4606
Venkatesh Rao ☀️
@vgr #4606
Delver at contraptions.venkateshrao.com, running summerofprotocols.com, host of /gloom ☀️
132347 Follower 505 Following
SoP 2025 program is announced! Highlights:
1. Curriculum development grants for faculty who want to develop courses about protocol science/studies/entrepreneurship — $500k pool — deadline April 1
2. Protocol science fiction story contest, $5k in prizes; AI-assisted stories not just accepted but encouraged — deadline April 14
3. A distributed AI x blockchains workshop in Bangkok April 21-25; regional travel support available — deadline to register March 21
Plus a lot more. Please share with any educators and SF writers you know
https://open.substack.com/pub/protocolized/p/sop-2025-accelerating-order?r=1cjha&utm_medium=ios
1. Curriculum development grants for faculty who want to develop courses about protocol science/studies/entrepreneurship — $500k pool — deadline April 1
2. Protocol science fiction story contest, $5k in prizes; AI-assisted stories not just accepted but encouraged — deadline April 14
3. A distributed AI x blockchains workshop in Bangkok April 21-25; regional travel support available — deadline to register March 21
Plus a lot more. Please share with any educators and SF writers you know
https://open.substack.com/pub/protocolized/p/sop-2025-accelerating-order?r=1cjha&utm_medium=ios
Yes! Mediocrity ftw :D
4731
Matthew McDowell-Sweet
@msms·a day ago
Why "Normal" Engineers Are the Key to Great Teams - IEEE Spectrum
“When people talk about world-class engineering organizations, they often have in mind teams that are top-heavy with staff and principal engineers, or that recruit heavily from the ranks of former Big Tech employees and top universities. But I would argue that a truly great engineering org is one where you don’t have to be one of the “best” or most pedigreed engineers to have a lot of impact on the business. I think it’s actually the other way around. A truly great engineering organization is one where perfectly normal, workaday software engineers, with decent skills and an ordinary amount of expertise, can consistently move fast, ship code, respond to users, understand the systems they’ve built, and move the business forward a little bit more, day by day, week by week.”
https://spectrum.ieee.org/10x-engineer
“When people talk about world-class engineering organizations, they often have in mind teams that are top-heavy with staff and principal engineers, or that recruit heavily from the ranks of former Big Tech employees and top universities. But I would argue that a truly great engineering org is one where you don’t have to be one of the “best” or most pedigreed engineers to have a lot of impact on the business. I think it’s actually the other way around. A truly great engineering organization is one where perfectly normal, workaday software engineers, with decent skills and an ordinary amount of expertise, can consistently move fast, ship code, respond to users, understand the systems they’ve built, and move the business forward a little bit more, day by day, week by week.”
https://spectrum.ieee.org/10x-engineer
Ok this idea is growing on me and is upgraded from minor to major. The options vs crypto example is peripheral. This is a way to view something much more fundamental — the explore-exploit tradeoff. For a fully certain value commodity, in a perfect competition market with instantaneous price discovery, the frontier collapses to a point at (1,1). A perfectly unbelievable fantasy collapses to (0, 0). The 45 degree ray is the fantasy-to-reality turnpike. As you learn more and the thing becomes more real, the frontier initially expands then contracts, with end points on the unit square perimeter. The (1,0) - (0,1) line is the explore-exploit transition frontier. When the frontier crosses that, you’re in fully convergent exploit mode. At that point you can start thinking about “IPO” or “exiting” the idea in some way, since either correlatedness and liquidity can be maxed out. The whole finance metaphor can be ditched. I’m calling it modal information theory — ie how possible worlds become real.
4606
Venkatesh Rao ☀️
@vgr·2 days ago
Just came up with a principle that may already have a name: Correlation-Liquidity Tradeoff (CLTO): The correlatedness of a speculative asset to an underlying value thesis and its liquidity form a tradeoff curve. Private startup stock has high correlation low liquidity. Crypto tokens have low correlatedness high liquidity. Former requires IRR fictions. Latter can experience wild excursions based on factors having nothing to do with the thesis. Between them is a Pareto frontier
Came up with this idea from a weird angle. I was actually thinking about how reward schedule tempo is slower than decision tempo, causing problems in reinforcement learning.
Eg you are a novelist who writes 1000 words everyday but the reward schedule is every 40k words when you finish and ship a novel and get market feedback.
Or a robot trying sequences of actions to accomplish a task. It only actually gets a reward when it finds a satisficing sequence. Agile iteration can increase feedback rate but not necessarily reward rate.
Came up with this idea from a weird angle. I was actually thinking about how reward schedule tempo is slower than decision tempo, causing problems in reinforcement learning.
Eg you are a novelist who writes 1000 words everyday but the reward schedule is every 40k words when you finish and ship a novel and get market feedback.
Or a robot trying sequences of actions to accomplish a task. It only actually gets a reward when it finds a satisficing sequence. Agile iteration can increase feedback rate but not necessarily reward rate.
Theory: By the time there’s trust in a product, there’s no need for trust in the token. If the token doesn’t enable actually clever mechanics (eigenlayer is a good example maybe?) it is superfluous perhaps
2458
rafa
@rafa·3 days ago
Name a crypto product with PMF that has a token that matters for their PMF
Industrial era societal capacity: Dinosaurs
Trumpism & friends: Asteroid
Protocols: Mammals
Trumpism & friends: Asteroid
Protocols: Mammals
Just came up with a principle that may already have a name: Correlation-Liquidity Tradeoff (CLTO): The correlatedness of a speculative asset to an underlying value thesis and its liquidity form a tradeoff curve. Private startup stock has high correlation low liquidity. Crypto tokens have low correlatedness high liquidity. Former requires IRR fictions. Latter can experience wild excursions based on factors having nothing to do with the thesis. Between them is a Pareto frontier
Came up with this idea from a weird angle. I was actually thinking about how reward schedule tempo is slower than decision tempo, causing problems in reinforcement learning.
Eg you are a novelist who writes 1000 words everyday but the reward schedule is every 40k words when you finish and ship a novel and get market feedback.
Or a robot trying sequences of actions to accomplish a task. It only actually gets a reward when it finds a satisficing sequence. Agile iteration can increase feedback rate but not necessarily reward rate.
Came up with this idea from a weird angle. I was actually thinking about how reward schedule tempo is slower than decision tempo, causing problems in reinforcement learning.
Eg you are a novelist who writes 1000 words everyday but the reward schedule is every 40k words when you finish and ship a novel and get market feedback.
Or a robot trying sequences of actions to accomplish a task. It only actually gets a reward when it finds a satisficing sequence. Agile iteration can increase feedback rate but not necessarily reward rate.
A video chat where @timber and I riff on all things protocol, and provide some sneak previews of SoP 2025 programming (to be announced on Friday) https://contraptions.venkateshrao.com/p/video-recording-protocols-preview
Don’t drink and delve
delve or delve not, there is no prompt
Substack note du jour https://substack.com/profile/2264734-venkatesh-rao/note/c-99998954
No mattter what the new media tech thing is, there will be a group of people saying “now it will be all about curation, curation will be very important.”
“Melissa Chavin, a US immigration lawyer based in London, said Ms Burke could be released either through voluntary departure or expedited removal, but the process of appearing in front of a judge as part of voluntary departure could be delayed because several dozen of the approximately 700 immigration judges had resigned or been fired by the Trump administration.”
This is so dumb. This woman *wants* to leave and can access the means to do so.
https://www.bbc.com/news/articles/c80y3yx1jdyo
This is so dumb. This woman *wants* to leave and can access the means to do so.
https://www.bbc.com/news/articles/c80y3yx1jdyo
The right mental model for getting people into crypto is not “adoption” but “immigration”
Crypto needs a Statue of Liberty, an Ellis island, and an Emma Lazarus type poem
Crypto needs a Statue of Liberty, an Ellis island, and an Emma Lazarus type poem
Are warps being retired?
It’s surprising that the X-files is so rewatchable, given that 2025 is overrun by real conspiracy theorists who are generally toxic and unpleasant people who should logically have ruined the show for us by now.
I think it’s because the conspiracy theorists on X files all the way up to Mulder are charming and relatable and want to believe and know awesome things that would be wild if true. Sadly the real-life ones of today are less interested in UFOs and telekinesis and more interested in finding ways to believe the worst possible things about people or groups they dislike, which would be awful if true, and require no magic or alien science .
A realistic Mulder today would be a partisan hack and hate the smoking man for being appointed by the other party. He would be obsessed with vaccines, birth certificates etc. Instead of being banished to the basement he’d be running a high-powered inquisition. The lone gunmen would be spreading rumors about people being pedophiles. Sculley would be an influencer.
I think it’s because the conspiracy theorists on X files all the way up to Mulder are charming and relatable and want to believe and know awesome things that would be wild if true. Sadly the real-life ones of today are less interested in UFOs and telekinesis and more interested in finding ways to believe the worst possible things about people or groups they dislike, which would be awful if true, and require no magic or alien science .
A realistic Mulder today would be a partisan hack and hate the smoking man for being appointed by the other party. He would be obsessed with vaccines, birth certificates etc. Instead of being banished to the basement he’d be running a high-powered inquisition. The lone gunmen would be spreading rumors about people being pedophiles. Sculley would be an influencer.
It’s honestly hard to like hacker news people 😐
Medieval lifetime travels leaderboard
———————————————————-
1. Ibn Batuta (1304-1378): 117,000 km (73,000 miles)
2. Zheng He (1371-1433): 50,000 km (31,000miles)
3. Marco Polo (1254-1324): 24,000 km (15,000 miles)
…
…
431, 776, 231. My ancestor, Delver Rao (1280-1350): 5 km (3 mi)
———————————————————-
1. Ibn Batuta (1304-1378): 117,000 km (73,000 miles)
2. Zheng He (1371-1433): 50,000 km (31,000miles)
3. Marco Polo (1254-1324): 24,000 km (15,000 miles)
…
…
431, 776, 231. My ancestor, Delver Rao (1280-1350): 5 km (3 mi)
Like every other living thing I’m just a dissipative system. Just more dissipated and less systematic.
What is the protocol equivalent of an “operator”?
This might be the most fun I’ve had writing an essay in a while. Felt some of that old going-nuts-with-an-idea energy flowing.
https://open.substack.com/pub/contraptions/p/discworld-rules?r=1cjha&utm_medium=ios
https://open.substack.com/pub/contraptions/p/discworld-rules?r=1cjha&utm_medium=ios
Any good frames that do AI things for money?
Accelerationism: If you floor the gas pedal hard enough, gasoline will meme itself into existence in your tank.
If you want peace, prepare for war
If you want war, prepare for peace
If you want the benefits of war without the costs of peace, lie through your teeth about everything
If you want war, prepare for peace
If you want the benefits of war without the costs of peace, lie through your teeth about everything
If you’re temperamentally thorough, you should look for domains where efficiency is basically impossible — open, messy, complex natural environment, topological disconnectedness, multi-scale structure
Lots of intrinsic friction that feels like a feature rather a bug. High information content per unit space time.
Lots of intrinsic friction that feels like a feature rather a bug. High information content per unit space time.
Let’s play guess who said it again. No googling or asking llms
“Always try to be nice, but never fail to be kind”
“Always try to be nice, but never fail to be kind”
The central dogma of SV is perhaps that you need high talent concentrations to scale. ‘If you can’t hire 500 engineers in a quarter when you hit takeoff you’re ngmi” belief system. Even with a lot of trad companies moving away from Bay Area for scale out, the belief in talent concentrations is still very high. The “SF is so back!” endemic meme underlines the strength of this dogma.
Crypto ecosystems are a clear counter-example. Outside of big events like Devcon no SV-style concentration exists anywhere. The implications of this have not yet sunk in. I’m convinced it’s not a special case. Any kind of digital tech can be effectively produced by decentralized talent networks. SV is addicted to single-geography concentrations for 2 reasons:
1. Investors are rich and lazy and want their investments down the street
2. Boomer/GenX managers don’t know how to manage virtually and don’t want to learn
Crypto capitalizes differently and has millennial leadership so broke free of the dogma
Crypto ecosystems are a clear counter-example. Outside of big events like Devcon no SV-style concentration exists anywhere. The implications of this have not yet sunk in. I’m convinced it’s not a special case. Any kind of digital tech can be effectively produced by decentralized talent networks. SV is addicted to single-geography concentrations for 2 reasons:
1. Investors are rich and lazy and want their investments down the street
2. Boomer/GenX managers don’t know how to manage virtually and don’t want to learn
Crypto capitalizes differently and has millennial leadership so broke free of the dogma
Hey folks! We've put the first 2025 SoP workshop together. It will be April 21-25 at CMKL University in Bangkok and focused on open distributed AI x blockchains (loosely d/acc inspired). The name "khlongs and subaks" refers to Thai and Balinese words related to rice agriculture, which is our motif/inspiration. Here is the call for applications. There will be technical, cultural, and public policy tracks (including a hackathon component). If you'd like to attend, please apply using the registration/application form. The workshop is free and limited travel support is available for those who need it.
Capacity is limited so we will be prioritizing people from the region and with the right background from the topic. Application deadline is March 15. Please share this call with specific people who might make good participants.
https://www.cmkl.ac.th/event/khlongs-and-subaks-open-distributed-ai-x-blockchain-protocols-in-southeast-asia
Capacity is limited so we will be prioritizing people from the region and with the right background from the topic. Application deadline is March 15. Please share this call with specific people who might make good participants.
https://www.cmkl.ac.th/event/khlongs-and-subaks-open-distributed-ai-x-blockchain-protocols-in-southeast-asia
4606
Venkatesh Rao ☀️
@vgr·20:58 04/03/2025
Becoming increasingly obvious that distributed AI is the killer app of blockchains. When models need to have financialized “trust but verify” relations with other models.
We’re doing a week-long workshop on this at CMKL in Thailand in April. If you’re working at this intersection in some substantive way (tech, policy, culture) and especially if you’re from SEA, we want you there. Stay tuned for the announcement in a few hours when SEA starts to wake up.
We’re doing a week-long workshop on this at CMKL in Thailand in April. If you’re working at this intersection in some substantive way (tech, policy, culture) and especially if you’re from SEA, we want you there. Stay tuned for the announcement in a few hours when SEA starts to wake up.
.png)
Khlongs and Subaks: Open Distributed AI x Blockchain Protocols in Southeast Asia - Carnegie Mellon - CMKL│Thailand
Apr 21, 2025 | CMKL University | The Summer of Protocols (SoP) program, funded by the Ethereum Foundation, in collaboration with CMKL University, Seapunk Studios, and Global Chinese Community, is excited to present Khlongs and Subaks: Open Distributed AI x Blockchain Protocols in Southeast Asia—an immersive 5-day workshop from April 21-25, 2025, at CMKL University in Bangkok.
www.cmkl.ac.th
Highly recommend. Anything nifty gets up to is bound to be fun
5478
niftynei
@niftynei·01:18 05/03/2025
ya’ll keccers is right, come hack on bitcoin at our first ever SXSW event in Austin next week!!
Trying to read 3 books in parallel
- Chip Wars
- Nvidia Way
- Giordano Bruno and the Hermetic Tradition
Look upon my readings ye mighty and despair
- Chip Wars
- Nvidia Way
- Giordano Bruno and the Hermetic Tradition
Look upon my readings ye mighty and despair
Reupping
4606
Venkatesh Rao ☀️
@vgr·17:38 21/02/2025
Has there ever been a token standard proposal that triggers a receiver end accept action and goes to a return address if it doesn’t happen? Ie “push-based claims” instead of “pull-based claims.” Default “reject” over “accept”
Proposal: a class of tokens that come with a default burn/return address and expiry date. Once sent, the recipient wallet “sees” it (like an airdrop) but unless it claims it within the window, the token goes to return address. So you just compare the issue timestamp, any claim timestamp, and the time constant, to determine who actually owns it. UX: past expiry period the status is unambiguous and permanent, and same as today’s tokens. The return address can only be the origin address or a known burn address like 0x0000…
Ie a heisen-coin that decoheres after a set period. Proposed name: looper or loop tokens. Or boomerang tokens. Can come in fungible and non-fungible varieties and replace current erc
Anyone want to develop this as an ERC with me 😬😰?
Proposal: a class of tokens that come with a default burn/return address and expiry date. Once sent, the recipient wallet “sees” it (like an airdrop) but unless it claims it within the window, the token goes to return address. So you just compare the issue timestamp, any claim timestamp, and the time constant, to determine who actually owns it. UX: past expiry period the status is unambiguous and permanent, and same as today’s tokens. The return address can only be the origin address or a known burn address like 0x0000…
Ie a heisen-coin that decoheres after a set period. Proposed name: looper or loop tokens. Or boomerang tokens. Can come in fungible and non-fungible varieties and replace current erc
Anyone want to develop this as an ERC with me 😬😰?
Here’s a Tamil joke
What do you call an informal ai to ai greeting
aiaiyo
What do you call an informal ai to ai greeting
aiaiyo
*Seinfeld voice* is this anything
The current political project of memeing a particular vision of America into existence is going to fail.
Been thinking fairly hard and skeptically about this conclusion because I’m politically predisposed to believe it and hope for it, but I’ve convinced myself it checks out and am starting to bet accordingly. No interest in making the case, sharing my reasoning, or persuading others, but figured the headline was worth sharing.
Been thinking fairly hard and skeptically about this conclusion because I’m politically predisposed to believe it and hope for it, but I’ve convinced myself it checks out and am starting to bet accordingly. No interest in making the case, sharing my reasoning, or persuading others, but figured the headline was worth sharing.
Sending a concerted new techlash gathering steam, targeting AI. Resembles anti-crypto in many ways.
I guess trial by techlash is a rite of passage.
Feels different from 10y ago though, due to tech maga shennanigans in DC
I guess trial by techlash is a rite of passage.
Feels different from 10y ago though, due to tech maga shennanigans in DC
Becoming increasingly obvious that distributed AI is the killer app of blockchains. When models need to have financialized “trust but verify” relations with other models.
We’re doing a week-long workshop on this at CMKL in Thailand in April. If you’re working at this intersection in some substantive way (tech, policy, culture) and especially if you’re from SEA, we want you there. Stay tuned for the announcement in a few hours when SEA starts to wake up.
We’re doing a week-long workshop on this at CMKL in Thailand in April. If you’re working at this intersection in some substantive way (tech, policy, culture) and especially if you’re from SEA, we want you there. Stay tuned for the announcement in a few hours when SEA starts to wake up.
I miss VCRs
This is dumb. The old Meeker decks were uninspired and marginally useful at best even when restricted to the tech economy. As a way of analyzing the macroeconomics of a country, the corporate-sectoral microeconomics approach is not even wrong. Basic deficit hawk thinking wrapped in technobabble to lay a paper trail of post-hoc justification for doge theater. https://www.axios.com/2025/03/03/us-government-fiscal-trump-doge
This note is too long to screenshot. Click over to substack if you want to read it.
https://substack.com/@contraptions/note/c-97695182
https://substack.com/@contraptions/note/c-97695182
The world is no longer consistent, available, or partition resistant. We’re supposed to be able to get 2/3 but now it’s at 0/3 🤬
At this point we should all just learn Russian
This is great. @timber has been firing on all cylinders lately https://open.substack.com/pub/protocolist/p/mold-rust-and-the-risk-society
