Funny to think about the long term implications of cavemanmaxxing being more token efficient
Will future generations speak caveman English?
🛂 GPT-5.5's CoT keeps leaking in the new Codex update
>Now we know how they got the token efficiency they cavemanmaxxed it.
> Every time, I think of that Kevin Malone quote: "why waste time say lot word when few word do trick."
At what point does compression stop being efficiency and start being lobotomy?
10
2
36
6,003
I'm American and it's hilarious to me I'm making it on Euro crypto talent lists
European crypto founders and operators are elite.
The talent density is absurd.
@diogomonica (Portuguese) - Anchorage
@StaniKulechov (Finnish) - Aave
@gavofyork (British) - Polkadot, Ethereum
@RuneKek (Danish) - Sky
@ilblackdragon (Ukrainian) - NEAR
@PaulFrambot (French) - Morpho
@_pgauthier (French) - Ledger
@borgetsebastien (French) - The Sandbox
@ArthurB (French) - Tezos
@koeppelmann (German) - Gnosis, Safe
@eric_demuth (Austrian) - Bitpanda
@ASvanevik (Norwegian) - Nansen
@jbaylina (Spanish) - Polygon zkEVM
@euler_mab (British) - Euler
@Zac_Aztec (British) - Aztec
@gluk64 (Ukrainian) - zkSync
@gakonst (Greek) - Paradigm, Reth
@bertcmiller (British) - Flashbots
@peter_szilagyi (Hungarian) - Ethereum Geth
e/acc Europe.
5
1
35
10,365
Great thread from @dataalways, who has picked up the mantle on posting creatures from the dark forest.
Meet 0x841: a fresh sandwich bot lurking in the dark forest--currently attacking about 1500 users per day, while going almost unnoticed. What's particularly scary: most of the victims are using private mempools and should be protected.
1
2
13
3,362
Pretty remarkable to think this problem went unsolved purely because of some strange, collective "block" that prevented humans from approaching it the right way
Naturally the next question is: where else might these blocks be?
23 years old with no advanced mathematics training solves Erdős problem with ChatGPT Pro. "What’s beginning to emerge is that the problem was maybe easier than expected, and it was like there was some kind of mental block.”-Terence Tao scientificamerican.com/artic…
5
1
30
6,781
The Internet feels like a minefield right now. Constant vigilance on every email I get, every phone call, DM, package you install, etc.
7
4
46
11,727
Very cool launch.
It feels to me like what you'd naturally come to if you took the latest and greatest in Ethereum research, but looked at it with an intensely pragmatic and product oriented lens. Probably indicative of what's to come from the @tempo team!
Blockchains still broadcast every transaction publicly. Every stablecoin payment leaks the amount, the sender, and the recipient.
We’re excited to share that Tempo is building Zones for businesses that need privacy: private blockchains that are interoperable with the rest of Tempo for stablecoin use cases like payroll, treasury, and settlement.
5
5
48
9,265
Happy to back @PropheticAI and excited to try what the team has been cooking!
We are pleased to share that we have successfully demonstrated the ability to increase lucidity in dreams.
This critical milestone opens the door for our devices to ship soon.
1
1
8
2,243
A summer research fellowship for truth seekers from the inimitable @fkpxls
Introducing Pace Summer Fellowship.
For compulsive truth seekers who care about doing great work in a contained period of time.
4 weeks. $10k. Two fellows. One publishable research artifact.
1
3
2,227
Between widespread distillation + increasing capabilities + compute crunch it makes sense for the frontier labs to keep their top models more closed. Wouldn't be surprised if we get Claude Code or Codex specific models that aren't generally available over APIs next.
2
4
1,462
Axios being compromised by North Korea is probably a sign of what's to come. Expect a lot more compromising of popular OSS packages in the future.
3
1
13
2,540
Crypto has spent years fighting this stuff, and NK years honing their craft on already paranoid crypto people. Attacks breaking containment into the broader software world is both unfortunate and inevitable.
1
4
1,679
A harness is a way of pulling future capabilities of a model into the present.
It gives the model structure that delivers better outcomes at the expense of generality.
Then with each release the model needs less & less structure for the same outcomes. Eventually none at all.
If you're sufficiently AGI pilled, no harness is the best harness
Opus 4.5 needed heavy harnessing, Opus 4.6 needed less
In the limit, the only "harness" AGI will need is context to the outside world
No thinking tricks, no prompts with human intelligence baked in
If you really want to know if AGI is here we need to be a direct pass through for model performance
1
4
1,717