Brendan Farmer
@bfarmer
Followers
8K
Following
4K
Media
23
Statuses
544
Assume this is bait, but "great men sitting around moaning about their feelings" is basically as old as western civilization. See Achilles in the Iliad sitting around moaning about his wounded pride and grief. It's difficult to find great men and women of the past who *didn't*
It is 100% true that great men and women of the past were not sitting around moaning about their feelings. I regret nothing.
2
1
9
🚀 New Plonky3 release just dropped. This is probably our most impactful and ambitious release so far: - MUCH faster lookups - High-arity folding - N-ary Merkle trees + Merkle caps - Major Poseidon2 optimizations - Poseidon1 support - And many more… Let’s break it down 👇
4
32
134
Hegseth [yelling]: and if you won't train WarClaude... we have an effective altruist who will.
2
0
7
Dario has claimed that inference revenue exceeds training cost for Anthropic's models, so training yields a positive return on investment. I'm curious whether "training cost" refers to the final run or total R&D compute cost (up to ~4-5x greater, includes experiments, failed
1
0
3
I was randomly thinking about this exact example, that it's somewhat easy (though impressive) to imagine LLMs proving isolated lemmas in math, or coding. I think it's harder to imagine an LLM spontaneously picturing itself in an elevator accelerating to the speed of light, and
Demis Hassabis just defined the real test for AGI. It’s more brutal than anyone expected. Train AI on all human knowledge. Cut it off at 1911. See if it independently discovers general relativity like Einstein did in 1915. If it can, we have AGI. If not, we’re still building
3
0
7
It would be interesting to look at failure rates for projects with organic pmf vs incentivized - ie exclude projects with significant drawdowns in TVL or fees after TGE.
Was curious if this was true and looks like it Among projects that reached some pmf (>10m in tvl or >1m/mo in fees), those that launched a token were +50% more likely to die compared to those that didn't launch a token
1
0
5
First Proof discussion: https://t.co/6GMKqkwSuC LLMs only solved 9, 10, with the caveat that important ideas for 9, 10 were previously published. OpenAI claim to have solved more problems using an unreleased model, but they may have violated the benchmark rules by having human
1
2
8
My previous favorite Bostrom pseud moment is the picture of him looking concerned in front of a whiteboard with nonsense complexity formulas like log log n^sqrt log log n, but “the optimal AI policy is to wait exactly 3.1 days if ‘safety’ increases 0.1% per year” takes the cake
My favourite table of made-up numbers is this one, which recommends waiting exactly 3.3 days to launch AGI if the risk is around 20% of immediate death and there is glacial safety progress.
2
6
124
I don't disagree with this - too hard to project the rate of progress. I do think there are some reasonable counterarguments. First, ARC-AGI-2 is a very easy benchmark for humans that LLMs perform poorly on. It's difficult (for me at least), to get an objective sense of where
The idea that AIs won't soon be able to invent new and useful mathematical structures seems to rest on a "ghost in the machine" style assumption about the nature of mathematical invention. My strong suspicion is that, within the next few years, we will see this assumption
4
1
8
RIP. One thing that I didn't realize was that JSTOR told the government that they didn't want to press charges and asked them to drop the case. The AUSA refused and would only discuss plea agreements involving federal prison time. Awful. https://t.co/vxqJi3ULsh
keker.com
Dan Purcell, one of Swartz' lawyers, writes about the spiteful and unreasonable charges that led to his suicide—and MIT's gutless support of his prosecutors.
6
3
25
America, famously founded on obedience to authority. Please tread on me.
13
1
34
> The L2 roadmap has been a catastrophic failure. One L2 did 5x more volume than every perp dex on Solana combined over the last 30 days.
Messages can travel round trip from NYC to tokyo 60 times in the span of a 12 second Ethereum slot. Saying that latency is fundamentally limited by the speed of light in the context of an ethereum block is like a runner saying he cant run any faster because the speed of light
8
7
227
I'm really excited that the legendary team at ECC is regrouping under a new structure, so that they can continue to build for Zcash without the shackles of Bootstrap's broken and misaligned nonprofit corporate structure. The potential unlock here is enormous. ECC transitioned to
71
158
775
I'm curious about "get as much valuable economic activity on the L1 as possible." Ethereum already hosts far more valuable economic activity than any other protocol. Lighter alone did $160b in volume over the past 30 days, >5x more than all perp dexs on Solana combined. But
This is a good post. Not because I agree with it, directionally I disagree with a lot. But Vitalik is finally telling people what he actually believes, and an indication on how he intends to run the Ethereum Foundation. This creates transparency for those around him, and means
1
2
28
Primer is amazing
Primer (2004) is the pure, unfiltered version of a time-travel film. No hand-holding, no explanations, just two guys breaking the timeline until it collapses on itself. Still one of the wildest debuts ever.
14
1
20
This originally came up in a conversation with @maxgillett and @TomasIsMe but this tweet reminded me of it: https://t.co/ob9XLDmpHd
this is something that a bunch of .hl are not going to like to hear spending ~$1 billion on $HYPE buy backs had zero long term effect on price coinbase, robinhood, nasdaq are all entering perps on crypto and equity that revenue should have been spent on growing a strong
0
0
4
Hyperliquid is a great product, but only one of these can be true: 1) 100% of fees going to burn tokens is good 2) HYPE's valuation is justified because high-growth tech companies trade at 30-40 P/S, and FDV/REV is < 40. The problem is that high revenue multiples assume
8
0
14
This debate comes back up every few months - I think it's interesting. Circle can't force a hardfork imo. Suppose there's a hardfork. The best option for Maker, Arbitrum, Aave, etc would be to redeem USDC on the USDC-supported fork and buy USDT on the decentralized fork. They
Let's run a little thought experiment: Imagine someone managed to compromise the USDC admin key and mint, say, ~$3B. That USDC would be immediately sold for uncensorable assets (BTC, ETH, DAI, etc). Circle would, as soon as they can, pause() USDC while they decide their next
11
4
57
Lot of respect for Jim and Radi and the rest of the team for pushing the space forward and being principled about building a sustainable business. Excited to see what they do next.
Irreducible's time has come to an end. Long story short, @radi_cojbasic and I ultimately came to the realization that we couldn't sustainably build the kind of deep tech business in ZK that motivated us.
3
0
14