llms / harness engineering / investoor / music prod / optimist
Joined November 2018
- Tweets1.2K
- Following456
- Followers2.3K
- Likes171K
So sorry to hear this, my condolences to you and your family.
I lost my Dad to an aggressive form of cancer earlier this year. He was diagnosed and then he passed two weeks later. He was 71. Still hard for me to believe it’s real sometimes. Sending prayers to you and your family
Cursor just released their own source/version control a couple months ago, Origin. i don’t think too many companies will use these tho, at least not for a while (most cos are currently locked into multi-yr contracts w github, gitlab, atlassian etc.)
ya just having fun. ggs.
i do think the "crypto is ai loser" and "ai x quantum" fud is largely overblown tho. full thoughts here if ur interested xitter.cf/climbcode28/status/210…
Cryptography is _the_ most sophisticated field in computer science. Bitcoin's SHA-256 hashing has a brute-force time complexity of ~O(2^256) - roughly 10^77 attempts (ask AI how astronomically large that number is). and that's just hashing. Bitcoin also uses elliptic-curve cryptography (ECDSA / Schnorr).
hashing / merkle trees / SHA-256 / pub-key cryptography / digital signatures - these are the foundations of modern digital security. banking / 2FA / passkeys / healthcare / govt systems / legal records, etc. all depend on cryptographic infrastructure.
if AI x quantum discovers practical shortcuts against widely used cryptography, we'll need fixes across _all_ portions of digital infrastructure, not just "crypto" or "banks". i.e., protecting our military's nuke codes, medical/legal records, bioweapons access, etc.
ofc AI will accelerate cryptographic research. my guess is - frontier US-based labs (who lead AI, don't see this changing) accelerate breakthroughs => cryptographers / standards bodies / industry leaders develop solutions in collaboration w/ frontier labs behind closed doors => institutions/protocols migrate to updated standards. this is already happening btw. NIST finalized its first post-quantum standards in 2024.
agree that Bitcoin has greater coordination challenges, but there's economic/political incentive for btc-core team and the broader network to adopt necessary upgrades.
am not dismissing the risk. but there's a massive distinction between theoretical vulnerabilities / mathematical breakthroughs, and practical exploits. AI advancing rapidly doesn't mean Bitcoin's cryptography is anywhere close to being broken (ask AI).
appreciate u sharing your thoughts, but with all due respect, would implore you to learn more about the field of cryptography (at least from a high-level) before making such definitive conclusions as you've made the past couple days.
"Quantum FUD and AI FUD is exactly the same" is the point i see most of the discourse missing.
AI making rapid mathematical breakthroughs ≠ BTC cryptography being on the verge of collapse (don't believe me, ask fable/astra and they will confirm this).
xitter.cf/climbcode28/status/210…
Cryptography is _the_ most sophisticated field in computer science. Bitcoin's SHA-256 hashing has a brute-force time complexity of ~O(2^256) - roughly 10^77 attempts (ask AI how astronomically large that number is). and that's just hashing. Bitcoin also uses elliptic-curve cryptography (ECDSA / Schnorr).
hashing / merkle trees / SHA-256 / pub-key cryptography / digital signatures - these are the foundations of modern digital security. banking / 2FA / passkeys / healthcare / govt systems / legal records, etc. all depend on cryptographic infrastructure.
if AI x quantum discovers practical shortcuts against widely used cryptography, we'll need fixes across _all_ portions of digital infrastructure, not just "crypto" or "banks". i.e., protecting our military's nuke codes, medical/legal records, bioweapons access, etc.
ofc AI will accelerate cryptographic research. my guess is - frontier US-based labs (who lead AI, don't see this changing) accelerate breakthroughs => cryptographers / standards bodies / industry leaders develop solutions in collaboration w/ frontier labs behind closed doors => institutions/protocols migrate to updated standards. this is already happening btw. NIST finalized its first post-quantum standards in 2024.
agree that Bitcoin has greater coordination challenges, but there's economic/political incentive for btc-core team and the broader network to adopt necessary upgrades.
am not dismissing the risk. but there's a massive distinction between theoretical vulnerabilities / mathematical breakthroughs, and practical exploits. AI advancing rapidly doesn't mean Bitcoin's cryptography is anywhere close to being broken (ask AI).
appreciate u sharing your thoughts, but with all due respect, would implore you to learn more about the field of cryptography (at least from a high-level) before making such definitive conclusions as you've made the past couple days.
ya all it takes is one hole is def correct, and is missing from much of the discourse. and agreed that attackers have advantage in speed / agility, whereas companies/protocols have to test / coordinate / reach consensus / fix / roll out upgrades, which all takes time.
but defenders have advantages too. frontier labs have access to models/capabilities not yet publicly available, and they're already collaborating with industry leaders (military / healthcare / banking / law) to identify/fix vulnerabilities before they're exploited.
ofc this doesn't guarantee attackers won't find something first. but there's also no compelling evidence that a practical break of BTC's cryptography is imminent.
NIST already standardized post-quantum algorithms in 2024, so we're not starting from zero. BTC does have unique coordination challenges to migrate tho, agreed. but i'd bet on them figuring it out, given economic/political incentives.
my main point is simply that AI making rapid mathematical breakthroughs ≠ BTC cryptography being on the verge of collapse (don't believe me, ask fable/astra and they will confirm this).
Cryptography is _the_ most sophisticated field in computer science. Bitcoin's SHA-256 hashing has a brute-force time complexity of ~O(2^256) - roughly 10^77 attempts (ask AI how astronomically large that number is). and that's just hashing. Bitcoin also uses elliptic-curve cryptography (ECDSA / Schnorr).
hashing / merkle trees / SHA-256 / pub-key cryptography / digital signatures - these are the foundations of modern digital security. banking / 2FA / passkeys / healthcare / govt systems / legal records, etc. all depend on cryptographic infrastructure.
if AI x quantum discovers practical shortcuts against widely used cryptography, we'll need fixes across _all_ portions of digital infrastructure, not just "crypto" or "banks". i.e., protecting our military's nuke codes, medical/legal records, bioweapons access, etc.
ofc AI will accelerate cryptographic research. my guess is - frontier US-based labs (who lead AI, don't see this changing) accelerate breakthroughs => cryptographers / standards bodies / industry leaders develop solutions in collaboration w/ frontier labs behind closed doors => institutions/protocols migrate to updated standards. this is already happening btw. NIST finalized its first post-quantum standards in 2024.
agree that Bitcoin has greater coordination challenges, but there's economic/political incentive for btc-core team and the broader network to adopt necessary upgrades.
am not dismissing the risk. but there's a massive distinction between theoretical vulnerabilities / mathematical breakthroughs, and practical exploits. AI advancing rapidly doesn't mean Bitcoin's cryptography is anywhere close to being broken (ask AI).
appreciate u sharing your thoughts, but with all due respect, would implore you to learn more about the field of cryptography (at least from a high-level) before making such definitive conclusions as you've made the past couple days.
climbcode retweeted
Bitcoin's proof of work has done around 2^96 hashes over all time. Breaking the ZKP's in Zcash requires a minimum of 2^120 elliptic curve operations assuming ECDLP. If you had a dedicated ASIC, an elliptic curve addition would cost around 10-100x more than SHA2.
So today breaking an Elliptic curve would cost around 1 billion Bitcoin-hashrate equivalents. AKA a billion times the hardware buildout of all Bitcoin mining ASICs, dedicated to break this.
If a hypothetical billion X improvement came in, you'd need a chip buildout on the level of bitcoin mining before you can break anything. While competing against all the hyperscalars for fab-space.
AKA plenty of warning time, for a recovery process that takes a couple day response time max.
this is one of the main reasons much of crypto has become gh3y crabs in a bucket over the past few yrs.
few onchain traders or "devs" have any code of ethics. most would sell their soul for a few extra $. e.g.
- "i hate cz, he caused 10/10, he ruined crypto" => price starts moving on a BSC coin => "ahh shii it's new genny runner!"
- "i hate pumpfun, it encourages vamping and bundling, it ruined crypto" => price starts moving up => "it's the top revenue generating protocol! go PUMP!"
am not saying i'm above this, but i mostly choose not to participate. would rather trade/invest in protocols actually building cool shit, or timeless memes like pepe. this is why I (and many others ) moved to equities ~ mid 2025. i trade both equities and crypto now, but will never participate in these onchain poor person vamp gh3y bundling shenanigans. most people with sizable capital never will either, unless they're fans of crime ofc
yep agreed. personal example i've seen in medicine:
> friend had intermittent medical issues all his life, doctors always told him "sorry, no idea what it is 🤷♂️"
> used Claude Code - did many rounds of deep research, scraped hundreds of thousands of medical journals' data, created a data model unique to his symptoms
> had multiple models (Opus / Sol at the time, was a few months ago) analyze the data model along with his symptoms he experienced througout his life
> models came to consensus that there was 3/4 chance he had young Parkinsons, should get tested for it
> he got tested, sure enough, he has young Parkinsons. is now patient 7 in new experimental young Parkinsons treatment
life completely changed by using AI tools effectively. pretty amazing lol