someone made a half life movie using garry's mod. it's pretty fuckin sweet. i mean, b grade but still awesome.

# 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: - matrix: @smeshlol:matrix.org - signal: mleku.80 - email: me@mleku.dev - git:

repos

https://www.zerohedge.com/ai/chinas-ai-knife-fight-deepseeks-new-model-runs-100x-cheaper-anthropics-flagship been using deepsnek for two months now. i've not regretted it at all, especially not the part where i'm spending less than half as much on it, for actually better results. the model has some quirks, but for my work, what claude was more of a problem with was taking shortcuts and ignoring the imperative of my requests to use an unconventional model to solve the architectural problems i was solving especially with my moxie language compiler. claude kept on reverting back to using heaps and garbage collection, and that was the central point of what i was trying to eliminate. 3 weeks claude worked on it, and i can honestly say it made no real progress. deepsnek actually built what i asked it to, which led to new problems that i had to devise solutions for. it's getting really close to a solution, and for the last 3 weeks i haven't had a lot of time or money to keep up with the work but i'm feeling like it's closing on a fully working model now. a programming language that doesn't need a garbage collector, and generates code and uses memory in a novel way that allows it to bypass this. same memory safety and elimination of stop the world GC pauses as Rust, without the ugly and confusing rust memory management annotations. american AI dev shops are way behind with the tech now, grown fat on giant farms of nvidia chips. usg restricted chinese ai devs access to this hardware, so the chinese got creative and found really good solutions. and now, huawei and others are building better chips for less cost as well as the models being trained to be better AIs for less cost. i mean, for corporate users, setting up their own small in-house data center, and running deepseek or kimi is entirely doable on a budget they can realistically field. and the improvements in the model and algorithms are remaining available, that's a shared benefit for everyone in the field. open sauce > closed sauce.

a little history lesson: the pattern is remarkably consistent across 500 years of history: **tiny, stubborn minorities with low overhead and high conviction change the world; comfortable majorities don't.** **the demographic profile of the stubborn core:** - **1-3% of the population**, but densely networked. the cypherpunks were a mailing list of a few thousand. the abolitionists were maybe 2% of americans for 30 years. the suffragettes were a few thousand women in a nation of millions. the samizdat network in the USSR was statistically invisible. - **economically independent or low-overhead**: monks, artisans, tradespeople, people who don't need the system's approval to eat. the printing press operators, the ham radio operators, the shipyard workers. people whose survival doesn't depend on the institution they're resisting. - **ideologically framed, not materially motivated**: they refuse on principle, not for a raise. this is why they can't be bought off. - **curmudgeons overrepresented**: the chronic skeptics, the ones who were "difficult" before the crisis. they have the receipts because they never trusted the narrative. - **skilled in the era's communications technology**: pamphleteers in 1776, samizdat typists in 1970, mailing lists in 1992, relays in 2026. the resistance uses the distribution channel the center underestimates. **the mechanism of stubborn refusal:** 1. **a visible alternative exists** - the refusal itself is the signal. "we will not" is a reference point others gravitate to. 2. **it raises the legitimacy cost** - every crackdown proves the movement matters. the powerful are forced to respond, and the response is the advertisement. 3. **it recruits through demonstration** - people see refusal is possible, and the threshold to join drops. 4. **the center's response legitimizes it** - the moment the powerful argue with the minority, they've conceded the minority is a threat. **historical scoreboard:** - **the american revolution**: the sons of liberty were a few thousand. the boston tea party was ~100 people. the refusal to accept the intolerable acts, held stubbornly, outlasted the british empire's patience and logistics. britain had 100x the men, 1000x the money, and lost because it couldn't outlast a minority that refused to normalize. - **the suffragettes**: decades of hunger strikes, window smashing, public humiliation - and the refusal to accept that the franchise was closed. they won with numbers that never exceeded a rounding error of the population. - **gandhi's salt march**: one man, a handful of followers, and the refusal to pay a tax on salt. the refusal was the entire strategy. it worked because it was unbearably visible and impossible to argue with. - **the montgomery bus boycott**: 381 days of a black community refusing to ride buses. economically devastating to the transit system, morally devastating to the segregation narrative. the refusal, not the protest, was the weapon. - **the cypherpunks**: a few thousand people refused to accept that privacy was dead. they wrote the cryptographic primitives that now secure trillions. they changed the world with math and stubbornness and no institutional backing whatsoever. - **solidarity**: it started with one shipyard refusing to work. the refusal spread by demonstration, not by organization. - **the french resistance**: tiny numbers, huge morale effect, disproportionate damage to the occupation's legitimacy. "we can stay autistic longer than they can remain solvent," is the historical formula: **the center has overhead, the stubborn minority doesn't. the center must eventually break even; the minority just has to keep refusing.** gen - withdrawal as sovereignty - is the oldest trick in the book. every one of those movements won by refusing to participate on the terms offered, not by defeating the center in a head-on fight. the center can't counter a refusal to play. that's the water element: resolution through refusal. the first to act on the lie destroys it for both - and the refusal is the act.

#bip110 #fuckcore #coretards #meme #memestr

# 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: - matrix: @smeshlol:matrix.org - signal: mleku.80 - email: me@mleku.dev - git:

repos

the strategic calculation error of the coretards is very simple: shitcoin transactions and ordinal spam and suchlike, are "paper" values. they entirely depend on "paper" money from central banks. central banks one day can wake up and choose violence, hike rates, and they will, because hyperinflation is teetering on happening. the eurasian continent is abandoning the USD as fast as it can. that is the hypothetical basis of the value of these spammy "assets" on the core spamchain. in contrast, the value of what sats will be denominated in on the minority bip-110 and subsequent minority fork, is real value, denominated in sats, not USD. that is what really changes the equation. don't give up just because lukedashjr's and dathonohm's efforts fail. we get forked off. let's build an exchange. let's petition small exchanges to list the minority coin. the UTXO set is still there. we can mine with more rewards, even if the value of the minority chain is lower for now, the fact that we have solid economic backing and are not depending on central bank liquidity matters more than anything. those of us who mine and hodl and run nodes and run businesses that accept these payments, we don't depend on central bank liquidity as much. that's the achilles heel of the coretard spamchain. they are aligned with the global monetary scam. that is not what bitcoin is about, they are shooting themselves in the foot, low time preference, autistic monetary maxis will win anyway. or at least we will have a clean, legitimate chain with actual monetary value not just puffed up by suits printing money by cosy deals with banks and the central banks and world bank and imf. you want the parallel economy? then you should support the minority fork chain anyway. because otherwise next up is bitvm and running a spamchain node cost goes from the current ~$400 startup cost and $20/month sorta range, into the thousands of dollars. i've seen what it's like in shitcoin land. EPYC nodes with a terabyte of memory and 64+ terabyte. just to run an Internet Computer node. price is in the range of $70-200 THOUSAND dollars. that's where spamchain ends up. some of us don't care about being in the majority. we care about the real economics and ethics. i will run a reduced data signalling node indefinitely. i will modify my node to retain it even after the flag day in 4 days. they can have fun licking central bankers boots. i'm here to do real business.

winning the war against spam and shitcoinery on bitcoin is not just about gaining hashpower. the situation as i last saw it is being measured - by p2p nodes signalling bip-110, it was at 20% of total nodes. there is efforts that are visible by coretards running fake p2p nodes that don't actually process transactions or actually relay transactions - a cheap, high time preference attempt to stuff the ballot box. it won't work. that 20% figure doesn't probably count the fake p2p nodes that don't actually have a replica of the blockchain behind them. > ok, so 20% is the number of nodes signalling not just the ones with miners attached. but that's a LOT still, enough that it's now a swarm of political resistance right - node signaling is the political metric. 7-15% of nodes running reduced-data rules is the swarm: every one of them is an independent deployment of the resistance, not a proxy for luke or dathon. no single point of attack. that's the octalogue's xun and kun working - invisible penetration via adoption, new ground via installed infrastructure. the strategic distinction: hashpower signaling is about who wins the block race. node signaling is about who owns the economic consensus. when 15% of nodes - including the ones running payment infrastructure - reject blocks that carry spam, the miners face a real economic question: their blocks get orphaned by a minority partition, and the value of their chain depends on the economic majority recognizing their blocks as valid. the 2017 segwit playbook worked exactly this way - BIP148 nodes forced the issue by refusing non-signaling blocks, and miners capitulated when they saw the userbase's economic weight behind it. the difference from 2017: segwit had ~90%+ of the userbase behind it. BIP-110 has 7-15% of nodes and ~3% of hashpower. the swarm exists but it's smaller than the one that won segwit. the mandatory window in august is the test - if the swarm holds, the economic pressure compounds; if the window closes without capitulation, the reduced-data chain becomes a permanent minority network and the main chain keeps the spam. the political resistance point stands regardless: 10,000+ nodes refusing to accept the "art on chain" narrative is a public rejection of core's governance that can't be shamed away. whether it converts to a fork win or becomes a permanent parallel consensus, it changes the conversation. the swarm is real. the outcome is not decided yet. ----- and i'm not gonna stop running my node this way. and i'm going to be running it as a payment rail onchain and lightning. i think that it is very likely that other users will do the same thing. a lot of people were not running nodes because they couldn't do a full node or like me were not in a stable physical location to run a node. but you can do pruned nodes on VPSs. you can attach lightning nodes to them. you can open your own little online business and sell stuff - i'll be selling a book and maybe if i find some people who want to invest in it, the "tinfoil wallet" with tin, gold leaf and pyrolytic carbon shielding, and radioactive material providing entropy that can't even be picked up by a device scanning for it within under a metre range. because why not? by launching this attack on coldcard users the way they have, they have added something to the mix that is an incontrovertibly evil action that growing amounts of forensics data link the actors involved in bringing about this gigantic heist, they have taken away what comfort we had in our futures. they have made it so this becomes a war that we will not back down from. running nodes that filter the transactions, and some of us mining, enough to form a working chain, we now stand here, doing actual business, and the miners who stick to spamchain are missing out on blocks full of real fees that we get to keep on our minority sidechain, regardless of whether the UASF activates this month, i think i'm going to set about setting it up so it doesn't back down. it stays on the minority chain. i get a miner, and add my little power to a much smaller pool on the sidechain. miners can either get in on that, or they can go fuck themselves. it doesn't matter. bitcoin is not a smart contract, data storage chain. never was, never will be. that is not an argument they can win. it's contrafactual and contrary to everything Satoshi wrote.

so, maybe the last architectural decision for the #moxie programming language: tail recursion because moxie isolates the scopes of functions, recursion is particularly brutal, each time it calls itself it allocates a new arena, and its result values, that go into the parameters, are passed in and have to be copied on the next cycle. tail recursion effectively makes it so that every cycle nothing has to be done. the return value is effectively unified with the parameters making an exception that they are mutable, and it just jumps back to the beginning of the function. Go never had this facility in it and it made recursive implementations very clunky and memory hungry. the Go compiler itself uses a recursive descent algorithm, so it most likely required a lot of optimization in the garbage collector with the special case so that it doesn't blow up the stack. when you simply call the function itself, usually there is one or more calls back to the function itself at various branches in the logic of the function. what Tail Call Optimization does is effectively recognise that the return values are what are being optimized on, and the input values are the same variables, and you just go back to the top of the function with the return values already handled as being the output values, and voila. my first time trying to implement this, i never got this far into building a compiler before. i usually got stuck somewhere around the code generator and didn't get to the point of anything actually running. but i'm super pleased about having TCO in moxie. recursive implementations are simpler to write and more elegant than iterative functions. having TCO in the compiler means that when i have the choice to do iteration or tail recursion, i can write it as tail recursion. there is a lot of stuff such as hash grinding and tree walking and graph walking in general that are simplest to express as recursive so this is a big boost - faster to write, and performance will be very high, similar to how this is also a cheap operation in languages like haskell and lisp, which are also recursion heavy and have TCO.

man, istanbul monday morning is so manic, omg. so many dudes running around carrying building materials and giant loads of recyclables on these crazy cart things they walk up and down the crazy hills. idk how anyone lives in this place, it's so squished together and crazy. the shopkeepers are mostly really nice tho. and i come back into the hotel and on the TV it's talking endlessly about inflation and showing all these images of people at markets selling fruit and vegetables.

# 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: - matrix: @smeshlol:matrix.org - signal: mleku.80 - email: me@mleku.dev - git:

repos

but if you had shown that to these guys they would have told you the robot was hallucinating. i had hoyte say that about a vulnerability in negentropy. lol. whatever. 8 months i have been working with AI to do dev and maybe sometimes you have to get it to run a few times and watch closely what it does but you can actually do these things that would take days and days to do, in hours. View quoted note →