SVRGN Weekly Digest #21 💫
A quick trip to meet some founders in person, and I drove a robot
This week was long, hence why you’re receiving this late Sunday evening. But energizing and fun. Hope you enjoy this week’s digest!
🤼 People
Stephane Gosselin - Founder of Frontier Research
We had a great catch up with Stephane regarding the state of crypto. There is mostly either verticalization in unproductive markets or re-iterations of concepts from cycles past. Much like the ghost of Christmases past, it doesn’t make us feel great. Luckily, there are a few stellar exceptions to this…
Herbert Mangesius - GP at VSquared
Herb is one of the first VCs I ever got to know. We organized a tech conference in Munich in 2018 for 2500 people called the Dahoam conference, and at least one hackathon if memory serves. VSquared just announced their second fund of an oversubscribed EUR214M and have done very well by being early to the deep tech trend, with investments like Isar Aerospace in 2018 and IQM.
🚀 Companies
Monava - Information through acoustics
Status: Revenue generating
Source: Network
Founders: Alexander
Why it’s cool:
The background of the name is “MONitoring AVAlanches”, and that was the first use case for the acoustics company. Given Alexander’s background as a pro skier, acoustics engineer and avalanche researcher, he knew the space very well. However, turns out we can use sound localization for a lot more than detecting avalanches!
Blockmesh - Monetize your excess bandwidth
Status: pre-launch
Source: CAVI
Founders: ?
Why it’s cool:
Delivering digital goods over blockchains is just the only way that makes sense, once the option is there. Curious to see how this works out given the rich history of filecoin and helium.
Schelling - Decentralized AI
Status: empty twitter page
Source: CAVI
Founders: Emad Mostaque
Why it’s cool:
You can say what you want about StabilityAI, but at least they did move the world more towards open models. Meta may have latched on to that, or maybe StabilityAI latched on to what the people at LMU were doing… Anyway, Emad is now building the next decentralized AI company.
💡 Ideas & Science
Situational Awareness by Leopold Aschenbrenner
This is a neat 165 page essay series by a former OpenAI employee. In it, Leopold Aschenbrenner lays out the argument for why AGI is imminent. It’s already happening, and if we extrapolate the trajectory, then we’ll be there by 2027. Critics will say LLMs are only predicting the next token, without “real reasoning or intelligence” behind it. Does it matter if we can use it to do intelligent-looking things? In particular, the thing that makes the intelligence scaling never stop is the ability to accelerate fundamental research into every subject, incl. how we make more intelligence.
A somewhat previously less discussed viewpoint of the essays is that of national security-relevance. AGI is the brain that can come up with nuclear weapons, it’s an endless source of round-the-clock von Neumann+Heisenberg+Oppenheimer-level geniuses. Of course every nation in the world will want to get to it first and control such a resource. It’s a lot more difficult to control than nuclear weapons, too. Cyber security takes on a new meaning entirely, when the weapons themselves live in the digital realm. In related news, OpenAI appointed former NSA director Paul M. Nakasone as a board member.
Of course, the major question marks to decide how this (humanity) goes, are still:
will model performance plateau?
when do we run out of (high-quality) data?
how do we coordinate the state-level capital and commitment needed to build multi-GW data centers?
OurNetwork Research on SocialFi
From the first headline saying “Farcaster was 50% bots until recently” to stark realities like “41% of Lens users and 49% of Farcaster users have fewer than 10 followers.” and friend.tech daily fees and volumes down 99% MoM, it’s not looking like these networks built strong… network effects… yet. Or I guess the products just aren’t mostly that interesting compared to the potential to farm potential airdrops. To be fair, probably there’s a boomer equivalent that someone said about twitter in the early days. The ideas make sense, so maybe it’s still just early.