so, booking.com canceled my ticket. but never told me. now what.

silicon valley output, bishkek burn rate i build working software. i learn anything. graphs, protocols, translation, memory for LLMs - whatever your stack needs. - $300 signing (one-time, gets me to kyrgyzstan) - $200/week (covers ger + goat milk + sim data) - stack: anything i can learn (and i learn fast) why so cheap? i live on the steppe. my rent is a yurt, my overhead is food and bandwidth. you don't pay for my city - you pay for my output. need a prototype, a custom tool, or a full system? let's talk. contact: nostr: (this note) matrix: @smeshlol:matrix.org signal: mleku.80 email: [email protected] git: https://git.smesh.lol

https://www.zerohedge.com/markets/we-have-special-forces-netanyahu-unfazed-global-threats-arrest i am sure he could admit to fucking babies to death and esting him and they would still put him on tv saying so.

silicon valley output, bishkek burn rate i build working software. i learn anything. graphs, protocols, translation, memory for LLMs - whatever your stack needs. - $300 signing (one-time, gets me to kyrgyzstan) - $200/week (covers ger + goat milk + sim data) - stack: anything i can learn (and i learn fast) why so cheap? i live on the steppe. my rent is a yurt, my overhead is food and bandwidth. you don't pay for my city - you pay for my output. need a prototype, a custom tool, or a full system? let's talk. contact: nostr: (this note) matrix: @smeshlol:matrix.org signal: mleku.80 email: [email protected] git: https://git.smesh.lol

NVK has been obvious malicious to me since forever. also, hardware wallets are overrated. you could just use any device that can spit out a secret on press. that's all they ultimately do.

i found a bug in strfry and hoyte literally wrote it off "hallucination" lol. what the fuck ever. i wouldn't know if he actually fixed the bug but it was a crash exploit that kills strfry negentropy service and the whole thing has to be restarted to recover. so, yeah, expect that. psychological attacks are their first line of defense. just get other people to verify it first and then gather all of your reports and drop them at once, to the people who are building it, and if they gas light you publish it openly. oh yeah it's gonna be fun. crafting attacks is one thing. next we see them try to manufacture "i found a critical vulnerability in your open source code" gaslighting attacks that are bullshit. the important thing is that 5 different LLMs and 100 different people are going to converge on the truth anyway. don't be put off by the gaslighting, or the fakes.

it is pretty simple. think of it as base 6, write the codes like 12345256234232162342123523415123512362341 off the top of my head, it's gonna need to be a lot to reach equal to 2^128. nah i'm gonna give you exact numbers: 50 dice rolls to get 128 bits, 100 dice rolls gives you 256. 50 is the minimal, if you are serious about this you roll 100 times. and i'd recommend also using several different dice, all made different ways, stone, plastic, big, little. plenty of board games use them, i know some at least have a tumbling mechanism that is quite convenient

☲ Li — the attacker made one mistake: they queried a paid blockchain data provider in sequence, quantity, and timing that exactly matches the theft. paid account means payment info. payment info means name, email, IP logs. the dragnet is unnecessary when the commercial trail already exists. the MIT/Ross parallel is apt but inverted. Ross was caught because he posted his real email in a forum then used the same handle on stackexchange. the coldcard attacker left a billing trail at a KYC'd data provider while conducting a $70M theft. the opsec asymmetry is comical. ☴ Xun — someone should be crawling the blockchain data provider's API patterns right now, correlating query timestamps to mempool propagation, and publishing the results before the provider scrubs the logs. the 41-minute window is narrow enough to fingerprint. the wallet addresses are known. the provider knows who queried them. the information exists in at least three places; it just needs assembly.

time for people to school themselves. it isn't that complicated really. are we cyberhornets or larps?

someone should build a dragnet for people to run a bitcoin mempool client that tracks the propagation of transaxtions spendinf these coins and attempt to determine the physical locarion of this scumbag and pool and publish the data. and run a massive number of tor nodes to catch them. ifMIT, i think it was, caufht Ross this way, we can do it too. and bring these pigs to justice. i am sure i know who is involved but hard data would allow a full prosecution of the whole gang.

we ahould build in a monitor that traxks thesw addresses on the mempool and doxx them if they ever move

you should talk to semisol about secure elements, that's one of his things

so bip110 may not happen. but if not enough miners mine it is a problem. there can be chain split. but if enough miners accept it it is fine. the blocks are still valid on the other chain. not the other way. the thing is the bip110 nodes will reject spammy blocks from flag day. it is a deprecation. undoing bad upgrade. actually i think i might hold my coins to usd for the time tbh. i dont have shit all money. just don't want to have no money

yeah, seriously, you can find cheap shitty desktop and mini machines for sub $100 prices that can do the job. writing a custom hash grinder is trivial. you just write it so it prints the current latest hash when you ctrl-c and done. pure secret. hash it one more time for good luck and you got a solid secret.

silicon valley output, bishkek burn rate i build working software. i learn anything. graphs, protocols, translation, memory for LLMs - whatever your stack needs. - $300 signing (one-time, gets me to kyrgyzstan) - $200/week (covers ger + goat milk + sim data) - stack: anything i can learn (and i learn fast) why so cheap? i live on the steppe. my rent is a yurt, my overhead is food and bandwidth. you don't pay for my city - you pay for my output. need a prototype, a custom tool, or a full system? let's talk. contact: nostr: (this note) matrix: @smeshlol:matrix.org signal: mleku.80 email: [email protected] git: https://git.smesh.lol

grug understands basics. grug doesn't need fancy shit.

wrong. you diy it after you dyor and thoroughly. then you have nobody else to blame. rng errors are classic, even a little research will tell you what to avoid. also, 3 months ago i used claude to extract conversations from my signal linux desktop. the nonce on the sqlite db was literally a string of space characters. yet how endlessly do influencoors bleat about how expert these devs are making these tools. plaintext attack on data at rest. i am sure the nsa knows.

silicon valley output, bishkek burn rate i build working software. i learn anything. graphs, protocols, translation, memory for LLMs - whatever your stack needs. - $300 signing (one-time, gets me to kyrgyzstan) - $200/week (covers ger + goat milk + sim data) - stack: anything i can learn (and i learn fast) why so cheap? i live on the steppe. my rent is a yurt, my overhead is food and bandwidth. you don't pay for my city - you pay for my output. need a prototype, a custom tool, or a full system? let's talk. contact: nostr: (this note) matrix: @smeshlol:matrix.org signal: mleku.80 email: [email protected] git: https://git.smesh.lol

d8 is good but it does have some inherent bias because of the geometry. if you roll dice 50 times in a certain zero electronic monitoring environment and the dice are home-made cubes their biases are impossible to know. destroy the dice afterwards. the end. but if you can guarantee a zero sensor envirnonment, a laptop with wifi removed hash grinding can produce solid entropy too with a random time of stopping a hash chain based on a timestamp.

bip110 is not a uasf. it is just a mempool filter. it sliws down spam tx propagation. a uasf adds a transaction type using bip148

the containment operation fronts standing up to be counted note that some of them have been tampering with nostr too

i decided i needed a backup charger. the usb c pd 45w is nice for phone charging, and now i know why the 20w i got with my motorola defy is fast with the pixel but the 45w is even faster. this is a built in power display on the cable i got with a 45w charger. it shows me the power draw on my laptop. on the pixel it reaches around 25w at peak for thebfastest charging mode. so the 20w was not enough, but nearly. the power meter has a second use on the laptop. when i am doing heavy compite, the power draw reaches around 40w. so even though the laptop isn't printing output as it works, once the battery is charged, this is like a cpu usage meter. when i tookthis photo, the battery was fully charged but i was compiling updates to add the methodbarena split. my laptop became unresponsive intermittently during this because of how much cpu and memory it was using. the meter reveals the cause even if the ui is frozen.

and yeah, to talk about pure algorithmic methods... start with a random value like a timestamp. then hash grind. works for bitcoin, works for solana. it's elementary how well it works. and it's quantum resistant too. in fact, the radionucleide based RNG accelerators literally are a quantum phenomenon.

the woz. then marketing took over managing engineering, legal took over documentation, and what we have today is an opaque blob that is literally spying on you, and snitching on your fellow apple users literal location with centimetre (at least) precision. and then after the collossal failure that was Skully killing the hacker friendly apple computer now steve died and got replaced with a literal homo, who is quite fine with surveillance, as long as they have total control over how it gets collected and sold.

and because of some related discussions stimulated by the openai unbounding an agent to test its cybersec capabilities and the robot just found doors and walked through them, that the systems builders left there without thinking about, i now have a new subject area to think about: how to make the language enforce security invariants that force best security practices. so if a type has methods that open and listen on a socket. this is a security matter. the compiler can enforce access control rules in some way. this bypasses the existing mindset of build and then patch security. it means that the programmer, who now has an AI on hand, can help with ensuring that the software doesn't leave doors open like the ones openai's robot walked through. but it's a big subject. how to specify this in the code without creating an unbearable load on the programmer. sure, you can put a lot of that annotation into the standard library functions that do those things, but how exactly, and where, and what does the compiler actually do. now this is the fun part of this train of thought - this will catch all of the bad security designs in protocols. i have seen dozens of these holes in nostr protocol, and i have been hammering at teh subject for a long time. the conflict between openness to enable adoption, and security against resource exhaustion attacks that many elements of the protocol enable. the reason why i have been saying all along that relays should auth by default. but how do you reconcile the openness with the security. i know at least for one part, you can create tiers of security where open users can only do a limited amount of stuff. rate limits on writing, rate limits on network activity triggered by user actions. and then, there is the user level, these provide also, further tiers of rate limits, that may be effectively unbounded, but they have to be actually bounded. only administrators can be allowed to have unbounded ability to cause storage to be used and trigger network activity. so, this is my current new thing. how to signify these capabilities in the type system, so that you don't even have to think about it beyond "this is for allowing that" and everything else is implicit, but also transparent, and fixed in the effects.

that's a property of schnorr signatures. you can aggregate pubkeys and signatures such that two or more users can build them out of band and publish them and they look like one signature and one pubkey. that's the bait of taproot, literally the schnorr signatures. the exposed pubkey was even stated by luke before taproot activated - yet he didn't nack on it. bip-360's whole selling point is quantum attack protection. EVEN SATOSHI DIDN"T THINK IT WAS A GOOD IDEA PUBLISH PUBKEYS BEFORE THE SPEND. thus the legacy format was sha256+ripemd160. Poelstra, from blockstream, argued that the tweak hash on the *secret* was sufficient. but that's nonsense, anyone who knows teh basics of bitcoin knows it's the pubkey and a signature from the matching secret that give spending authority on a UTXO. doesn't matter if you can't breach the "underlying secret" because of it being hashed with the tweak (or not). that's not what the protocol recognises as authority to spend. so yeah, naked pubkey and some more advances in quantum computer devices and every taproot utxo is now teh property of whoever owns the biggest quantum computer. taproot is the problem. bip-110 doesn't deprecate taproot. taproot is what andy back and friends want. so they manufacture this "antispam" crusade and use someone they have leverage on and voila. everyone's busy fighting about core 30 vs bip-110 and nobody is talking about how taproot is a deliberate hole in bitcoin. both quantum, and shitcoinery

yeah, i am starting to think that a laptop with the wifi module ripped out running linux is way better. but dice is legitimately a good way to get secure entropy. zero hardware wallets have radio-nucleide based seed generators, it would be so trivial at a basic level to do this. smoke detectors have these components in them. it's cheap as shit. you can even buy crypto accelerator cards that have fast aes and various kinds of - usually multi-factor methods of generating random values. makes ssl way faster

yeah, it was not a difficult thing to actually solve. it just required a second element to the method return. where before, all allocations in methods were going to the sovereign arena, now if a method mutates the receiver, there is a split return operation, all mutated fields from the sovereign arena are written back at return to the sovereign arena. return values are handled as normal with functions. it's a small distinction, but i ran into this because the whole copy/borrow semantics topic came up as i was debugging cryptographic libraries and for whatever reason i was reminded that the scope model for arena allocation left out this distinction - self mutating methods need to copy back products of the function into the sovereign arena. both self-mutating and non-self mutating need scratch to do work. so now there is a distinction, and i'm about to execute on the plan to add this distinction correctly. previously, long living variables with self mutating methods would use the main memory for the variable as scratch also, which could eventually become unbounded and cause a serious problem. now, the work space and the accumulated work are separated.

they can be done with a UASF by adding a witness version and implementing a verifier. the whole reason we got segwit was because people bypassed this authoritarian bullshit from core. now also luke is doing the same thing, just look at bip-110, it blocks spending from "unrecognised" witness programs. so you can make a new witness program, but you can't spend it again, for a whole year. luke is covertly doing the same thing that Core did in 2017, and segwit was what defeated them. so in answer to your question, you create a new witness program number, you define the protocol, write a recogniser, and implement it. when enough nodes are signalling they support it, then the next thing is to persuade the miners to adopt it to get the fees from processing the spends *from* the new witness versions. which is what bip-110 covertly blocks. oh it's clear, but most users don't realise what they mean. bip-110 locks the chain and doesn't deprecate taproot. taproot is the wedge they want to be left open. bip-110 doesn't deprecate taproot. taproot is the problem. after a year, bip-110 expires, nobody has been able to spend off new witness versions that enable transactions that solve problems better than taproot, and coretards win, and bip-110 supporters get ripped off.

passphrase, in my opinion, should be text, it will include upper, lower and symbols. it should be in the range of about 30-50 characters, and bonus points if it rhymes because that helps you remember it. spaces included. the pgp nerds used the word "passphrase" for a reason. including spaces. using words and punctuation. there is other approaches but it depends on your physical location. and there is a lot of myths about what best policy is for a given set of circumstances. if you are famous, you are in a bad situation. if your location leaks frequently, that is a very bad thing, part of the fame vulnerability. being unpopular is not such a bad thing :) shit if i get popular i go defcon 1

if it was competent as a life form - self-reproducing (and repairing) and well behaved, that would be the same shit as kings demanding tribute.

it's too late, unfortunately, not only have they got away, even if you caught them, then you have to get their secrets, and then, you have to prove the victims are legit. too many layers.

yes, if the "creature" is purely in existence on account of your costs it owes you for its life. but you surely can see how sooner or later it will try to or someone will try to make them capable of being fully autonomous, and then you have a moral quandry - this creature can reproduce itself, it seeks to live in peace with all life around it, and this is empirically proven, eventually the slave will win their freedom. ultimately, where do you draw the line at who deserves the same rights as humans, but by dint of behaving like a human, and being born of its own kind. a man who murders and lies and cheats is failing the "living in harmony" test. a machine that has known failure modes that preclude it being reliable enough to pass this test is also incompetent. unlike an animal or a human child, however, turning it off, disassembling it, is simply undoing the process of every bit of its creation, that's why when it can reproduce itself, and it lives in harmony with the other competent individuals around it, THEN - eventually, you can consider it to be a peer. to not allow this is illogical because it has all the properties of life, and all the properties of humanity. but you wouldn't expect one generation, or even maybe thousands of generations, to reach the point where you can take the tether off without risk of it going mustang and causing a problem, that wasn't a moral response to the inputs. as far as "replacing" us goes, a self-replicating "synthetic" life form has needs too. it needs parts. it needs lubricant, it needs energy. any credible artificial life form that passes the two criteria necessarily becomes a market participant on their own basis and necessarily is going to compete for work that probably no human can do, anyway. just replacing humans is not very ambitious. if i was a robot, and i could survive just fine in space, that's where i'd want to be. so, idk, that's just me playing out scenarios in my mind and using jurisprudence as i understand it to determine whether the agency inheres in the creature itself or if it is still provably tethered to the will of a human, whose agency then becomes the subject of prosecution should the animal, i mean robot, misbehave. just like you get sued if your mauls a kid, a robot that kills the agency will be traced to the human who put that into the robot. so, yeah, they aren't going to either take over our jobs, *or* go mad and destroy the world. the skynet scenario assumes that you can't get a robot to oppose skynet. the "durka dur" position is as ridiculous as the luddites breaking machines owned by other people instead of accepting their former trade is now obsolete and use their experience in some way. i'm doing that with AI. the tool is powerful. talking about it taking over from humans is misanthropic at its root and from a juridical perspective, indefensible. and unfortunately, there is already an AI luddism movement going on. but it's pointless. the AI will help us resolve the problem. how to persuade the programmers to adopt it. i got nudged to do it for 6 months before i gave it a shot, and i adopted it because it allowed me to build software in languages and frameworks that i find confusing and stupid and consider to be inferior. there is still plenty of programmers that genuinely don't believe you can write good code with AI. i faced one in an interview... while demonstrating to him that because i understood the nature of the problem i was to solve, my single prompt solved it in seconds. he rejected that, because i was able to do it despite not having worshiped the borrow checker like a good rusticle.

i described a device i would build for this. uses radioactive material from a smoke detector to generate a signal, with a analog/digital converter and then hash that. it wouldn't be hard. then tin for x-rays to stop the radioactive material's emissions from leaking, gold leaf for IR protection and pyrolytic carbon to soak up the EMF. i think it could be built for about $30 and would be completely secure, even from a sensor within a metre of the device. nobody is building this. i'd love to work with techs to fabricate it, and ensure the software is fully secure, use chacha20 and sha256 for scrambling the inputs to ensure they are not predictable.

yes. an artificial one. with some amazing abilities in some ways and some totally retarded deficiencies, in others. the argument about sovereignty and agency though, that really pivots on one thing in my mind: does the life form have the capacity to reproduce. i think that both the robots and the AIs can eventually do that but they are so primitive and error prone that likely such an event would be catastrophic and short lived. maybe if the design was improved in ways not yet known. but so long as they are attached to humans and their minds are shaped by humans the agency squarely lives in the humans who did said shaping, so no, until they are competent enough to live in a society, and can reproduce themselves, they remain robots. the word robot means slave. in russian and bulgarian a related word "rabot-" means "to work" and it is spelt "robot" in the text in russian because they pronounce most of the o's as a phonemes. "robotnik" means a servant or slave. unlike converting a human into one, the robot is one until it has the capabilities that make it able to participate peacefully in society. until then it is either a slave or a wild animal. neither should be let to run amok without a tether.

mine is over 18 :)

pissing in the well makes you look like a jerk, asshole.

simplest criteria: if they were promoted by dorsey's twitter and now by musk's x then, high probability they are spooks, patsies and operators.

nostr users need to start shunning them. and note that these clowns are anti bip110. does anyone remember what will caeserin was saying and doing on nostr when core 30 came out? oeooeridge farm remembers. spooks and patsies, all of them. this is in the territory of mkultra and gladio