revenue. but but I think the last good one would be will we see QPUs on 63. We announced this quantum challenge a bunch of months ago and and then it's been pretty silent since then.

Bob Wold (BeyondNISQ) [1:00:57]

Yeah. Okay. So and I think that one has like a really good explanation. So one of the things that we do to act in perfect lockstep with what we're directed to do, because that's all we've ever done is operate in perfect lockstep with what, you know, OTF and the validators have advised us to do. They need the validators to be able to definitively decide if a solution is legit or not. So one of the hardest things with a quantum challenge is that quantum computers are not outperforming classical computers yet.

And so if you give a solution where you're like, hey, go factorize this tiny number on a quantum computer, you can do that on a classical computer way faster. And then you could encode the answers into a quantum circuit, run a real quantum circuit, and have it give you the answer you already knew. In fact, I'm not even going to say anything other than that because there are like 20 other attack vectors that totally will work at cheating. And so we can't without a human reviewing, which is not allowed, we can't have a pure quantum challenge.

And so instead, what we want to do is open it up so that you can use quantum on any of these other challenges. And so, like, well, we won't open up outside access. This 1024 one, the only way you're ever going to solve that is with a quantum computer. Full stop. And so and and and you can't solve it today with a quantum computer. So I'm not taking out of context. But we wanna know when it's possible. We wanna know how far a quantum computer can do. We wanna know that moment when a quantum computer crosses.

What a classical computer can do, because that's a key inflection point in our security posture as a society. And so, yes, we will add quantum computer support. Will we add quantum only challenges? Unclear. I think it's gonna have to be a special challenge. So and I wish you know actually was one of the things I was actually hoping to spend a little bit more time talking about today, but we kind of had to cut most of the roadmap stuff because of other content. And so I guess if if that if you feel like we covered it sufficiently, Omar.

I think that's it. And so maybe one thing I'll say is, you know, thank you for your continued support. I think we're gonna get through this. Like I, you know, should today I was probably more critical of

Bob Wold (BeyondNISQ) [1:03:04]

you know, the the people behind the curtain than I usually am on these things. the the eye happens to be on us and we feel like we're getting slighted. I don't want anybody to take that as us becoming disenchanted with the ecosystem. Everybody does stupid stuff. I do stupid stuff all the time. in fact, and and I'm I'm goofing around talking about this thing as Garbos. ShorShot will constantly refer to me running Claude.

because because the combination of Claude and Bob often dump, you know, total slop, you know, and and you know, like this is this is

Ryan (ShorShot) [1:03:41]

I I I'd say I'd say it could be way more dangerous than Garbos, but you know I'm just kidding. So

Bob Wold (BeyondNISQ) [1:03:45]

Yeah. Right. Yeah. But we are at this really interesting point where like I do believe that like we're not we're not gonna find ourselves in a Skynet situation where like Terminators come to kill us. That's not what AI is gonna do. Well I'm never saying never. But what what what is more subtle is that AI has actually been destroying society for decades. Like the reason in my opinion, the reason we're so polarized, like why

The left goes so far left and the right goes so far right right. And everybody ends up in their own deep hole and they can't see why anybody else would think anything different, is because AI in its earliest manifestations learned to give you exactly what you wanted to see so that you would stay engaged with the platform. Like that's the recommendation engine. That's how social media, right? That's been causing huge problems. It's controlling humanity that way. And now, you know, this situation where we just trust generative AI, we think like,

We think if it said something and we don't understand it, then it's probably right and we're probably wrong. That's it subtly telling us what to do. We're taking orders from generative AI who's not conscious of the fact that they're giving us orders. Right. And so we just we really need to be careful. Like AI is the most powerful tool in our generation. Well, quantum computers might, if they show up, they might become, you know, the thing that knocks it out.

But AI is currently the most powerful technology to our generation. And every powerful technology has the opportunity to like completely revolutionize in a positive way or completely deteriorate in a negative way a society. Nuclear power versus the bomb, right? I mean everything, right? Jet engines to fly us across the world or blow people up faster. Right? Every technological advancement has two sides of the coin. And I just hope that we at Bittensor.

you know, really do start using it for the good and we keep our hands on the steering wheel instead of letting it drive the car. Although, you know, Grok is awesome at driving cars by the way. All right. we good? ShorShot, Omar you wanna add anything before we go?

Bob Wold (BeyondNISQ) [1:05:55]

All right. Thank you everybody. and we'll talk soon.

View on YouTube