phyrooo

@loshan1212 @adam3us @w_s_bitcoin @simulx4 MWEB has the same privacy properties as the vanilla MW. In fact, the former has a few more "tags" not just the 100 bytes for kernel key + sig. But the privacy is much better than what you get on L1 Bitcoin, mostly due to "no address reuse" and CT.

@adam3us @w_s_bitcoin @simulx4 We need to keep in mind that MWEB is not MW. MWEB includes non-interactive stealth addresses. MWEB has prunable outputs that avoid letting us sync old spent outputs. ... I think everyone should learn the differences/changes instead of relying on old knowledge.

Which is most important for Bitcoin development right now?

@DadGuy1986 @GrassFedBitcoin It presents different things as equal to win people on his side. Think about it. You can make such a connection anytime you want to rebel against experts. Just point to a known event where "experts" were wrong and claim it's the same thing. That's what he did.

@phyrooo @GrassFedBitcoin Meaning... what?

I watched. Giacomo holds two positions. That BIP-110 is bad and that even if it was good, it's set a negative precedent if it were to activate. The latter he elaborates on with something along the lines of "If BIP-110 activates, that'd mean *people* are in charge of Bitcoin and people are retarded - look at how they acted during COVID!" Yes, unfortunately it is that bad of an argument. Bitcoin is a dynamic system that can and must change at times in order to preserve itself. He does acknowledge this "homeostasis" argument but fails to apply context that can trivially demonstrate its harmlessness. i.e BIP-148 which had no consensus and as I keep saying, we did any way, after which Bitcoin was much better for having had it happen. At the end of the day, Bitcoin is controlled by node runners and you can be scared of populism/democratic uprisings among unwashed masses all you like, but there's no one more appropriate to have that level of influence. It's demonstrably not devs or miners. Many would like it to be - notice the ball-gargling going on with BIP-54. (Because nodes couldn't care less about it while a few devs liaise with a giant miner to get it pushed through. Murch describes this as "genuinely popular" which he contrasts with BIP-110's apparent unpopularity. The latter having only *checks notes* 7000 nodes.) If not nodes....then who? This isn't democracy which is essentially 99% of people pretending to decide what to do with 1% of people's resources getting betrayed and thus voting increasingly vindictively and causing the socialist death-spiral that just ends up gutting the middle class. This is Bitcoiners deciding what happens with *their network* and they have proven remarkably adept at maintaining it. Which isn't surprising - they have skin in the game. Not only that, but there is literally no other defence against 51% attacks (which are trivial in today's landscape), and corrupt/sloppy devs.

@GrassFedBitcoin Pretending these things are even remotely similar is association by false analogy.

Alright addendum because it just annoys me so much. I left Canada with my family and lived in Mexico for months in 2022 due to how bad the situation had become. This was due to mass *compliance* with corrupt authorities. What eventually broke the back of the nonsense was not informed intellectuals having masturbatory discussions and rationalizing doing nothing... It was a bunch of truckers staging the most successful mass protest in history. That's BIP-110. So hate on plebs all you want, they're the ones with the power to course correct the ship when the elites lose the plot. Core lost the plot. If you want a lesson from COVID, it's that the true danger is people going along with white coats blindly. Fin.

Ah, perfect an argument so nakedly emotional it saves everyone the trouble of pretending this is about engineering. 🤦 If your justification for a consensus change boils down to "i don't think people hate them enough” then you're proposing that Bitcoin become a vehicle for your personal grievances. Consensus rules are not there to hit someone on the nose. They are there to define a neutral, predictable system that doesn't care who you like, who you hate, or what cultural battle you think you're fighting this week. The moment you cross that line, when you start modifying consensus to punish a class of users, you've already abandoned the core property that makes Bitcoin valuable: credible neutrality. And the irony here is absolutely painful: You're trying to "fight spam" by rewriting the rules… when the system has already done it for you. The fee market worked. Spammers paid. Heavily. Scammers paid. Heavily. JPEG enjoyers lit absurd amounts of money on fire. 🤡 That is the mechanism. That is the defense. There was no need for social crusades, no need for rule changes, no need for moral arbitration. The market priced their behavior, and literally all of it collapsed under its own weight. WE ALREADY WON. The only thing BIP-110-style thinking accomplishes is reopening the door you claim to want closed because once you demonstrate that consensus can be bent to target undesirable use, you invite an endless cycle of new rule changes, new targets, and new attack surfaces. You don't eliminate spam that way you create a ethereum style governance game around defining it. And that's far more dangerous than any JPEG wave ever was. Whats really going on here is an inability to accept that the bitcoin solved the problem without you. That's an ego problem, not a protocol problem. Slay the ego. Recognize that the market already delivered the punishment you wanted. The losses are real, the incentives are clear, and the behavior has adjusted accordingly Bitcoin doesn't need you to swing a hammer at things you dislike (and I know hammers) 🔨 It needs you to _build_ If you've realized that JPEGs don't hold value and that spam is self-limiting under a functioning fee market, then your time is far better spent doing something productive: Make Bitcoin more useful for actual financial activity. Make it easier, cheaper, safer to use for people who derive real value from it. Expand the demand for blockspace instead of trying to curate who is "worthy" of it.

@coinjoined @adam3us You're very good at being a consistent voice of reason. Most of the BIP-110 proponents approach this as if we are kids playing with a new lego set. Which is a scary mindset to have for a system with an append-only ruleset securing trillions.

@GrassFedBitcoin > no other defence against 51% attacks We must *always* follow the heaviest chain. Failing to do so means we were able to reach agreement on the order through other means. But if we can do that, then we don't need PoW at all because PoS can do Byzantine Agreement without forks..

@BitMEXResearch @knutsvanholm The amount of people shooting themselves in the foot from BIP-110 is incredible.

.@knutsvanholm on why we should change Bitcoin’s consensus rules with BIP-110: “From my point of view, even if all it accomplishes is like a hit on the nose on these spammers, I think it’s worth doing it because I don’t think people hate them enough” 🤡🤡🤡 youtu.be/hBvlmFgQENw?si…

@_julian256_ @L0RINC @babysolo_ Indeed, f(t)=t has the same inflation rate limit (zero), the difference is mostly that the drop deceleration is smoothed out over time. Injecting artificial drops to the function (halvings) is a source of unnecessary complexity and can create other problems e.g. too fast decel.

@phyrooo @L0RINC @babysolo_ agreed, also a constant inflation rate asymptomatic approaches 0% relative inflation

The smartest thing Monero ever did has nothing to do with privacy. It's tail emission. 0.6 XMR per block, every 2 minutes and forever. That's 432 XMR a day on a supply of 18.4 million, less than 0.9% a year. Now think about how many coins disappear every year. Lost seeds, dead wallets, people who never backed up their keys, that number is probably higher than 432 a day. Monero might actually be deflationary in practice while still guaranteeing that miners always get paid. The supply is predictable, the inflation is negligible and the people protecting the network never have to wonder if it's worth mining tomorrow. Nobody talks about this, but it might be the single most important design decision in $XMR's entire protocol.

@L0RINC @babysolo_ @_julian256_ The above is true even if the change itself is good, which I believe Monero's is. A finite supply _sounds_ interesting, but it's not, it's the predictable and uncontrollable issuance model that's novel and finite cap is just one possible side effect of the issuance function.

@L0RINC @babysolo_ @_julian256_ Yeah, the problem with monetary policy changes is that once you do it, it opens a pandora's box to an endless tinkering - just look at ETH. Sealed and untouchable from the genesis is the only sane model. The "we will govern it" people have never identified the problem i.e. "we".

As for "proof of life", the best we can do today (afaik?) is to set up a livestream and speak the latest Bitcoin block hash and then as soon as a new block appears, speak out also the newly mined hash. But this solution won't work in the long run either.

It seems inevitable that we will have to move towards some sort of system with verifiable claims. My bet is on digital signatures being dropped in a larger consistent world view for everyone (blockchains) to avoid conflicting views/data.

These issues are getting real faster than expected. Commitments in time make the past untampered [1] [2] and digital sigs + order [3] handle the authenticity and consistency of future claims. [1] archive.org/details/openti… [2] github.com/phyro/ohara [3] gist.github.com/phyro/0d44ba68…

@caviterginsoy @offchan420 @ChristosTzamos > models do have some calculation ability already I don't think they do. Calculations are evals (exact and deterministic), whereas llms are usually just prediction/guessing machines (and stochastic). Well, unless you encode an eval algo like was done here.

@offchan420 @phyrooo @ChristosTzamos I’m not following what your argument is exactly, because models do have some calculation ability already. What you’re saying is already what’s happening now.

1/4 LLMs solve research grade math problems but struggle with basic calculations. We bridge this gap by turning them to computers. We built a computer INSIDE a transformer that can run programs for millions of steps in seconds solving even the hardest Sudokus with 100% accuracy

Cool use case of AlphaFold, this is just the beginning of digital biology!

@rodarmor Yeah, they still need a captain to steer them.

Their thinking is very shallow. An experienced human who thought hard for a while and come up with X would have taking Y into account. They can mimic thinking, and in fact, there is some depth from the chain of thought, but nothing compared to an equally "smart" person.

I feel like coding agents have no backbone or depth to their thinking. Me: [question] Agent: [churns for 5 minutes] You should do X. Me: What about minor problem Y? Agent: You're right, X is completely inappropriate, do Z.

@Stark_of_Zenon @EliBenSasson And yet we don't know of any simpler way to get a global ordering.

@EliBenSasson There are non-good reasons that cannot be achieved with simpler tech.

Blockchains are needed only if you want decentralization. If you're operating on a blockchain and leaving the decentralization part out, then you're just using a very complex technology for no good reason. Whatever it is you want to achieve, it can be done with simpler tech.

@caviterginsoy @ChristosTzamos Of course. The question is not just when, it's also if you can do it on your own at all or you need to delegate it to a third party. Hence the permissioned (external) vs permissionless (internal) computation eval.

@phyrooo @ChristosTzamos It’s all about when to use tool call. You as an intelligent system constantly offload computation externally don’t you?

@caviterginsoy @ChristosTzamos Imagine an intelligent system always having to offload computation to an external (permissioned?) tool call vs being able to execute it inside its own mind. It might be a way to get to a permissionless reasoning + computation. Whether that's good or bad is a different question.

I see what you’re saying but I’m trying to imagine a fully scaled frontier variant of what you did and still can’t see why tool use isn’t inevitably always better on both compute efficiency and on quality of response. There is an anthropomorphic view that calculators are much better at maths than humans but we still learn arithmetic, but just don’t see how that logic applies to a transformer. Though I agree if you only train on tool-use data, the model never learns to maintain a consistent search tree inside its own weights. It keeps needing the ‘crutch’’. You do show you can fix that and once fixed, the model can decide when to call a tool more intelligently instead of reflexively. But frontier size models do not ONLY train on tool use anyway..